Apf 2015 5 1 by Hogrefe

Volume 5 | Number 1 | 2015 ISSN-L 2192-0923 • ISSN-Print 2192-0923 • ISSN-Online 2192-0931

AVIATION PSYCHOLOGY AND APPLIED HUMAN FACTORS www.hogrefe.com/journals/apahf

Editor in Chief Ioana Koglbauer Associate Editors Cristina Albuquerque André Droog Hinnerk Eißfeldt Harald Kolrep Monica Martinussen Michaela Schwarz Matthew Thomas

1|15

Aviation Psychology and Applied Human Factors

Editor-in-Chief Associate Editors

Editorial Board

Publisher

Production Subscriptions Advertising/ Inserts ISSN Copyright Information

Cover Pictures

Publication Subscription Prices Payment

Electronic Full Text Abstracting/ Indexing Services

Ioana Koglbauer, Graz University of Technology, Institute of Mechanics, Kopernikusgasse 24/IV, 8010 Graz, Austria, Tel. +43 (316) 873-4111, E-mail journal@eaap.net Cristina Albuquerque, Lisbon, Portugal André Droog, Groningen, The Netherlands Hinnerk Eißfeldt, DLR, Hamburg, Germany Harald Kolrep, HFC, Berlin, Germany Monica Martinussen, University of Tromsø, Norway

Michaela Schwarz, Austro Control, Vienna, Austria Matthew Thomas, Westwood Thomas Associates, Goodwood, Australia

Andrew Bellenkes, Monterey, CA, USA Rob Lee, Canberra, Australia Robert Bor, London, UK Wen-Chin Li, Cranfield, UK Guy Boy, Melbourne, FL, USA Carol Manning, Oklahoma City, OK, USA Thomas Carretta, Dayton, OH, USA Dietrich Manzey, Berlin, Germany Sidney Dekker, Mount Gravatt, Australia Lena Mårtensson, Stockholm, Sweden Mike Feary, Moffett Field, CA, USA Peter Maschke, Hamburg, Germany Shan Fu, Shanghai, PR China Randy Mumaw, Seattle, WA, USA Kazuo Furuta, Tokyo, Japan Peter Murphy, Franklin, Australia Hans Giesa, Hamburg, Germany Jan Noyes, Bristol, UK Don Harris, Coventry, UK Teresa Oliveira, Lisbon, Portugal Brent Hayward, Albert Park, Australia Jan J. Roessingh, Amsterdam, The NetherBrian Hilburn, Haddonfield, NJ, USA lands David Hunter, Peoria, AZ, USA Gideon Singer, Linkoping, Sweden Peter Jorna, Lage Vuursche, The Netherlands Anthony Smoker, Sidlesham, UK Hung-Sying Jing, Taiwan, ROC Mark Wiggins, Sydney, Australia Wolfgang Kallus, Graz, Austria Ann Williamson, Sydney, Australia Hogrefe Publishing, Merkelstr. 3, D-37085 Göttingen, Germany, Tel. +49 551 99950-0, Fax +49 551 99950-425, E-mail publishing@hogrefe.com, Web http://www.hogrefe.com North America: Hogrefe Publishing, 38 Chauncy Street, Suite 1002, Boston, MA 02111, USA, Phone (866) 823-4726, Fax (617) 354-6875, E-mail publishing@hogrefe.com Regina Pinks-Freybott, Hogrefe Publishing, Merkelstr. 3, D-37085 Göttingen, Germany, Tel. +49 551 99950-428, Fax +49 551 99950-425, E-mail journalsproduction@hogrefe.com Hogrefe Publishing, Herbert-Quandt-Str. 4, D-37081 Göttingen, Germany, Tel. +49 551 99950-900, Fax +49 551 99950-998 Marketing, Hogrefe Publishing, Merkelstr. 3, D-37085 Göttingen, Germany, Tel. +49 551 99950-423, Fax +49 551 99950-425, E-mail marketing@hogrefe.com ISSN-L 2192-0923, ISSN-Print 2192-0923, ISSN-Online 2192-0931 Ó 2015 Hogrefe Publishing. This journal as well as the individual contributions and illustrations contained within it are protected under international copyright law. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photo-copying, microfilming, recording or otherwise, without prior written permission from the publisher. All rights, including translation rights, reserved. Reprinted with permission (images 1–6, clockwise from top left): Image 1, Lufthansa Technik AG, Hamburg, Germany; image 5, Ó UK MOD Crown Copyright 2011, SAC Mark Dixon, RAF, reproduced under the UK Open Government License; image 6, DFS Deutsche Flugsicherung GmbH, Hamburg, Germany. Published in 2 issues per annual volume. Calendar year subscriptions only. Rates for 2015: Institutions US $294.00/1210.00/£168.00; Individuals US $143.00/1102.00/£82.00. Postage and handling US $8.00/16.00/£5.00. Single copies US $147.00/1105.00/£84.00 + postage and handling. Payment may be made by check, international money order, or credit card, to Hogrefe Publishing, Merkelstr. 3, D-37085 Göttingen, Germany, or, for North American customers, to Hogrefe Publishing, 38 Chauncy Street, Suite 1002, Boston, MA 02111, USA. The full text of Aviation Psychology and Applied Human Factors is available online at www.psyjournals.com and in PsycARTICLES. Abstracted/indexed in PsycINFO, PSYNDEX, and Academic Index. APAHF Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors Volume 5, No. 1, 2015 Official Organ of the European Association for Aviation Psychology (EAAP) and the Australian Aviation Psychology Association (AAvPA)

Contents Editorial

Original Articles

APAHF in Practice

Book Reviews News and Announcements

Taking Off 2015 With Aviation Psychology and Applied Human Factors Ioana Koglbauer

Farewell to Don Harris and Introduction to the New Editor-in-Chief André Droog

Safety Culture and Safety-Relevant Behavior in Air Traffic Management: Validation of the CANSO Safety Culture Development Concept Michaela Schwarz and K. Wolfgang Kallus

Experimental Evaluation of the AIRWOLF Weather Advisory Tool for En Route Air Traffic Controllers Ulf Ahlstrom

Pilots Who Are Perceived as Unsociable Are Perceived as More Likely to Have a Mental Illness: An Affective Perspective Scott R. Winter and Stephen Rice

A Request for Regulatory Revision: Instructions for Passenger Bracing for Emergency Landings Kyung In Yoo and Jan M. Davies

Toward Evidence-Based Decision Making in Aviation: The Case of Mixed-Fleet Flying Timothy J. Mavin, Wolff-Michael Roth, Kassandra Soo, and Ian Munro

Why Don’t Women Fly? Monica Martinussen

The Best Paper Award

Meetings and Congresses

Aviation Human Factors Related Industry News

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1)

Ó 2015 Hogrefe Publishing

Taking Off 2015 With Aviation Psychology and Applied Human Factors Since October 2014 I have the privilege to be the Editor-inChief of the journal Aviation Psychology and Applied Human Factors. In the past I have been involved as reviewer for several journals and last year I joined the Editorial Board of the International Journal for Aviation Psychology. I enjoy the process of helping fellow scientists to improve their work and get it published. As a pilot and aviation psychologist I have a strong interest in promoting this journal as a high-quality international platform for scientific and practical exchange on aviation psychology and human factors. Don Harris performed an outstanding work with launching the journal as an Editor-in-Chief and leading it at a high level for the past 4 years. I thank him in the name of the APAHF community. Taking over from Don Harris is a great opportunity and I believe that the journal has considerable potential for further growth. The journal has a very strong and committed team of Associate Editors and Editorial Board members, as well as external reviewers which ensures that manuscripts are expertly reviewed. In Hogrefe Publishing, Göttingen, Germany we have a very competent and efficient partner for producing and distributing the journal. And now – fasten your seatbelts because Captain Koglbauer and her crew of Associate Editors will fly even higher! We plan several actions to increase the impact and public awareness of the journal, and to attract a greater share of excellent manuscripts. This year we have introduced the annual ‘‘Best Paper Award’’ which recognizes

Ó 2015 Hogrefe Publishing

excellence in aviation psychology and human factors research of a scientific paper, research note, or practitioner paper submitted to the journal Aviation Psychology and Applied Human Factors. We also plan to accelerate the review process and publish research results in as timely a manner as possible. In the near future we will introduce the Editorial Manager, an online peer-review web-based system for manuscript submission and management. We will also adhere to higher ethical standards and promote full transparency and disclosure of the contribution of study sponsor(s) to the published manuscripts. For a proper resource management, the APAHF community can contribute by submitting their excellent papers to the journal and by promoting the journal within their scientific networks. Enjoy the flight!

Correspondence Address Ioana Koglbauer Editor-in-Chief Graz University of Technology Institute of Mechanics Kopernikusgasse 24/IV 8010 Graz Austria Tel. +43 316 873-4111 E-mail journal@eaap.net

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):1 DOI: 10.1027/2192-0923/a000076

Editorial

Editorial Farewell to Don Harris and Introduction to the New Editor-in-Chief The journal Aviation Psychology and Applied Human Factors started in 2011 as an enterprise of the EAAP in collaboration with the Australian Aviation Psychology Association (AAvPA). Don Harris took on the demanding task of Editorin-Chief of the rising journal most energetically. His 4-year term as the key figure of the journal ended in September 2014 when Don stepped down from the EAAP Board. After being appointed Professor at Coventry University it became more and more difficult for him to combine his more than full time job with the likewise time consuming journal activities. Don has done a great job in the past 4 years and the EAAP is very grateful to him for his amazing human performance. The search for a new Editor-in-Chief of the APAHF journal took some time but fortunately not too long. Ioana Koglbauer from Austria entered the EAAP board in 2014 and was interested and willing to continue the work. Ioana received her PhD in Psychology at the University of Graz, Austria, in 2009, spent post-doctoral fellowships at EUROCONTROL Experimental Centre in Brétigny, France and at Brightline Avionics, Austria. Currently she is Key Human Factors Researcher at Graz University of

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):2 DOI: 10.1027/2192-0923/a000078

Technology where she is teaching Human Factors for more than 3 years. She has published several international journal and conference articles in the areas of aviation psychology and human factors. She has served as a chair of many international conferences in aviation psychology, such as the International Symposium on Aviation Psychology in Dayton Ohio and the EAAP Conference. She is also an editorial board member of The International Journal of Aviation Psychology. We are delighted to welcome Ioana as the head of the APAHF journal. As stated in her editorial also published in this issue of APAHF, she expresses her ambition to develop the journal into a leading forum in the aviation sciences. Correspondence Address André Droog EAAP President Van Speykstraat 49B 9726 BK Groningen The Netherlands Tel. +31 (0)50 526-5104 E-mail adroog@planet.nl

Ó 2015 Hogrefe Publishing

Safety Culture and Safety-Relevant Behavior in Air Traffic Management Validation of the CANSO Safety Culture Development Concept Michaela Schwarz1 and K. Wolfgang Kallus2 1

Austro Control GmbH, Vienna, Austria, 2Department of Psychology, University of Graz, Austria

Keywords: air traffic management, safety behavior, safety culture, safety management system, validation

The Civil Air Navigation Services Organisation (CANSO) represents the interests of air navigation service providers (ANSPs) worldwide. In 2010, the CANSO Waypoint Strategy stated: By 2013 a significant number of CANSO members . . . will have achieved an implementing level of maturity for safety culture objectives as measured by the safety management system (SMS) maturity level. That means a positive safety culture is developing, although still immature. Safety culture is measured and improvement plans address the need for individuals to be aware of, and support, the organisation’s shared beliefs, assumptions and values regarding safety. (CANSO, 2009, pp. 17–18)

cultural aspects within their safety management systems (SMSs). CANSO has therefore developed a safety culture concept taking into account different levels of SMS maturity in organizations. This paper supports the validation of the CANSO safety culture development concept, looking at the relationship between safety culture and safety-relevant behavior (SRB) on the air traffic management (ATM) job. Readers will be offered a concise overview of the background of safety culture and SRB. Results will also provide advice on how to measure and improve safety culture and behavior to comply with current regulations and sustain high safety standards in the increasingly demanding ATM environment.

Safety Culture and Safety Climate

Following CANSO, the European Commission (2011, 2013) mandated ANSPs to regularly assess and improve their level of safety culture as one of the three key performance indicators for safety. In Europe the majority of ANSPs have managed to implement a standardized safety culture assessment approach promoted by the European Organisation for the Safety of Air Navigation (EUROCONTROL). However, across the CANSO region especially in Africa, Asia-Pacific, the Middle East, and South America, ANSPs are only starting to consider

The term safety culture was first mentioned in the investigation of the 1986 Chernobyl nuclear power plant accident, where a poor safety regime was declared as a major contributing factor to the accident (International Atomic Energy Agency [IAEA], 1986). The safety culture concept originated in the organizational culture (Schein, 1985, 2010) and anthropology literature (Guldenmund, 2000, 2010). Organizational culture reflects the usual and habituated way of thinking and acting within an organization. It is distinct from organizational climate, which is described as

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17 DOI: 10.1027/2192-0923/a000068

Original Article

Abstract. Since 2010, air navigation service providers have been mandated to implement a positive and proactive safety culture based on shared beliefs, assumptions, and values regarding safety. This mandate raised the need to develop and validate a concept and tools to assess the level of safety culture in organizations. An initial set of 40 safety culture questions based on eight themes underwent psychometric validation. Principal component analysis was applied to data from 282 air traffic management staff, producing a five-factor model of informed culture, reporting and learning culture, just culture, and flexible culture, as well as management’s safety attitudes. This five-factor solution was validated across two different occupational groups and assessment dates (construct validity). Criterion validity was partly achieved by predicting safetyrelevant behavior on the job through three out of five safety culture scores. Results indicated a nonlinear relationship with safety culture scales. Overall the proposed concept proved reliable and valid with respect to safety culture development, providing a robust foundation for managers, safety experts, and operational and safety researchers to measure and further improve the level of safety culture within the air traffic management context.

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Original Article

the enduring and perceived quality of the inner environment and features of an organization, which can be influenced by the people working within the organization (von Rosenstiehl & Nerdinger, 2011). In ATM, the safety culture concept was applied to partially explain the Milano Linate collision on takeoff in 2001 and the Ueberlingen (Lake Constance) midair collision in 2002 (EUROCONTROL, 2008). The Lake Constance investigation report (Bundesstelle für Flugunfalluntersuchung [BFU], 2004) recognized lack of safety culture as one of the systemic causes, stating that ‘‘safety and risk management are required for the development of a healthy safety culture. They were not present and had to be developed together with the expertise’’ (BFU, 2004, p. 91). Since the inception of safety culture, a debate has been ongoing about the differences between safety culture and safety climate. According to Mearns and Flin (1999): Safety climate best describes employees’ perceptions, attitudes, and beliefs about risk and safety, typically measured by questionnaire surveys providing a ‘‘snapshot’’ of the current state of safety. Safety culture is a more complex and enduring trait reflecting fundamental values, norms, assumptions and expectations, which to some extent reside in societal culture. (Mearns & Flin, 1999, p. 5) For a detailed discussion and arguments in favor of these separate but related concepts, readers are referred to Heese (2012). Regular (repeated) safety climate assessments provide an insight into the underlying safety culture. This paper picks up the CANSO definition of safety culture as ‘‘the enduring value, priority and commitment placed on safety by every person and every group at every level of the

organisation’’ (CANSO, 2009, p. 11). According to the CANSO Standard of Excellence for Safety Management Systems (CANSO, 2009), safety culture is the enabler for an effective SMS taking into account different levels of safety culture (Fleming, 2000; Parker, Lawrie, & Hudson, 2006) in line with the SMS maturity pathways: initiating, planning, implementing, and managing as well as continuous improvement (CANSO, 2009). Figure 1 shows the relationship between safety culture and the SMS maturity pathways.

Measuring Safety Culture The current tools available to assess safety culture in ATM are the CANSO Safety Culture Development Questionnaire (SCD-Q; CANSO, 2009) and the EUROCONTROL Safety Culture Measurement Toolkit (SCMT; Mearns, Kirwan, & Kennedy, 2009). Both tools originally based their safety culture model on eight underlying components: informed (shared safety information), reporting (open communication of mistakes), just (fair treatment of employees), learning (continuous improvement) and flexible culture (adaptation to unexpected situations), management’s safety attitudes (genuine commitment to safety), safety-relevant behavior (e.g. involvement, teamwork, responsibility) and risk perception (awareness of operational threats) in line with the broad range of existing literature (Ek, Akselsson, & Johansson, 2003; Ek & Arvidsson, 2002; Guldenmund, 2000; Hopkins, 2006; Reason, 1997; Wiegmann et al., 2002). For detailed descriptions of each safety culture component and expected outcomes on each development level, refer to Heese (2012). The CANSO safety culture development concept combines a traditional questionnaire

Figure 1. Safety culture development and safety management system (SMS) maturity levels combined. Adapted from The CANSO Standard of Excellence in Safety Management Systems, by the Civil Air Navigation Services Organisation (CANSO, 2009, 2014) and Fleming (2000). Ó 2009 by CANSO. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Ó 2015 Hogrefe Publishing

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Table 1. Safety culture, SMS maturity and safety-relevant behavior

Safety culture components (CANSO, 2009, 2014) Informed culture Reporting culture

Learning culture Just culture

Management’s safety attitudes Safety-relevant behavior Risk perception

All within the organization openly seek and exchange safety information. Management systems and approach encourages employees to challenge procedures/practices and people in their quest to improve safety performance. There is a clear and published policy on how to dialogue with judicial authorities and media. Lessons from within the organization and different industry sectors are used to enhance the organization’s safety culture approach. The organization flexibly adjusts to changes/disturbances and sustains required operations under expected and unexpected conditions. Management cooperates and supports customers, suppliers, and contractors to improve their safety standards. All personnel are proactive and committed to improving safety. Documentation and practice reflect the use of both proactive and predictive methods to inform the organization about inherent risk levels.

Safety communication

Safety (rule) compliance

Resilient safety behavior

Safety leadership Safety citizenship behavior/ safety participation Safety knowledge

Note. SMS = safety management system.

technique to assess safety climate with safety-related interviews related to the previous operational work shift, to accurately assess and understand safety culture artefacts within an organization.

CANSO supports the safety culture pyramid model (Patankar & Sabin, 2010) considering multiple layers of an organization to assess safety culture: safety values, safety strategies, safety climate, and safety performance. Safety performance is positioned on the top of the pyramid and indicates the everyday SRB that can be observed on the job (Heese, 2012). SRB ‘‘reflects the extent to which every level of the organization behaves such as to maintain and improve the level of operational safety’’ (Piers, Montijn, & Balk, 2009, p. 6). A search of the SciVerse Scopus research database on the article title, abstract, and key words ‘‘safety-(relevant) behaviour AND aviation/air traffic management/air traffic control,’’ results in less than 30 articles. Broadening the search to other high-risk industries such as nuclear power, oil and gas, or transportation in general leads to more hits. However, most of the studies relate to the occupational health and safety domain (physical, mental and social well-being of workers; International

Labour Organisation, 1998) rather than to operational safety (dynamic, integrated management of air traffic and airspace; International Civil Aviation Organization, 2007). Some articles point toward safety citizenship behavior (stewardship, voicing own opinions, helping coworkers etc.; Didla, Mearns, & Flin, 2009), safety communication (talking about safety issues), safety compliance (adhering to rules and procedures; Bosak, Coetsee, & Cullinane, 2013), safety participation (participating in voluntary safety activities; Lu & Yang, 2011, and safety knowledge (familiarity with safety practices; Griffin & Neal, 2000; Hofman & Morgeson, 1999), as well as safety leadership (managers are leading by example; Martinez-Córcoles, Gracia, Tomás, & Peiró, 2011; O’Dea & Flin, 2001). Most articles focus on ‘‘traditional’’ safety management approaches, considering the absence of or lack of SRB as ‘‘unsafe’’ or ‘‘risky’’ behavior (Reason, 1997). Recent research proposes resilient safety behavior (managing uncertainty, anticipation; Malakis & Kontogiannis, 2011) inspired by the new resilience engineering approach based on Hollnagel et al. (2006, 2011). Table 1 provides a complete picture of the CANSO safety culture development concept by mapping safety culture components (CANSO, 2009) against the highest of five SMS maturity levels, as required by the CANSO Standard of SMS Excellence (CANSO, 2009, 2014) and associated SRB.

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Safety Culture and SRB on the Job

Original Article

Flexible culture

SMS maturity (highest level-5) (CANSO, 2014, Appendix A, pp. 17ff.)

Safety-Relevant Behaviors (Didla, Mearns & Flin, 2009; Bosak, Coetsee & Cullinane, 2013; Lu & Yang, 2011; Griffin & Neal, 2000; Hofman & Morgeson, 1999; MartinezCórcoles, Gracia, Tomás, & Peiró, 2011; O’Dea & Flin, 2001)

Original Article

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Validating Safety Culture Measures

Sample

Validity relates to the assessment of psychometric properties of items used to measure a latent construct. ‘‘Emerging paradigms replace prior distinctions of face, content-, and criterion validity with the unitary concept ‘construct validity,’ the degree to which a score can be interpreted as representing the intended underlying construct’’ (Cook & Beckman, 2006, p. 166). A number of validated safety culture assessment tools are available for other high-risk industries such as oil and gas (Offshore Safety Questionnaire; Mearns, Whitaker, & Flin, 2001), health care (Patient Safety Culture Questionnaire; Steyrer, Latzke, Pils, Vetter, & Strunk, 2011), and aviation (Aviation Safety Climate Scale; Evans, Glendon, & Creed, 2007). Zohar (2010) claims that research over the past 30 years has demonstrated safety culture as a robust predictor of safety outcomes (criterion validity) across industries and countries. On closer inspection, the majority of papers focus on the occupational health and safety domain (Clarke, 2006). In fact empirical evidence that safety culture improvement programs have an actual influence on operational safety performance (criterion validity) remains yet to be demonstrated (Mearns et al., 2013). This paper therefore supports the evaluation of the CANSO safety culture development concept by assessing its psychometric properties (construct and criterion validity).

The trial sample consisted of 50 licensed terminal ATCOs across two different locations (average age: 26–35 years, average length of service: 7–14 years) equaling a 62.5% response rate. The main sample consisted of 282 (50.8% response rate) ATM staff (55% licensed en-route and terminal ATCOs, 26.6% ATSEPs, 16% METs and 2.5% unknown) across 10 different sites. The majority of participants were male, within the 26- to 35-year and 36- to 45year age group, with an average length of 15–29 years of full-time service in the company. Approximately 12% of respondents were also managers, 23% supervisors; 16% had an additional safety role (e.g., local safety committee [LSC]1), and 40% were also trainers or on-the-job instructors. The follow-up sample consisted of 32 (84.2% response rate) members of the LSC. Demographic data cannot be reported for confidentiality reasons (groups smaller than 10 participants).

Hypotheses

The CANSO Safety Culture Development Questionnaire

Tied in with the empirical evidence, this study aimed to evaluate construct and criterion validity for the CANSO Safety Culture Development Questionnaire. The following hypotheses were postulated: Hypothesis 1 (H1): CANSO components are reliable and valid with respect to safety culture across different occupational groups and assessment dates (construct validity). Hypothesis 2 (H2): Safety culture scores will predict retrospectively reported safety-relevant behavior on the job (criterion validity).

Method The study was conducted between 2011 and 2013 in three data collection phases (trial, main study, and follow-up) with ATM staff including licensed air traffic controllers (ATCOs), air traffic safety electronics personnel (ATSEPs), and meteorologists (METs).

Measures The measures consisted of the CANSO SCD-Q and a safety-related reconstruction interview associated with the previous shift (Safety-Related Reconstruction Interview [SRECO]) as detailed in the following.

The CANSO SCD-Q consists of a shortlist of 40 items in English selected from a database (CANSO, 2011) of more than 900 questions previously used to assess safety culture in the aviation industry (universities, airlines, ANSPs). The 40 items were originally grouped under eight facets of safety culture based on content analysis by a group of 10 subject matter experts (safety managers, operational staff, and human factors experts) in the ATM industry. The items are formulated as statements covering individual, management, and organizational perspectives (e.g., ‘‘I. . .,’’ ‘‘My immediate manager. . .,’’ ‘‘My Company. . .’’). The answer format is a 4-point Likert scale ranging from 1 to 4 (1 = strongly disagree, 2 = disagree to 3 = agree and 4 = strongly agree) and including a separate not applicable box. A typical example item is ‘‘My immediate manager listens to our views on safety’’ (facet: Management’s Safety Attitudes). The SCD-Q closes with 10 ‘‘tombstone’’ (demographic) questions referring to occupational role; region; length of service; additional managerial, supervisory, training, or specific safety roles; time in those roles; full-/parttime employment; and age group – all previously found to be potential influencers on safety culture.

An LSC consists of operational staff members responsible for the implementation and maintenance of a mature safety culture in support of occurrence reporting and investigations. Their principal duty, among others, ‘‘is organising safety briefings for the collection and dissemination of national and international safety-related information in close cooperation with local unit management’’ (Austro Control, 2013, p. 6).

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Ó 2015 Hogrefe Publishing

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Safety-Related Reconstruction Interview

Research Variables Safety culture development scales (facets) were used as independent variables (IVs) and demographic data as control variables (CVs), and retrospectively reported SRBs served as dependent variable (DV).

Procedure

May 2012 (following the SCD-Q administration). Participation was completely randomized (initials written down on paper, drawn by lot) to make sure that each unit and location was representative in terms of participants and personal attitudes toward safety (e.g., negative, neutral, positive). Additionally unit chiefs/regional managers were invited to volunteer for an interview to set a positive example and personally encourage participation by their teams. Participants were allocated 1 hr during working hours to take part in the interview. Interviews were recorded using standard smart-phones. All participants had to sign a consent form declaring that they understood the eligibility requirements (voluntary basis, use of unique code to match data, anonymity guaranteed, no access to raw data for their employer).

Analysis Questionnaire The raw questionnaire data were entered into SPSS and analyzed following a stepwise questionnaire development procedure, as outlined in Kallus (2010). Data labeled not applicable were coded as missing values. Missing values were deleted listwise, unless otherwise stated. To demonstrate construct validity, data were subject to an examination of their internal consistency (reliability), principal component analysis (PCA) including varimax rotation with Kaiser normalization as well as Pearson’s correlations and logistic regression using IBM SPSS statistics 21.

Questionnaire Interview All operational safety-relevant ATM personnel and their direct line managers were invited to participate in this study by a personal letter written by the chief officer of operations. The questionnaire package was distributed to participants in March 2012 in paper-and-pencil version in English, together with a translated version in German (mother tongue) for reference. Participants were instructed to complete the English version to allow future data comparisons across other ANSPs in the CANSO region. Participants were allocated 15–20 min to complete the questionnaire during working hours. It was agreed with top management, unit chiefs/ regional managers, and staff union representatives that participation was entirely voluntary, and confidentiality as well as anonymity would be ensured through use of a unique personal code. This code was used to match data sets (questionnaire and interviews) for further analysis and was replaced with consecutive numbering afterward to make sure that data stored would be deidentified. The questionnaire package was administered a second time to LSC members in December 2013 to confirm that components underlying safety culture were stable over time.

A total of 81 SRECOs were conducted by four professionally trained and independent interviewers in the period March–

Interview recordings were transcribed following a standard transcription procedure (McLellan, MacQueen, & Neidig, 2003). The interview coding system for analysis followed a qualitative content analysis according to Mayring (2010). In total, 97 variables were derived from responses to the original 25 open interview questions. Some of these variables corresponded directly to an interview question; other variables corresponded to spontaneous answers given by participants. The occurrence of each variable (yes = 1, no = 0, missing = not asked) was enumerated for all participants (N = 81) and entered into SPSS for further analysis. Reliability analysis of four independent raters on a random sample of three interviews resulted in an adequate intraclass coefficient (Cronbach’s alpha) of .79. For further analysis, 38 (of the overall 97) variables that corresponded directly to interview questions were selected to be related to the six SRBs derived from the literature. The final allocation is found in the Appendix. In a final step, the questionnaire and interview data files were merged based on their unique six-digit IDs. Unfortunately it was only possible to match 51 (out of 81) codes across the different occupational groups (40.4% ATSPEs, 28.8% ATCOs, and 30.8% METs). To demonstrate criterion validity, 51 cases were finally tested on significant correlations (Spearman Brown) and prediction of group membership (logistic regression).

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Interview

Original Article

The SRECO was developed to supplement the CANSO SCD-Q. The objective of SRECO is to assess the organizational context in which participants operate and gain insight into actual behaviors on the job relevant to safety culture (criterion). SRECO is based on the Reconstruction Interview for the Integrated Task Analysis (ITA) for Air Traffic Controllers (Kallus, Barbarino, & VanDamme, 1998) associated with the previous work shift. The first part includes 12 open questions related to specific safety-relevant situations that have occurred during the previous work shift and how they were managed with respect to operational safety. The second (general) part of the interview consists of 13 open questions regarding safety culture and SRB on the job such as safety citizenship behavior, safety communication, safety compliance, safety knowledge, safety leadership, and resilient safety behavior. In total, the interview consists of 25 questions lasting on average of 30–40 min.

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Table 2. Reliability of Safety Culture Scales on trial (N = 50) and main sample (N = 282) #

Component

Original item allocation

k (43)

Cronbach’s alpha (N = 50)

k (24)

Cronbach’s alpha (N = 282)

1 2 3 4 5 6 7 8

Informed culture Reporting culture Just culture Learning culture Flexible culture Management’s safety attitudes Safety-related behavior Risk perception

(1a),5,18,32,34,42 2,11,12,22,24,44 9,17,26,29,35,41 6,19,20,21,38 3,10,23,30,40 8,25,28,33,37 7,13,16,27,36,39 4,14,15,31,43

5 6 6 5 5 5 6 5

.614 .676 .490 .675 .585 .106 .702 .435

4 4 4 3 3 0 4 2

.621 .524 .443 .521 .642 n/a .685 .513

Notes. aItem 1 was not part of the original CANSO questionnaire and was introduced as an ice breaker item in the trial sample. Item 1 was therefore excluded from analysis and not used in the main sample. Items marked in grey were removed in the course of the initial validation exercise due to insufficient reliability.

Original Article

Results Construct Validity (H1) Trial Sample The original CANSO questionnaire consisted of a shortlist of 40 items. These were expanded into 44 items in a basic review before administering the SCD-Q (Heese, Kallus, Artner, & Marek, 2011) to avoid asking questions for two themes at the same time in one question. Reliability analysis on the eight safety culture scales on the trial sample (N = 50) is presented in Table 2. The eight safety culture scales did not reach an acceptable level of internal consistency (Cronbach’s a > .70) according to Tabachnick and Fidell (2007). Twenty items (marked in grey in Table 2) with corrected item-total correlations (ri (T-i)) smaller than .30 had to be excluded from further analysis. A PCA was conducted based on the remaining 24 items to explore the number of underlying factors. Despite the small sample size (N = 50) the KaiserMeyer-Olkin (KMO) coefficient for sampling adequacy was 0.69, just above the required minimum (0.60). Bartlett’s test of sphericity confirmed the data was appropriate for factor analysis (v2 = 576.2, df = 276, p = .000). A PCA based on eigenvalue with varimax rotation extracted eight factors accounting for 76.35% of total variance. However, clustering did not agree with the original CANSO subject matter experts’ item allocation to the eight scales. A new (aggregated) two-factor solution achieving sufficient Cronbach’s alpha values for internal consistency was attempted during initial validation (Heese & Kallus, 2012). The overall goal of the initial validation exercise was to reduce the total number of items from 43 to a maximum of 32 items (4 items per subscale). As a result of the exclusion of 20 items due to insufficient reliability, the scale management’s safety attitudes was lost altogether, and the items in the remaining scales were reduced (compare Table 2). To accommodate for some of the items lost, it was decided to 2

introduce eight new items from the original CANSO safety culture database. The eight items once again were selected in consultation with subject matter experts. Three new items were included to accommodate for the lost dimension (management’s safety attitudes), three new items were introduced for risk perception, one item was added for flexible culture, and another one was chosen for learning culture. All items have been used previously to assess safety culture by ANSPs and/or universities in the aviation context, but have not yet been assessed based on their psychometric properties. Main Sample Following initial validation, the revised version of the SCDQ consisted of 32 items. Twenty-four were equal to the original questionnaire validated in the trial; eight new items were added. Once again reliability analysis of the eight scales on the main sample (N = 282) did not achieve the acceptable level of internal consistency (Cronbach’s a > .70) (see Table 2). It was therefore decided to perform another PCA with varimax rotation on item level to explore the underlying factor structure. According to Tabachnick and Fidell (2007), items with loadings smaller than .40 were suppressed. The sample was confirmed to be adequate for factor analysis (KMO = .88, v2 = 1,970.9, df = 496, p = .000). The PCA resulted in a five-factor solution (32 items) accounting for 50.64% of total variance. The final rotated factor solution is presented in Table 3. Based on the PCA, three items (i3, i15, i16) were removed, because they did not load on any of the five factors. A fourth item (i20) loaded poorly on informed culture, but did not fit into this category based on its content. This item was also removed to ensure face validity. Descriptives and intercorrelations for the 28 items on five subscales are reported in Table 4. All safety culture scales showed highly significant intercorrelations, r(281) = .48-.64, p < .01, demonstrating convergent validity. However, correlations between just culture2 and the remaining scales were only

‘‘Just culture’’ means a culture in which front line operators or others are not punished for actions, omissions or decisions taken by them that are commensurate with their experience and training, but where gross negligence, wilful violations and destructive acts are not tolerated’’ (EU 691/2010, L291/3; European Commission, 2014).

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Ó 2015 Hogrefe Publishing

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Table 3. Rotated component matrix for SCDQ-32 (N = 282) Component 1

.758 .666 .632 .594 .591 .554 .533 .521 .435

.416

.770 .656

Original Article

(18) Top management shows visible appreciation for the knowledge and experiences of all employees. (27) There is a positive safety culture led by top management in this company. (12) Top management has a clear understanding of risks associated with all kind of operations. (8) I believe that all levels of management generally trust employees. (28) Management is interested in employees’ views on safety. (9) Top management has adequately prepared employees for changes to operations. (2) When a problem arises, it is the most competent person or team who gets to solve it. (1) Employees ideas are shared with the appropriate level of management. (29) Employees are encouraged to put forward ideas and suggestions for improvements concerning work. (24) Everyone in our unit is kept informed of any changes which may affect safety. (32) Staff receive feedback regarding the progress and results of safety hazards that they report. (26) Staff is actively involved in identifying and resolving safety concerns. (31) Staff is always involved in the management of safety matters. (4) The company keeps me well-informed about safety issues. (11) Management implements long-term solutions to safety issues that address the underlying causes and go beyond the current minimum demands, when necessary. (22) A risk management process is used to manage any safety risks associated with changes to the operational environment. (20) Managers have sufficient authority to stop unsafe practices. (16) Before finalizing the decision process, people with differing opinions get the opportunity to present their opinion. (3) In my work, I am expected to take some risk to get the job done better. (14) There is a good, open exchange of information between my immediate manager and me. (13) My immediate manager listens to our views on safety. (30) I believe that employees generally trust their immediate manager. (5) Immediate managers encourage employees to continually seek improvements that will enhance performance and safety. (6) My immediate manager always follows through on safety related matters until they are resolved. (7) Managers actively encourage safe work practices. (23) Management goes above and beyond regular minimums when it comes to issues on safety. (17) Everyone is given sufficient opportunity to make suggestions regarding safety issues. (19) It is acceptable to make suggestions for change concerning somebody else’s area of responsibility. (25) I am comfortable reporting safety concerns with no fear of punishment. (15) Employees learn from mistakes made by their coworkers on how to do their work better. (10) I always report my own mistakes that might impact on safety in future. (21) I know what behavior is acceptable and what behavior is unacceptable in the company.

.619 .571 .564 .527 .486 Remove Remove Remove .807 .767 .653 .603

.466

.562 .527 .454 .647 .527 .509 Remove .758 .719

Notes. Extraction: principal component analysis; Rotation: varimax with Kaiser normalization; Rotation converged in eight iterations. Items marked in grey show item allocation to extracted factors.

moderate, r(281) = .22-.29, p < .01, suggesting discriminant validity for the just culture concept. The five factor solution was subsequently cross validated on all available sub samples (n = 75 ATSEPs, KMO = .40, v2 = 805.0, df = 496, p = .000; n = 155

ATCOs, KMO = .83, v2 = 1368.4, df = 496, p = .000, n = 45, KMO not available) to verify that the five factors were stable across three different occupational groups. The results confirmed the five-factor solution for the ATSEP sample to account for 61% of total variance and the

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Table 4. Descriptives and intercorrelations of the CANSO Safety Culture Development Questionnaire (SCD-Q) Scales (N = 282) #

Components

k (28)

SCD1

SCD2

SCD3

SCD4

SCD5

1 2 3 4 5

Informed culture Reporting and learning Just culture Flexible culture Mgt’s safety attitudes

7 5 2 8 6

1.63 1.77 2.16 1.39 1.87

0.51 0.48 0.47 0.50 0.56

1 .510** .284** .496** .482**

1 .250** .541** .635**

1 .221** .291**

1 .508**

Note. Mgt = management; SCD = safety culture development. **Correlation significant on the .01 level, listwise: N = 281.

Original Article

Table 5. Reliability of Safety Culture Development Scales across occupational groups (N = 282) #

Item Numbers*

k (28)

Cronbach’s a N = 282 (OVERALL)

Cronbach’s a n = 75 (ATSEPs)

Cronbach’s a n = 155 (ATCOs)

1 2 3 4 5

4,11,22,24,26,31,32 1,5,17,19,25 10,21 2,8,9,12,18,27,28,29 6,7,13,14,23,30

7 5 2 8 6

.817 .723 .388 .843 .837

.752 .635 .537 .837 .809

.825 .745 .397 .844 .826

Cronbach’s a n = 45 (METs) .815 .762 .144 .845 .845

Notes. *Items 3, 15, 16 and 20 were removed. 1 = Informed Culture; 2 = Reporting/Learning Culture; 3 = Just Culture; 4 = Flexible Culture; 5 = Management’s Attitudes towards Safety. Values marked in grey show Cronbach’s Alpha values for internal consistency below .70 (cut-off score).

ATCO sample to account for 53% of total variance. The five-factor solution could not be computed on the MET sample, because it did not fulfill the minimum sample size required to run a PCA (Tabachnick & Fidell, 2007). Results from subsequent reliability analysis of the extracted five factors on all samples are reported in Table 5. All components except just culture (marked in gray) achieved the acceptable level of internal consistency (Cronbach’s a > .70) indicating that the five-factor solution was stable across different occupational groups except for just culture. The component reporting and learning culture just missed the .70 cutoff in the ATSEP sample. To verify whether the five extracted components indeed measured one latent variable safety culture, another PCA for the main sample (N = 282) on component level was conducted (KMO = .79, v2 = 366.1, df = 10, p < .001. The PCA based on eigenvalue confirmed one underlying factor. However, when setting the number of factors to extract to two as suggested by the initial validation, the component just culture loaded on a second latent component.

subsample 2012 (n = 48) and the 2013 (N = 32) sample. Table 6 summarizes Cronbach’s alpha values for internal consistency. In the 2012 sample, a Cronbach’s a > .70 for internal consistency was achieved for all safety culture scales except for just culture. In the 2013 sample, a sufficient Cronbach’s alpha for internal consistency (see Table 6) was only achieved for flexible culture (a = .79) and management’s safety attitudes (a = .71). Moderately significant Pearson’s correlations were found between t1 (2012) and t2 (2013) relating to informed culture and reporting and learning culture, r(32) = .35, p < .05; just culture and flexible culture, r(32) = .35, p < .05; just culture and management’s safety attitudes, r(32) = .41, p < .05; as well as reporting and learning culture and management’s safety attitudes, r(32) = .41, p < .05. Retest reliability was only achieved for two safety culture scales (flexible culture and management’s safety attitudes).

Criterion Validity (H2)

Finally retest reliability of the five safety culture scales was tested on the follow-up sample regarding participants holding additional specific safety roles. This sample was chosen because participants holding additional specific safety roles are especially trained to promote safety culture and are hence considered to assess safety culture within the organization more accurately. For the analysis, internal consistencies and correlations were calculated comparing the associated

H2 postulated that safety culture scores would predict SRB. The six SRBs as well as an overall SRB score (sum across all six SRBs) were used as criterion variable to perform regression analysis. Descriptives and intercorrelations for the six SRBs as well as the overall SRB are reported in Table 7. On average, 18 SRBs (SD = 4.2) were stated across all occupational groups (ATCOs, ATSEPs, and METs) per interview associated with the previous work shift. The overall SRB score correlated significantly with all SRBs except safety compliance. When looking at the detailed SRBs, only resilient safety behavior and safety communication, r(49) = 30, p = .033, and resilient safety

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Ó 2015 Hogrefe Publishing

Follow-Up Sample

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Table 6. Reliability of Safety Culture Scales across assessment date (2012/2013) #

Safety Culture Development Scales

k (28)

1 2 3 4 5

Informed culture Reporting and learning Just culture Flexible culture Management’s safety attitudes

7 5 2 8 6

Cronbach’s a t1 (2012) N = 48

Cronbach’s a t2 (2013) N = 32

.821 .794 .348 .847 .876

.549 .604 .258 .787 .705

Note. Values marked in grey show Cronbach’s alpha values for internal consistency below .70 (cut-off score).

Table 7. Descriptives and intercorrelations between 38 safety-relevant behaviors (N = 51) Safety-relevant behaviors

k (38)

1 2 3 4 5 6 –

SCB Safety communication Safety compliance Safety knowledge Safety leadership Resilient safety behavior Overall SRB

4 6 4 7 9 8 38

1.73 2.39 1.83 3.81 4.22 4.08 18.08

0.95 1.35 0.87 1.41 1.83 1.64 4.20

1 1 .082 .053 .273 .083 .267 .474**

2 1 .175 .082 .192 .304* .534**

1 .143 .036 .111 .151

1 .089 .276* .541**

1 .022 .561**

1 .675**

Notes. SCB = safety citizenship behavior; SRB = safety-relevant behavior. (+) Percentage of favorable (k) number of items. *Significant at the .05 level; listwise N = 49.

behavior and safety knowledge were significantly positively correlated, r(49) = .38, p = .055. All other correlations were not significant. Correlation of SRBs and Safety Culture Scales In preparation for regression analysis, the relationship between the overall SRB score and the six detailed SRBs with the five safety culture scales was investigated. Only one weakly significant Pearson’s correlation between safety communication and just culture was found, r(49) = .29, p < .05. This correlation disappeared when controlling for occupational role, region, and length of service in a partial correlation. In addition, another moderate partial correlation between safety knowledge and informed culture, r(43) = .31, p < .05, was revealed. Since Pearson’s correlation analysis did not pick up any further correlations, the relationship between each SRB and the safety culture scales was investigated using bivariate scatterplots. Results showed that there may be a curvilinear relationship between all five SRBs (except safety compliance) and just culture. All other SRBs showed a nonlinear relationship with safety culture scales, which would not have been captured by Pearson’s correlation analysis (Tabachnick & Fidell, 2007).

r(49) = .33, p = .018, and additional training/instructing role, r(49) = .31, p = .033; and negatively with length of service, r(50) = .33, p = .019, and age, r(50) = .34, p = .016. Further significant positive correlations between occupational role and safety citizenship behavior (SCB), r(50) = .30, p = .035, safety knowledge, r(51) = .60, p < .000, and resilient safety behavior, r(49) = .40, p = .004, were found. Safety Leadership correlated negatively with age, r(49) = .38, p = .007, and positively with additional managerial role, r(49) = .47, p = .001. Additional training/instructing roles correlated positively with safety knowledge, r(49) = .35, p = .013, and resilient safety behavior, r(47) = .34, p = .021. Logistic Regression

In a next step, correlations (missing values were excluded pairwise) with demographic data were checked. Overall the SRB score correlated positively with occupational role, r(51) = .41, p = .003, additional managerial role,

After investigating correlations between SRBs and potential predictors, data were entered into a logistic regression model. Logistic regression for discrete outcome variables was chosen, because it is relatively free of restrictions (Tabachnick & Fidell, 2007, p. 441) with respect to sample size, normality, linearity, and homoscedasticity. Although the overall SRB score fulfilled requirements for linear regression, the detailed SRBs did not satisfy these prerequisites. To run a logistic regression, SRBs were converted into a dichotomous variable by performing a median split (low/ high SRB). Then the five safety culture scales were entered as covariates in one block (ENTER method), and demographics that correlated significantly were entered as covariates in the second block, where applicable categorical variables (additional managerial and training/instructing role) went into the third block. The converted overall

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Correlation of SRBs and Demographic Data

Original Article

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

SRB score and the six detailed SRBs went in as DVs separately for each regression model. The logistic regression model for overall SRB score demonstrated significant standardized beta-coefficients for demographic data. However, only the regression model for safety culture scales was not significant. Therefore the five safety culture scores were also converted into a grouping variable (favorable = agree and strongly agree; unfavorable = disagree and strongly disagree) and entered into six separate regression models. The Hosmer-Lemeshow goodness of fit test statistic was calculated for all six models and did not prove significant, indicating that the observed rates versus the expected rates were a good match. For regression model 1, grouped safety culture development (SCD) scales (Block 1) and occupational role (Block 2) were entered as IV, and grouped SCB was entered as DV. This time just culture, b = 2.506, v2(5, N = 50) = 11.812, p = .037, R2 = 28.4, Wald = 6.193, p = .013, was found to be a significant predictor for SCB3 resulting in a 28.4% improvement in the classification table. For regression model 2, grouped SCD scales (Block 1) and safety communication (DV) were entered. Reporting and learning, b = 1.773, v2(5, N = 50) = 11.712, p = .039, R2 = 29.7, Wald = 3.645, p = .056, and just culture, b = 1.743, v2(5, N = 50) = 11.712, p = .039, R2 = 29.7, Wald = 3.746, p = .053, tended to significantly predict safety communication, adding 29.7% value. For regression model 3, grouped SCD scales (Block 1) and safety compliance (DV) were explored. Management’s safety attitudes significantly predicted safety compliance, b = 1.718, v2(5, N = 50) = 12.193, 2 p = .032, R = 29.5, Wald = 3.948, p = .047, adding 29.5% added value. For model 4, grouped SCD scales (Block 1), occupational role (Block 2), training role categorical (Block 3) and grouped safety knowledge were entered. For model 5, grouped SCD scales (Block 1), age (Block 2), and additional managerial role categorical (Block 3) and safety leadership went in. For model 6, grouped SCD scales (Block 1), occupational role (Block 2), additional training role categorical (Block 3), and resilient safety behavior (DV) were put in. Models 4–6 did not produce significant results for SCD scale predictors. H2, assuming criterion validity (safety culture scores predict SRB on the job), could therefore only be partly be confirmed for three of five components: reporting and learning culture, just culture, and management’s safety attitudes. No significant results could be found for informed and flexible culture.

Figure 1) within ATM, as required by the CANSO Standard of Excellence for SMS (CANSO, 2009, 2014). H1 postulated that CANSO components (Table 1) are reliable and valid with respect to safety culture across different occupational groups and assessment dates. Initial validation results on a trial sample (N = 50) did not achieve sufficient Cronbach’s alpha values of > .70 for internal consistency (Tabachnick & Fidell, 2007) and also could not confirm eight underlying components. A further 20 items had to be excluded during initial validation due to insufficient reliability values. A new and improved version of the CANSO SCD-Q, with 32 items, underwent further validation on the main sample (N = 282) and again did not prove reliable (Table 2). Clustering again did not agree with the original item allocation by CANSO. Therefore a PCA on item level was calculated and revealed five of the original eight components to be valid with respect to safety culture (Table 3). The five components underlying safety culture were informed culture, reporting and learning culture (merged together), just culture, flexible culture, and management’s safety attitudes. A cross-validation on the ATCO and ATSEP samples confirmed the five-factor solution (Table 4). Results generally resonated with recent findings based on confirmatory factor analysis by Mearns et al. (2013), who have reviewed their initial safety culture model moving away from Reasons’s Five (Reason, 1997; informed, reporting, just, learning, flexible culture) to higher order factors such as involvement in safety, reporting and learning and prioritization of safety. All components except just culture were found to be reliable across occupational groups (Table 5). One possible explanation for the lack of reliability is that just culture was left with only two reliable items. Analysis on factor levels suggests just culture to be loading on a latent variable different from safety culture. One possible explanation is that just culture had the lowest intercorrelations with the remaining safety culture facets (Table 4). Nevertheless it was decided to keep the component just culture, as it was deemed highly relevant for the development of a positive and proactive safety culture (CANSO, 2009; European Commission, 2013). Retest reliability on a sample of 32 participants holding specific safety roles (LSCs) confirmed flexible culture and management’s safety attitudes to be stable across two assessment dates (Table 6). Informed culture and reporting and learning culture only proved reliable for the 2012 sample. Again just culture did not achieve sufficient reliability.

Discussion

Criterion Validity (H2)

Construct Validity (H1)

H2 hypothesized that safety culture scales would be able to predict any retrospectively reported SRB on the ATM job, implying that they are somehow related. Six SRB groups derived from SRECOs related to the previous shift and an overall SRB score across these six behaviors served as cri-

The CANSO safety culture development concept (questionnaire and interview) was introduced as one option to regularly measure and improve safety culture (compare 3

‘‘Safety Citizenship Behaviours can be defined as behaviours that are discretionary, not directly or explicitly recognised by the formal reward system, and that in the aggregate promote the effective functioning of the organisation’’ (Didla, Mearns, & Flin, 2009, p. 476).

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Ó 2015 Hogrefe Publishing

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Therefore, validation mostly had to be based on reliability analyses to demonstrate that components were stable across occupational groups and assessment dates. Tabachnick and Fidell (2007) propose a minimum of 10 observations per variable as a rule of thumb to conduct PCA. In this study, the initial 44 items would have required a minimum sample size of 440 participants for the trial. The sample size of the main study (N = 282) to validate the final item set of 32 items would have required 320 participants. As a minimum requirement, the KMO coefficient for sampling adequacy was used to test appropriateness of factors analysis wherever sample size was too small. Secondly, components with only two items (just culture) are generally not kept as separate scales if they do not achieve the minimum Cronbach’s alpha for internal consistency of .70. However, since just culture is considered as one of the main key performance indicators for safety by CANSO and the European Commission, it was decided to keep the component separate. Results related to just culture must therefore be replicated in future studies, once the facet has been confirmed to be reliable. Thirdly, SRB was still assessed based on subjective self-reports in interviews. Although the objectivity of selfreported behaviors derived from interviews associated with the previous work shift is deemed higher compared with subjective attitudes assessed through questionnaire, underlying behavior measurements in this study still remained subjective. Fourthly, logistic regression was based on a very small sample size, and significant results could only be achieved by grouping safety culture components and SRBs. Such data manipulation impacts the explanatory power of results and must therefore be taken into account when interpreting logistic regression results. Lastly, the use of a unique personal code to match questionnaire and interview data needs to be reconsidered to avoid unnecessary data loss, as participants tend to forget or choose different codes over time.

Conclusion and Outlook

Firstly, the sample sizes in the trial (N = 50) and follow-up sample (N = 32) were too small to allow factor analysis.

The findings reported and discussion above add significantly to existing research (which has thus far mainly focused on workplace safety) by providing initial psychometric validation of the CANSO safety culture development concept based on SRB related to the ATM job. This study sets the foundation for future studies to demonstrate full construct and criterion validity by providing a list of retrospectively reported SRBs that can be objectively observed on the frontline (see Appendix). The CANSO safety culture model (Figure 1) links the safety culture levels to SMS maturity levels, offering a solid basis for managers, safety experts, and safety researchers to kick off safety culture improvement programs. However, the correlations between safety culture levels and actual SMS maturity scores and safety performance (e.g., occurrence data) within organizations remain to be demonstrated. This can only be achieved if CANSO members who have implemented a standardized safety culture survey approach agree to share their data. It is therefore recommended that the CANSO Safety Standing

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Methodological Limitations

Original Article

terion variables. Safety culture assessment tools so far have only been validated based on safety attitudes and safety behavior reported in subjective questionnaires (Evans et al., 2007; Mearns et al., 2001; Steyrer et al., 2011). This study is the first to attempt to relate safety attitudes measured in the SCD-Q to SRBs associated with a previous work shift. On average, participants reported 18 SRBs independent from occupational group. SRBs only showed some significant intercorrelations (Table 7), indicating low convergent validity. According to Lu and Yang (2011), Pearson’s correlations based on survey responses between safety compliance, safety communication and safety participation should be at least r = .50. SRBs also only correlated with two of five safety culture scales (informed culture and just culture). Examination of bivariate scatterplots showed that there may be a curvilinear relationship between all five SRBs (except safety compliance) and just culture. All other SRBs showed a nonlinear relationship with safety culture scales. H2, hypothesizing that safety culture scores would predict SRB on the job, was partly confirmed for three of five components: reporting and learning culture, just culture, and management’s safety attitudes. No significant results could be found for informed culture or flexible culture. Significant regression results found a positive standardized beta-coefficient for just culture groups (favorable/unfavorable) as predictor for SCB, and a negative standardized beta-coefficient for management’s safety attitudes groups as predictor for safety compliance. As the component just culture did not achieve sufficient reliability, results must be interpreted with care. A positive beta-coefficient means that the value of the DV is expected to increase as the value of the predictor increases. Conversely, negative beta-coefficients point to decreased DV values as the predictor value increases. The negative beta-coefficient for safety compliance is in line with recent research by Bosak et al. (2013), who investigated safety climate as predictors for risk behavior in the chemical industry. They also established ‘‘that employees’ risk behaviour is negatively correlated to management commitment and priority of safety’’ (Bosak et al., 2013, p. 356). Although there is some indication that safety culture scales predict SRB, full criterion validity demonstrating that all safety culture scores predict SRB is yet to be confirmed in future studies. Mearns et al. (2013) also claim that true criterion validity demonstrating that safety culture interventions predict safety performance remains difficult due to the lack of experimental control in the dynamic field of ATM. Most organizations have various interventions (new tools, process, training) and other potentially influencing factors (major reorganizations, staff union negotiations, cost cutting) going on at the same time. It is therefore almost impossible to solely relate increased safety performance to one program (i.e. safety culture improvement) alone.

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Committee promote a standardized safety culture assessment process across their members. This is crucial to ensure further validation of the CANSO safety culture concept on a larger sample size across different national and cultural settings. Although the majority of researchers tend to be very successful in producing significant results through large sample sizes in subjective questionnaires, the behavior assessed always remains self-reported and highly subjective. Proactive safety management and resilient engineering approaches therefore promote over-the-shoulder observations as part of day-to-day safety surveys (Isaac, Brooks, Jordan, & McCabe, 2009). Future research is proposed to integrate SRBs on the job (Appendix) into objective overthe-shoulder observations to further validate the safety culture development concept.

Austro Control. (2013). Organisation des ACG Local Safety Committee [Organisation of the ACG Local Safety Committee], (Unpublished internal document). Bosak, J., Coetsee, W. J., & Cullinane, S. J. (2013). Safety climate dimensions as predictors for risk behaviour. Accident Analysis and Prevention, 55, 256–264. Bundesstelle für Flugunfalluntersuchung. (2004). Untersuchungsbericht Überlingen/Bodensee [Investigation report Überlingen / Lake Constance]. Retrieved from http:// www.bfu-web.de/EN/Publications/Investigation%20Report/ 2002/Report_02_AX001-1-2_Ueberlingen_Report.pdf?__ blob=publicationFile Civil Air Navigation Services Organisation. (2009). The CANSO standard of excellence in safety management systems. Hoofddorp, NL: Author. Retrieved from http://www. canso.org/cms/streambin.aspx?requestid=E71C36FB-88044FD1-B412-3296E2510EA8 Civil Air Navigation Services Organisation. (2011). The CANSO Safety Culture Database (Unpublished data). Civil Air Navigation Services Organisation. (2014). The CANSO Standard of Excellence in Safety Management Systems. Retrieved from https://www.canso.org/sites/default/files/ CANSO%20Standard%20of%20Excellence%20in%20SMS_1. pdf Clarke, S. (2006). A meta-analytic review of safety climate and safety performance. Journal of Occupation Health Psychology, 11, 315–327.

Cook, D.A., & Beckman, T.J. (2006). Current concepts in validity and reliability for psychometric instruments. Theory and application. The American Journal of Medicine, 199, 166.e7–166.e16. Cooper, M. D., & Phillips, R. A. (2004). Exploratory analysis of the safety climate and safety behavior relationship. Journal of Safety Research, 35, 497–512. Didla, S., Mearns, K., & Flin, R. (2009). Safety Citizenship Behaviour: A proactive approach to risk management. Journal of Risk Research, 35, 497–512. Didla, S., Mearns, K., & Flin, R. (2010). Safety citizenship behaviour: A proactive approach to risk management. Journal of Risk Research, 12, 475–483. Ek, A., Akselsson, R., Arvidsson, M., & Johansson, C. R. (2003). Safety culture in air traffic management: Air traffic control. Paper presented at the 5th USA/Europe ATM 2003 R&D Seminar, 23–27 June, Budapest, Hungary. Retrieved from http://www.atmseminarus.org/seminarContent/ seminar5/papers/p_042_S.pdf Ek, A., & Arvidsson, M. (2002). Safety culture in the Swedish Air Navigation Services. Paper presented at the European Academy of Occupational Health Psychology, Vienna, 4–6 December. EUROCONTROL. (2008). Safety culture in air traffic management: A white paper. EUROCONTROL/FAA as part of the AP15 on Safety Research. Available upon request from Barry Kirwan (AP15 Cochair). European Commission. (2011). Commission Implementing Regulation (EU) No 1216/2011 of 24 November 2011 amending Commission Regulation (EU) No 691/2010 laying down a performance scheme for air navigation services and network functions. Retrieved from http://eur-lex.europa.eu/Lex UriServ/LexUriServ.do?uri=OJ:L:2011:310:0003:0005:EN: PDF European Commission. (2013). Commission implementing regulation (EU) No 390/2013 of 3 May 2013 laying down a performance scheme for air navigation services and network functions. European Commission. (2014). Regulation (EU) No 376/2014 of the European Parliament and of the Council of 3 April 2014 on the reporting, analysis and follow-up of occurrences in civil aviation, amending Regulation (EU) No 996/2010 of the European Parliament and of the Council and repealing Directive 2003/42/EC of the European Parliament and of the Council and Commission Regulations (EC) No 1321/2007 and (EC) No 1330/2007. Retrieved from http://eur-lex. europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014 R0376&from=EN Evans, B., Glendon, A. I., & Creed, P. A. (2007). Development and initial validation of an Aviation Safety Climate Scale. Journal of Safety Research, 38, 675–682. Fleming, M. (2000). Safety culture maturity model: Offshore Technology report 2000/049. Edinburgh, UK: Health and Safety Executive. Retrieved from http://www.hse.gov.uk/ research/otopdf/2000/oto00049.pdf Griffin, M. A., & Neal, A. (2000). Perceptions of safety at work: A framework for linking safety climate to safety performance, knowledge and motivation. Journal of Occupational Health Psychology, 5, 347–358. Guldenmund, F. W. (2000). The nature of safety culture: A review of theory and research. Safety Science, 34(1–3), 215–257. doi: 10.1016/S0925-7535(00)00014-X Guldenmund, F. W. (2010). (Mis)understanding safety culture and its relationship to safety management. Risk Analysis, 30, 1466–1480. doi: 10.1111/j.1539-6924.2010.01452.x Heese, M. (2012). Got the results, now what do you do? Safety culture transformation from theory into practice. Aviation Psychology and Applied Human Factors, 2(1), 25–33. doi: 10.1027/2192-0923/a000020

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Ó 2015 Hogrefe Publishing

Original Article

Acknowledgments The authors are indebted to the Austro Control Safety, Security and Quality Department for the continuous support and funding of the safety culture development project. Special thanks to the CANSO safety culture regional champions for providing their subject matter expertise in reviewing the validated item set. Heartfelt thanks go to all unit chiefs and regional managers for their outstanding personal commitment in getting the message across to their troops and following up on response rates. Thanks to all operational teams for their important contributions made by completing the questionnaire and interviews. Thanks also to Kerstin Gaisbachgrabner and Sylvia Peißl for the data assessment, entry, administration, and analysis support.

References

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Mayring, P. (2010). Qualitative Inhaltsanalyse. Grundlagen und Techniken [Qualitative content analysis: Fundamentals and techniques] (11th ed.). Beltz: Weinheim. McLellan, E., MacQueen, K. M., & Neidig, J. L. (2003). Beyond the qualitative interview: Data preparation and transcription. Field Methods, 15, 63–84. Mearns, K. J., & Flin, R. (1999). Assessing the state of organizational safety: Culture or climate? Current Psychology, 18(1), 5–17. doi: 10.1007/s12144-999-1013-3 Mearns, K. J., Kirwan, B., & Kennedy, R. (2009). Developing a safety culture measurement toolkit (SCMT) for European ANSPs. Napa, CA: ATM. Retrieved from http://www. atmseminar.org/seminarContent/seminar8/papers/p_021_HF. pdf Mearns, K., Kirwan, B., Reader, T. W., Jackson, J., Kennedy, R., & Gordon, R. (2013). Development of a methodology for understanding and enhancing safety culture in air traffic management. Safety Science, 53, 123–133. Mearns, K., Whitaker, S. M., & Flin, R. (2001). Benchmarking safety climate in hazardous environments: A longitudinal, interorganizational approach. Risk Analysis, 21, 771–786. O’Dea, A., & Flin, R. (2001). Site managers and safety leadership in the offshore oil and gas industry. Safety Science, 37, 39–57. Parker, D., Lawrie, M., & Hudson, P. (2006). A framework for understanding the development for organisational safety culture. Safety Science, 44, 551–562. Patankar, M. S., & Sabin, E. J. (2010). The safety culture perspective. In E. Salas & D. Maurino (Eds.), Human factors in aviation (2nd ed., pp. 95–122). Amsterdam: Academic Press / Elsevier. Piers, M., Montijn, C., & Balk, A. (2009). Safety Culture Framework for the ECAST Safety Management System and Safety Culture Working Group. Guidance on Organisational Structure. Retrieved from https://easa.europa.eu/essi/ecast/ wp-content/uploads/2011/08/WP1-ECASTSMSWG-Safety Cultureframework1.pdf Reason, J. T. (1997). Managing the risks of organizational accidents. Aldershot, UK: Ashgate. Schein, E. H. (1985). Organizational culture and leadership: University of Illinois at Urbana-Champaign’s academy for entrepreneurial leadership historical research reference in entrepreneurship. Rochester, NY: SSRN. Retrieved from http://ssrn.com/abstract=1496184 Schein, E. H. (2010). Organisational culture and leadership (4th ed.). San Francisco, CA: Jossey Bass. Steyrer, J., Latzke, M., Pils, K., Vetter, E., & Strunk, G. (2011). Development and validation of a patient safety culture questionnaire in acute geriatric units. Gerontology, 75, 481–489. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics, (5th ed.) Pearson Education. Von Rosenstiel, L., & Nerdinger, W. (2011). Grundlagen der Organisationspsychologie. Basiswissen und Anwendungshinweise. 7. Auflage [Basic principles of organizational Psychology]. Stuttgart: Schäffer-Poeschel. Wiegmann, D. A., Zhang, H., Thaden von, T., Sharma, G., & Mitchell, A. (2002). A synthesis of safety culture and safety climate research. Savoy, IL: University of Illinois. Retrieved from http://www.humanfactors.illinois.edu/Reports&PapersPDFs/TechReport/02-03.pdf Zohar, D. (2010). Thirty years of safety climate research: Reflections and future directions. Accident Analysis & Prevention, 42(5), 1517–1522. doi: 10.1016/j.aap. 2009.12.019

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Original Article

Heese, M., & Kallus, K.W. (2012). Marrying safety culture and organizational resilience: Construct validation of the Safety Culture Maturity Questionnaire. In Proceedings of the 30th Conference of the European Association for Aviation Psychology (pp. 303–309). September 25-28, 2012, Villasimius, Italy. Heese, M., Kallus, K. W., Artner, W., & Marek, T. (2011). Safety culture maturity assessment in air traffic management: Readyto-go? Initial validation and correlations with related concepts. In Proceedings of the 16th International Symposium on Aviation Psychology (pp. 7–12). Dayton, OH, May 2–5, 2011. Hofman, D. A., & Morgeson, F. P. (1999). Safety-related behaviour as a social exchange: The role of perceived organizational support and leader-member exchange. Journal of Applied Psychology, 84, 286–296. Hollnagel, E., Pariès, J., Woods, D., & Wreathall, J. (2011). Resilience engineering in practice: A guidebook. Farnham, UK: Ashgate. Hollnagel, E., Woods, D., & Levenson, N. (2006). Resilience engineering: Concepts and precepts. Farnham, UK: Ashgate. Hopkins, A. (2006). Studying organisational cultures and their effects on safety. Safety Science, 44, 875–889. doi: 10.1016/ j.ssci.2006.05.005 International Atomic Energy Agency. (1986). The Chernobyl Accident. International Nuclear Safety Advisory Group (INSAG). Report No. 1 International Atomic Energy Agency. Retrieved from http://www-ns.iaea.org/committees/ insag.asp. International Civil Aviation Organization. (2007). PANS-ATM procedures for air navigation services: Air traffic management (15th ed.) (Doc 4444, 2007, Amdt 5). Montreal, PQ: ICAO. International Labour Organisation. (1998). Technical and ethical guidelines for workers’ health surveillance. (Occupational Safety and Health Series No. 72). Retrieved from http:// www.ilo.org/wcmsp5/groups/public/—ed_protect/—protrav/— safework/documents/normativeinstrument/wcms_177384.pdf Isaac, A., Brooks, V., Jordan, N., & McCabe, M. (2009). Preventing the drift into failure: How do we know when we get it right. EUROCONTROL Hindsight, 8, 25–26. Retrieved from http://www.skybrary.aero/bookshelf/books/568.pdf Kallus, K. W. (2010). Erstellung von Fragebogen [Questionnaire construction]. UTB: Stuttgart. Kallus, K. W., Barbarino, M., & VanDamme, D. (1998). Integrated task and job analysis of air traffic controllers (HUM.ET1.STO1. 1000-REP-03). Brussels: Eurocontrol. Lu, C.-S., & Yang, C.-S. (2011). Safety climate and safety behaviour in the passenger ferry context. Accident Analysis and Prevention, 43, 329–341. Malakis, S., & Kontogiannis, T. (2011). Cognitive strategies in emergency and abnormal situations training: Implications for resilience in air traffic control. In E. Hollnagel, J. Pariés, D. Woods, & J. Wreathall (Eds.), Resilience Engineering in Practice: A Guidebook (8, pp. 101–117). Farnham: Ashgate. Marsh, H. W. (1989). Confirmatory factor analyses of multitraitmultimethod data: Many problems and a few solutions. Applied Psychological Measurement, 13, 335–361. Martínez-Córcoles, M., Gracia, F., Tomás, I., & Peiró, J. M. (2011). Leadership and employees’ perceived safety behaviours in a nuclear power plant: A structural equation model. Safety Science, 49, 1118–1129.

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Received January 13, 2014 Revision received May 5, 2014 Accepted for publication October 15, 2014 Published online April 10, 2015

Original Article

Michaela Schwarz (née Heese) works as safety management and human factors expert at Austro Control GmbH in Vienna and is affiliated with the Department of Psychology at the University of Graz in Austria supervised by K. Wolfgang Kallus. Her main research interest is linking safety culture, safety behavior and organizational resilience, and in air traffic management. She is also vice chair of the Austrian Aviation Psychology Association (http://www.aviation-psychology.at) and Secretary General of the European Association for Aviation Psychology (http:// www.eaap.net).

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3–17

Univ. – Prof. DDr. K. Wolfgang Kallus has been the head of the section on work, organizational, and environmental psychology at the Department of Psychology, University of Graz, Austria, since 1998. His main research interest focuses on psychophysiological measurements in aviation including spatial disorientation training for pilots, human factors in maintenance, and safety culture in air traffic management. He is also chair of the Austrian Aviation Psychology Association (http://www.aviation-psychology.at) and president of Psychophysiology in Ergonomics (http://pie-iea.org).

Correspondence Address Michaela Schwarz (Heese) Austro Control GmbH Schnirchgasse 11 1030 Vienna Austria Tel. +43 5 1703-1038 E-mail michaela.schwarz@austrocontrol.at

Ó 2015 Hogrefe Publishing

M. Schwarz & K. W. Kallus: Safety Culture and Safety-Relevant Behavior in Air Traffic Management

Appendix Table A1. Thirty-eight retrospectively reported safety-relevant behaviors Safety-relevant Behavior 1. Safety Citizenship Behavior

Variables 1. 2. 3. 4.

Proactively seeking safety information Attending regular debriefings of non-routine situations Assisting to others if required Attending to admin tasks/paperwork when bored

5. 6. 7. 8. 9. 10.

Talking about unsafe behavior Addressing unsafe behaviors with supervisor/team leader Providing quality handovers Talking about safety at work regularly Advising trainees/newcomers about safe behavior Advising others on safety standards

3. Safety Compliance

11. 12. 13. 14.

Deviating from procedures in favour of safety Adhering to procedures in account of efficiency Bypassing procedures on a regular basis Applying local procedures

4. Safety Knowledge

15. 16. 17. 18. 19. 20. 21.

Being Being Being Being Being Being Being

5. Safety Leadership

22. 23. 24. 25. 26. 27. 28. 29. 30.

Perceiving safe behavior from trainers/instructors Perceiving individuals are held accountable for safety Perceiving a positive attitude to safety by top management Perceiving safety first led by top management Perceiving a positive attitude by line management Being reminded about safe behavior by supervisors/team leaders Perceiving safe behavior from supervisors/team leaders Receiving sufficient support from managers in terms of safety Receiving safety support from supervisors/team leaders

6. Resilient Safety Behavior

31. 32. 33. 34. 35. 36. 37. 38.

Drawing back on additional resources Using additional planning tools Resuming other roles/responsibilities Proactively increasing safety buffers Restoring safety buffers Managing/adopting to high workload (stress) Managing/adopting to low workload (monotony) Asking for help/advise when uncertain

Ă&#x201C; 2015 Hogrefe Publishing

aware of consequences for unsafe behavior aware of recurrent safety trainings aware of emergency procedures training aware of official error reporting systems aware of reporting channels for bad procedures able to name the local safety committee member aware of having a say in designing procedures

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):3â&#x20AC;&#x201C;17

Original Article

2. Safety Communication

Experimental Evaluation of the AIRWOLF Weather Advisory Tool for En Route Air Traffic Controllers Ulf Ahlstrom

Original Article

Federal Aviation Administration, Atlantic City International Airport, NJ, USA Abstract. The objective of this study was to examine the potential benefits for air traffic controllers from the use of automated weather advisories. In a part-task simulation, we used a weather support tool called Automatic Identification of Risky Weather Objects in Line of Flight (AIRWOLF) that (a) detects conflicts between aircraft and hazardous weather, (b) alerts the controller, and (c) generates automatic weather advisories. During the simulation, air traffic control (ATC) subject matter experts responded to AIRWOLF alerts and either provided weather advisories to pilots via radio in two main conditions or provided data link communication in a third control condition. Automated advisories (a) eliminate the need for a manual production of weather advisories; (b) reduce the production time, voice duration, and overall advisory duration; and (c) reduce the cognitive workload associated with the dissemination of weather advisories. The results showed that the AIRWOLF tool could support air traffic controller weather avoidance actions and provide accurate and timely weather advisories to pilots. The weather advisory tool could support en route controllers for the safe, efficient, and strategic efforts required to handle adverse weather conditions in the en route environment. Keywords: air traffic, data link communication, en route controller, weather advisory, weather hazards, weather probe

Adverse weather conditions are hazardous to flights and contribute to reroutes and delays (Wong, Pitfield, Caves, & Appleyard, 2006). According to McCrea, Sherali, and Trani (2008), 24.83% of all delays in 2005 were weatherrelated, contributing to billions of dollars in additional costs. Among the primary weather hazards are convective activity (i.e., thunderstorms), icing, turbulence, and reductions in ceiling and visibility (Ahlstrom, 2005). Because of the serious impact of hazardous weather on aviation, past research has explored ways to help air traffic management (ATM) decide on safe routes and to improve flight efficiency. For example, with regard to an ATM tool, Steiner, Bateman, Megenhardt, and Pinto (2009) assessed probabilistic aviation impact forecasting during convective storms. The researchers anticipated that the proposed probabilistic aviation impact forecast could be used by air traffic controllers, traffic flow management, and airline dispatchers for increased safety of flight decisions. Similarly, Mitchell, Polishchuk, and Krozel (2006) have proposed a model that determines the probability distribution of airspace capacity given a probabilistic weather forecast. Nguyen, Alam, Tang, and Abbass (2007) explored methods for finding dynamic weather avoidance trajectories in traffic-constrained en route airspace. The researchers found that their approach not only avoids dynamically moving weather and aircraft conflicts but also provides optimal route choices for pilots. For capacity estimation of en route airspace, Krozel,

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35 DOI: 10.1027/2192-0923/a000070

Mitchell, Polishchuk, and Prete (2007) applied a complexity metric under a range of operational and weather conditions. Their experiments revealed a capacity compromise between the benefits of Free Flight (Hoekstra, van Gent, & Ruigrok, 2002) versus unidirectional traffic flows. During weather constraints, the sector throughput was highest for Free Flight which did not restrict the flow of aircraft. However, Free Flight also yielded higher airspace complexity than unidirectional flows. Therefore, while Free Flight conditions improved capacity during adverse weather conditions, throughput capacity was traded off with airspace complexity, which, if too high, will reduce capacity. While most of the research has focused on ATM weather avoidance solutions on a larger scale, research has also focused on weather avoidance solutions at the air traffic control (ATC) sector scale. One example is a study by Vigeant-Langlois and Hansman (2002). They proposed a trajectory-based forecast method that could aid air traffic controllers in keeping aircraft away from hazardous weather. In their model, there is a differentiation between 4D tubes (which represent aircraft trajectories) and 4D unsafe volumes (which represent hazardous weather areas). When used by an air traffic controller, the model can predict if there is an intersection of tubes and volumes (i.e., an aircraft–weather interaction), allowing the controller to choose trajectories that are free of aircraft–weather interactions. The Vigeant-Langlois and Hansman framework is an

2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

2015 Hogrefe Publishing

controllers to provide accurate and timely weather advisories to pilots.

Background Researchers have reported about the benefits from shared Next Generation Weather Radar (NEXRAD) on controller–pilot interactions. For example, Farley, Hansman, Endsley, Amonlirdviman, and Vigeant-Langlois (1998) examined the importance of traffic and weather information in rerouting situations. During the simulation, participants resolved traffic and weather conflicts in scenarios where researchers manipulated the sharing of traffic and weather data provided by digital data link. The researchers found that when no weather displays were available, the controllers’ responses indicated an awareness of only 40% of the weather-related conditions. In contrast, in conditions where controllers had access to shared weather displays, controllers demonstrated a 93% awareness of weather-related conditions. Hansman and Davison (2000) reported similar results for controller–pilot interactions during convective weather scenarios. In current US en route operations, en route controllers have access to precipitation information provided to the Display System Replacement via the Weather and Radar Processor (WARP) using NEXRAD data (Moosakhanian, Higginbotham, & Stobie, 2005). However, controllers integrate weather information (i.e., graphical and text-based format) and traffic data manually while providing advisories to pilots. Once the controller decides to issue a weather advisory to an aircraft, there is an initial production phase in which parameters such as aircraft location, distance to weather, directions, and weather area coverage are extracted from the situation display and used in the weather advisory. This initial phase is entirely manual and cognitive in nature; the controller has no system support during this phase. The content and phraseology for weather advisories are defined in Federal Aviation Administration (FAA) Order JO 7110.65V: Weather and Chaff Services (FAA, 2014), which states that the issuing of weather and chaff information is done either by (a) defining the area of coverage in terms of azimuth (by reference to the 12-hour clock) and the distance from the aircraft or by (b) indicating the general area width and area coverage in terms of fixes or distance and direction from fixes. (FAA, 2014, para. 2-6-4). One weather advisory example in FAA (2014) is made up of four parameters: intensity of the precipitation area (e.g., Heavy to Extreme), area of coverage (e.g., between 10 o’clock and 2 o’clock), distance from the aircraft to the precipitation area (e.g., 15 miles), and the width of the precipitation area (e.g., 25 miles in diameter). Using ATC phraseology, the weather advisory takes the following form: ‘‘Heavy to Extreme precipitation between 10 o’clock Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Original Article

important concept for trajectory-based weather forecasts, and it could potentially improve sector throughput, increase flight safety, and reduce controller workload. Another weather probe example is a set of User Request Evaluation Tool (URET) enhancements (i.e., problem resolution support capabilities) that can support air traffic controllers (Kirk & Bolczak, 2003) at the sector level. These support capabilities are labeled problem analysis, resolution, and ranking (PARR) and include analysis and resolution support for severe weather avoidance. The tool output can be displayed as current and forecasted weather areas, which can support controllers by displaying aircraft and their relation to areas of hazardous weather. The tool set also allows controllers to perform trial planning and to find aircraft routes that are clear of hazardous weather. A similar concept has also been evaluated by Love, Chan, and Lee (2009) using the Center Terminal Radar Approach Control (TRACON) Automation System (CTAS). The researchers modified the CTAS algorithm to avoid traffic conflicts and simultaneously reroute aircraft around convective weather hazards. The result showed a 91% conflict resolution for high-traffic scenarios and moderate weather and showed a 72% conflict resolution for high-traffic scenarios and severe weather. In today’s ATC system, there are no weather decisionsupport tools at the air traffic controller workstation. Besides a precipitation display and text-based weather advisories, controllers have no ability to display aircraft and weather hazard conflicts within their airspace. As a result, air traffic controllers must manually integrate weather information and traffic information while making decisions during adverse weather operations. Although previous research has shown benefits in capacity and safety of flight from severe weather avoidance tools (i.e., weather probes), less research has assessed the verbal communication of weather information between controllers and pilots. For example, the use of forecast tools and weather probes could provide controllers with information about safe reroutes that avoid weather hazards. However, during ATC operations, this information needs to be communicated effectively to pilots. In today’s ATC, controllers verbally transfer weather and traffic information to pilots via radio. There is no other air–ground communication mode in the National Airspace System (NAS), such as a data link capability (i.e., electronic transfer of communication), that could assist controllers during weather avoidance operations. Besides radio frequency congestion, there are also ATC work-related factors like traffic load and workload that can negatively affect radio communications during adverse weather conditions. A recent safety report by the Civil Air Navigation Services Organisation (2013) states that a great number of ATC loss-of-control accidents were associated with a failure by controllers to provide proper weather information to pilots. Also, previous reports by the National Transportation Safety Board (2006) describe some examples where pilots either received inadequate weather advisories from controllers or received none at all. Therefore, there is a need to assess controller weather information systems and air–ground integration modes that enhance controller weather avoidance actions and support

Original Article

U. Ahlstrom: AIRWOLF Weather Advisory Tool

and 2 o’clock, 15 miles. Precipitation area is 25 miles in diameter.’’ In previous work, we assessed ways to optimize the development of controller weather displays (Ahlstrom, 2005). Optimal weather displays could enhance controllers’ weather situation awareness and assist controllers when relaying important weather information to pilots. However, today’s manual production of weather advisories via radio adds another constraint on controllers during high-workload periods. This further emphasizes the need to explore ways to (a) support the controller with weather advisories and (b) optimize weather-related communication between the controller and the pilot. One way of integrating weather information at the controller workstation is to let the ATC system keep track of weather hazards. Rather than providing controllers with more weather information, the system could use the weather data and automatically track weather hazards and sector traffic and alert the controller of impending conflicts. This could enhance weather avoidance operations even further as there would be no need for controllers to manually correlate weather and traffic data (current ATC operations). In the following section, we describe an automatic weather probe that tracks aircraft and weather hazards and alerts the controller to impending conflicts. In addition, the weather probe automatically generates weather advisories for sector aircraft.

Automatic Identification of Risky Weather Objects in Line of Flight In previous research, we outlined a general weather probe algorithm for air traffic controllers (Ahlstrom & Jaggard, 2010). We named the weather probe concept after its function: Automatic Identification of Risky Weather Objects in Line of Flight (AIRWOLF). The AIRWOLF tool operates by (a) calculating a future aircraft position for an aircraft (e.g., 15 min along the aircraft’s projected route) and then (b) determining whether the route to the future position intersects with a known weather object (e.g., precipitation or convection area). If there is an intersection between an aircraft route and a weather object, the probe will detect it and treat it as a conflict. The probe performs this calculation for all aircraft at their associated altitudes. This is an iterative calculation that the algorithm performs for each aircraft every time there is a flight data update in the system. The AIRWOLF tool checks the aircraft’s flight plan for route information for flat tracking aircraft (i.e., aircraft that conform to their specified routes). If an aircraft is free tracking (i.e., aircraft that are not flying their routes), the probe uses the aircraft’s current heading to perform the future position calculation. The time parameter for the calculation of the future aircraft position can be adjusted to account for different sector needs (i.e., sector size, traffic patterns, and so forth). The AIRWOLF tool can also use a function that is defined by the FAA based on weight that determines whether a particular aircraft is small, large, or heavy. This implies that the tool can be programmed to Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Figure 1. The weather conflict alert. probe for one set of weather hazards for small aircraft (e.g., visibility, ceiling, precipitation, convection, icing, and turbulence) and another set of weather hazards for large and heavy aircraft (e.g., heavy and extreme precipitation, severe and extreme turbulence). When the AIRWOLF tool detects a future conflict between an aircraft and a hazardous weather object, it alerts the controller. In our current implementation, the controller was alerted by a weather conflict number in line 0 of the data block. When the tool generates a weather conflict alert, the controller can display the aircraft route and the weather hazard area (see Figure 1). In this paper, we will focus on the probing of WARP data, although the AIRWOLF tool can read and probe any gridded weather data sets.

Automatic Weather Advisories For the AIRWOLF probe output, the system displays an aircraft that is in conflict with a precipitation hazard (as we have shown in Figure 1). Because the location of the aircraft and the precipitation hazard are known to the system, the algorithm can compute and generate an automated weather advisory. In our current advisory tool implementation, the automated weather advisories (i.e., free-text message) are displayed in a Weather Advisory View, following the en route automation modernization (ERAM) view concept (see Figure 2). Using the automated Weather Advisory View, controllers can read automated weather advisories to pilots via radio. Automated weather advisories could support the controller by (a) eliminating the need for a manual integration of traffic and weather data, (b) reducing cognitive load by eliminating the need to manually produce a weather 2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

(a) allow us to measure a sufficiently large sample of weather advisories, (b) control when and how many advisories were issued, and (c) measure advisory time durations that were not confounded by controller workload effects. In the following paragraphs, we describe the simulation conditions and the data recordings. Figure 2. The automatic weather advisory.

Participants Seven male ATC subject matter experts (SMEs) participated in the data collection activity. Their mean ATC experience was 27.2 years (SD = 11.21) and mean age was 54.7 years (SD = 11.12, range 35–65). All SMEs had extensive prior ATC experience from working as certified professional air traffic controllers. Three of these SMEs participated in workload assessments by wearing a flexible headband for functional near infrared (fNIR) spectroscopy recordings. All SMEs had knowledge of ATC weather avoidance operations and were familiar with the AIRWOLF weather probe functionality.

Purpose The main purpose of this study was to explore ways to improve weather information dissemination for en route controllers as they assist pilots to avoid hazardous weather. This entailed an evaluation of a weather display and a weather decision-support tool, as well as an evaluation of the production, dissemination, and content of weather advisories. An additional purpose was to assess differences in the cognitive workload associated with manual and automatic weather advisory production. As a baseline and control condition for future evaluations, we also explored data link communications as an air–ground communication mode. To evaluate the potential benefits of automated weather advisories, we conducted an AIRWOLF weather support tool part-task simulation. Several questions were of particular importance for this effort. First, are there any time reductions in the production and dissemination of automatic weather advisories compared with manual advisories (i.e., today’s ATC)? Second, are there any reductions in the cognitive workload associated with the use of automated weather advisories compared with manual advisories? Based on the potential benefits from the use of automated weather advisories compared with manual advisories, we predicted that automated advisories would take less time to generate, be more accurate, and require less cognitive effort during ATC operations.

Apparatus Throughout the project, we used our DESIREE simulator for testing and development. DESIREE is a realistic and advanced simulator of en route ATC and replicates the functions and user interfaces of the ERAM system. During the part-task simulation, the SMEs used a display interface that simulates the ERAM system. We used one controller workstation (radar [R]-side) equipped with a situation display (2,048 · 2,048 pixels), keyboard, auxiliary input device, and trackball device that are the same as the equipment used in current en route field operations. We measured cognitive workload by a portable fNIR system (Izzetoglu, Bunce, Izzetoglu, Onaral, & Pourrezaei, 2007). The fNIR technology uses specific wavelengths of light to measure changes in the relative ratios of deoxygenated and oxygenated hemoglobin during brain activity. The continuous wave fNIR system we used is connected to a flexible forehead sensor pad that contains four light sources (peak wavelengths at 730 nm and 850 nm) and 10 detectors. This configuration generates a total of 16 measurement locations or voxels per wavelength. With two wavelengths and dark current recordings for each of the 16 voxels, the system generates a total of 48 measurements for each 2-Hz sampling period.

Simulation Conditions

Method There are many factors that influence the number and duration of weather advisories that air traffic controllers issue during real-world operations. In the present part-task simulation, we were interested in measuring time durations and therefore needed to create a focused test that would 2015 Hogrefe Publishing

We created two different weather and traffic scenarios using a generic en route sector. The generic sector (Sector 07) is part of a larger generic ATC center and provides a realistic ATC environment with navigational aids, simple radio frequencies, and basic ATC operating procedures (see Figure 3). The sector is a high-altitude sector ( 200 · 125 nautical miles) with boundaries from flight level (FL) Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Original Article

advisory, and (c) reducing the total time it takes a controller to provide a weather advisory to a pilot. Using our Distributed Environment for Simulation, Rapid Engineering, and Experimentation (DESIREE) simulator capabilities, we also assessed a data link capability where the controller can send the weather advisory to pilots electronically. This future capability could enhance weather-related communications even further by eliminating voice transmissions altogether.

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Original Article

Figure 3. The generic en route Sector 07 with boundaries, jetways, and fixes.

240 and above. The traffic flow consisted of arrivals on J21, J100, and J22, and overflights through the sector. The weather hazard areas (moderate to extreme precipitation) in both scenarios were moving into the south portion of the sector (as illustrated in Figure 1), conflicting with the routes of North and West arrivals going through the sector. To control for when and how many weather advisories had to be issued in each scenario, we used the AIRWOLF probe running in two different modes. This created three different weather advisory conditions: manual, automatic, and data link. The main focus was the comparison between the automatic and the manual conditions. The data link condition was solely a control condition that was of interest for future evaluations. In the first mode, called the manual condition, which simulated today’s en route ATC system, we used the tool alerting function (i.e., number indicator in line 0 of the data block) to indicate to the participants that a pilot was in need of a weather advisory. All other alerting functions of the AIRWOLF probe (i.e., route and weather hazard display) were suppressed during this condition. Each participant manually (a) composed a weather advisory (using the parameters and format defined in the section Background) by extracting and integrating information about the aircraft route and weather area (WARP precipitation display) and (b) transmitted the weather advisory to the pilot via radio. Weather information was provided by the precipitation display, center weather advisory (CWA), and by the significant meteorological information (SIGMET) advisory; no

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

additional weather information was available to the participant. We encouraged participants to use the example advisory phraseology (defined in the section Background) as much as possible, and we posted an advisory cheat sheet example at the controller workstation. In the second AIRWOLF mode, called the automatic condition, the tool also displayed an alert in line 0 of the data block. Simultaneously with the weather conflict alert, the system displayed an automatic weather advisory in a Weather Advisory View on the situation display. The participants could select the line 0 alert number and display a highlighted aircraft route and weather hazard area, as illustrated in Figure 1. In this condition, the participants read the automatic weather advisory to the pilot over the radio. In a third, control condition, called the data link condition, the weather alerting function was the same as in the automatic condition. However, no voice transmission was necessary because the participants could send the automatic weather advisory directly to the pilot via data link. Because this condition involved neither the production of an advisory nor a voicing of any advisory content, this condition served only as a reference baseline for future studies. With two weather–traffic scenarios and three weather advisory conditions, each participant ran six simulation runs in a counterbalanced order. Each simulation run lasted 30 min, with an average of 6.5 aircraft in the sector at any given time (minimum = 0 at startup, maximum = 15, yielding a medium traffic load). The AIRWOLF weather alert function was adjusted so that all weather conflicts occurred 2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

while the aircraft was inside the sector under control. If the participant did not vector any aircrafts, the simulator system generated 15 and 22 weather conflicts during the first and second weather–traffic scenarios, respectively. We used two different weather–traffic scenarios for the purpose of generating variability in the advisory parameters (e.g., distance from aircraft to weather, extent of weather area). To reduce the participants’ workload due to traffic load, we eliminated all initial aircraft-to-aircraft conflicts by separating all arrival traffic streams by altitude. However, after an aircraft entered the sector, participants performed any ATC operations deemed necessary based on traffic load and weather hazards.

Procedure

Data Collection During the simulation runs, we recorded and time-stamped all weather conflict alerts, voice alerts, and push-to-talk (PTT) events (i.e., pushing down the key and releasing the key on the microphone). During the data link control condition, we also recorded and time-stamped when the SME selected an aircraft in the Weather Advisory View and clicked on the send button. Viewed in sequence for the manual condition, first, we presented a weather alert displayed in line 0 of the data block (alert). Next, the SME manually formulated a weather advisory and subsequently pushed down on the microphone button (push) and voiced the advisory over the radio (voice). When the voice communication was completed, the SME released the microphone button (release). This completed the data recording chain for the manual condition with alert, push, voice, and release. For the automatic condition, we presented a weather alert displayed in line 0 of the data block (alert). Simultaneously with the alert, the system displayed an automatic weather advisory. Next, the SME pushed down on the microphone button (push) and read the advisory over the 2015 Hogrefe Publishing

radio (voice), releasing the microphone button when done (release). For the data link control condition, we presented a weather alert displayed in line 0 of the data block (alert). The SME selected the aircraft from the Weather Advisory View list and clicked the send button to send the weather advisory to the pilot.

Data Analysis Traditionally, researchers use the null hypothesis significance testing (NHST) framework when analyzing their data (Gigerenzer, 2004). The goal of this framework is to unveil statistically significant effects by rejecting a null hypothesis of no effect. However, the NHST testing framework is asymmetric – researchers can never accept evidence for the null, they can only reject the null and infer the probability (e.g., p = .021) of getting an outcome (or more extreme values) given that the null hypothesis is true, or p(data|null). But this is contingent upon following the strict a priori requirements that (1) researchers stop the data collection when N (number of subjects) equals a predetermined threshold (i.e., the ‘‘stop rule’’), (2) they only have a particular outcome to test, and (3) they never perform any more tests on the same data ever again. Of importance for this study is the fact that the absence of a statistically significant effect is not the same as the presence of a null effect. We need to capture and assess the presence of null outcomes as well as the presence of weather advisory effects. An implementation of a weather advisory tool is contingent upon practical and meaningful effects on air traffic controller performance and workload. However, if some aspects of the weather advisory usage (i.e., advisory time duration and efficiency) are the same (for practical purposes) for automated and manual advisory procedures, we need to detect and provide evidence for null effects as they imply a lack of benefits. And if there is a lack of benefits we do not pursue the advisory methodology any further, as data collection is resource demanding and costly. Besides the inability to accept null effects, there are other problems with the NHST framework. One criticism is that it overstates the evidence against the null hypothesis (Wagenmakers, Wetzels, Borsboom, & van der Maas, 2011). Another criticism is that there is no single and unique p value for any given data set; all data sets have many different p values because the p value is determined by the experimenter’s intentions (Kruschke, 2013; Wagenmakers, 2007). This is particularly problematic, as researchers often gauge effects in relation to their p values. The computation of p values assumes that the sample size was fixed in advance and that the researchers collected data until that fixed threshold was reached. The algorithm does not assume that researchers intended to collect data during a fixed time interval. The algorithm tries to figure out what the probability is of getting a certain sample statistic given that it was sampled from a null distribution if the intended experiment was repeated ad infinitum. This means that in Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Original Article

Before the part-task simulation began, each SME was briefed on the different weather advisory tasks and, thereafter, performed three practice scenarios (manual, automatic, and data Link). During both the practice and data collection runs, the SMEs used simulation-pilot commands to vector aircraft away from weather hazards. We used this procedure in our lab during scenario development and testing, and it was appropriate for the present part-task simulation because there was no need for actual simulation pilots. None of the practice scenarios were used during the data collection. At the start of the simulation, each SME was given a SIGMET advisory and a CWA that described the overall weather pattern. The SME was instructed to pay special attention to aircraft and weather areas (given the known thunderstorm conditions) and to provide weather advisories as needed. Each SME ran the six simulation conditions over the course of several days.

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Original Article

many situations the p values are calculated using the wrong sampling distribution (Dienes, 2011). Other problematic issues include NHST’s demand that analysts make corrections for multiple tests (O’Keefe, 2003) and the inability to incorporate historical data into new analyses (i.e., information from previous studies cannot be used as priors for subsequent analyses). There are also problems when analyzing small-sample data and data from unbalanced designs. Missing data values destroy the regularity and straightforwardness of an NHST analysis. This becomes more complex with an increasing number of missing values. Analysis of variance (ANOVA) split-plot designs get particularly difficult, or impossible to perform, when there are a different number of subjects for the between-subjects variable (for example, see Harrison et al., 2014).

Limitations of the Current Study The current study used a small-N data set from a simulation where ATC controllers provided weather advisories to pilots. This simulation was originally scheduled to run for 4 weeks with the goal of recruiting as many participants as possible (i.e., the ‘‘stop rule’’). However, due to a redirection of research priorities and project funding, the simulation was terminated prematurely. These constraints have important implications for data interpretation. In the applied psychology and human factors domain, researchers often end up with small data sets for their analysis and subsequent conclusions. Because of a lack of sufficient funding, time, study participants, or a combination thereof, the researcher is often left with a data set that is not randomly sampled, is unbalanced, and contains fewer observations than originally intended. However, applied research and applied evaluation efforts are important, as they often guide an organization in how to implement a future enhancement or to evaluate current practices with regards to efficiency and safety. Consequently, we need to use all of the data we have but at the same time be aware of constraints and choose optimal analysis methods for small and unbalanced samples. For the reasons stated above, we used a Bayesian framework, as it is more suitable for applied research with small samples than are NHST analysis methods. The Bayesian framework has fewer methodological constraints, does not rely on p values, can be used to accept null outcomes, is easier to apply to small-N samples and real-world data, and can lead to richer (and in many cases different), more substantive conclusions.

Bayesian Estimation In contrast to NHST analyses, modern Bayesian analysis can be used to assess whether a null value is true rather than being false. Bayesian analysis is also very flexible and works equally well with small and large sample sizes, and the data set need not be balanced as in NHST. Therefore, Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

for our data analysis, we perform Bayesian parameter estimation using a model and framework detailed in Kruschke (2013). Bayesian analysis computes the credibility of parameters or hypotheses, given our actual data, without being dependent on a researcher’s intentions (i.e., ‘‘stop rule’’). The analysis yields a complete distribution of credible values and their uncertainty in the posterior distribution. When performing multiple comparisons using a Bayesian analysis, there is no need to make corrections for multiple tests as with NHST, because for each additional test, we only view the outcome from different perspectives in the multidimensional parameter space. There is only one posterior distribution; it does not change as we perform multiple tests on the same data set. Our Bayesian estimation framework used Markov chain Monte Carlo (MCMC) simulations to determine the posterior distribution of likely means (M ), standard deviations (SD), and effect sizes (ES ) given the actual data. We used JAGS (Plummer, 2003, 2011) called from R (R Development Core Team, 2011) via the package rjags, using adapted program code from Kruschke (2011, 2013). This procedure involves generating a large number of representative combinations of parameter values from the posterior distribution, and then using those values to generate an approximation of the posterior. When we have a posterior distribution with a large sample of representative parameter values, we can evaluate the mean of a parameter distribution or the difference between values of different parameters (e.g., SD and ES ). We use a separate decision rule to convert the posterior distribution to a specific conclusion about a parameter value. Together with each posterior distribution, we plotted a black horizontal bar that represented the 95% high density interval (95% HDI). Every value inside the 95% HDI has a higher probability density (i.e., credibility) compared with values that are outside the HDI. Therefore, values contained within the 95% HDI represent the most credible values of the parameter. When we explore differences between parameter values in a contrast, we compute these differences at each step in the MCMC chain and plot the differences along with the 95% HDI in a histogram. Posterior histograms show what differences are credible and the uncertainty of those differences. If the value 0 (implying zero difference between parameters) is not contained within a 95% HDI for a histogram of differences, we say that the difference is credible. If, on the other hand, the 95% HDI includes the value 0, we cannot say that the difference between parameter values is credible, because a difference of 0 is indeed among the credible outcomes. To assess the presence of null effects, we use a region of practical equivalence (ROPE). The ROPE contains values that, for all practical purposes, are the same as a null effect. We chose a ROPE that spans an ES (Rosnow & Rosenthal, 2003) of 0 ± 0.2, which Cohen (1992) labeled a small effect and which would likely have no meaningful practical consequences for the conditions examined here. To determine the presence of null effects, we used the 95% HDI in conjunction with the ROPE. If the 95% HDI for a posterior distribution falls completely within the ROPE margins, we declare the presence of a null effect, and we accept the null 2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Results For all of the analyses in this section, we used the pooled data from the two weather-traffic scenarios.

Weather Advisories: Total Time Durations In a first part of the analysis, we compared the total weather advisory duration for the automatic and the manual conditions (from alert to release). Figure 4A shows the advisory duration data (N = 183) for the automatic and the manual conditions and the data link control condition. Figure 4B shows the data for the posterior difference between the manual and the automatic conditions (top right), the mean of the difference (top left), the SD of the difference (bottom left), and the ES of the difference (bottom right). The distribution of lines over the difference data (top right) are representative examples of posterior predictive distributions (i.e., predictive checks), created from selecting random steps in the MCMC chain. The purpose of plotting these distributions is to make sure that there is a reasonable fit between the model and the data. The black horizontal bar represents the 95% HDI (which will be shown in all subsequent contrast histograms). The vertical dotted axis shows the proportion of the posterior distribution that is below and above the value 0 (also included in all subsequent contrast histograms). The ROPE spans an ES of 0 ± 0.2. As illustrated in Figure 4B, the durations for the automatic weather advisories were credibly shorter than the durations for the manual advisories, with a posterior mode for the difference of 5.11. There was also a credible ES

2015 Hogrefe Publishing

for the difference, with a mode of .698, and the 95% HDI completely outside the ROPE. We also performed exploratory analyses that showed a credibly shorter advisory duration for the data link control condition compared with the automatic and manual conditions. For the comparison between the automatic and the data link control conditions, we found a mean of the difference of 16.2 (95% HDI from 15.5 to 16.9) and an ES of 5.72. For the comparison between the manual and the data link control conditions, we found a mean of the difference of 18.9 (95% HDI from 17.7 to 20.2) and an ES of 3.88.

Weather Advisories: Production Time One of the potential benefits of using automated weather advisories is that there is no need to manually extract distances and directions from an aircraft to weather and to compose a weather advisory. By comparing the time difference between alert and push, we can evaluate how the production duration varies between these two advisory conditions. Figure 5A shows the advisory production data for the automatic and the manual conditions and the data link control condition, N = 183. Figure 5B shows the difference data between the automatic and manual conditions (top right), the mean of the difference (top left), the SD of the difference (bottom left), and the ES of the difference (bottom right). There was a credible difference in the production duration between the automatic and the manual conditions, with a shorter duration for the automatic condition giving a posterior mode of the difference of 3.71. The ES for this difference was also credible, with a posterior mode of 0.561, and the 95% HDI completely outside the ROPE. To compose a weather advisory in the manual condition, the SMEs went through an initial production phase in which traffic and weather parameters were manually extracted from the display (mental integration). In the automatic condition, no such mental integration was needed, because the SMEs were provided with an automatic weather advisory. This results in a production phase for automatic weather advisories that is credibly shorter than the corresponding phase for manual weather advisories. We also explored how the production duration for the data link control condition compared with the automatic and the manual conditions. For the data link control condition, the production phase consists of the time duration between the alert and the time when the SME selects the aircraft from the Weather Advisory View. There was no credible mean difference between the automatic and the data link control conditions, M = 0.08, 95% HDI from 1.01 to 1.12, ES = 0.025, as 91% of the posterior distribution fell within the ROPE area. However, there was a credible mean difference between the manual condition and the data link control condition, M = 2.17, 95% HDI from 0.56 to 3.75, ES = 0.31, with a shorter production phase for the data link condition.

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Original Article

outcome. We do this because all of the credible values within the 95% HDI are, for all practical purposes, the same as a null value. If, on the other hand, the entire ROPE falls outside the 95% HDI, we reject the presence of a null effect. It is quite common for human-in-the-loop simulation data to include data points that are far away from central data points. To accommodate this in our analyses, we used robust estimation. Robust estimation is called robust because it can accommodate outliers in the data. In this study, we accomplish this by using Bayesian models with t distributions rather than normal distributions. A t distribution with tall-tails (small df ) can ‘‘reach out’’ and accommodate values that lie outside the central data. A normal distribution, on the other hand, will instead increase the standard deviation and (incorrectly) ‘‘move’’ the mean of the distribution toward the outlier. For all analyses, we used 200,000 samples to derive the posterior distribution: 1,000 steps to ‘‘tune’’ the samplers; 5,000 steps to ‘‘burn-in’’ the samplers, while running three chains and saving every step in the chain. For all analyses, we use priors that were vague and noncommittal on the scale of the data.

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Original Article

(A)

Figure 4. (A) Total advisory duration data for the manual and automatic conditions and the data link control condition. (B) The difference data for the total time durations, the mean of the difference, the SD of the difference, and the effect size (ES) of the difference. HDI = high density interval.

(B)

Weather Advisories: Voice Durations Figure 6A shows the time durations for the voice (push to release) for the automatic, the manual, and the data link control conditions, N = 183. (Note: The duration for the data link control condition is not voice but instead the time duration from the alert to selecting the aircraft [select] in the Weather Advisory View.) Figure 6B shows the data for the difference between the automatic and the manual conditions (top right), the mean of the difference (top left), the SD of the difference (bottom left), and the ES of the difference (bottom right). There was a credible difference in the voice duration between the automatic and the manual conditions, with a shorter duration for the automatic condition (mode = 0.935). There was also a credible ES with a mode of

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18â&#x20AC;&#x201C;35

0.419, and the 95% HDI was completely outside the ROPE. We also explored how the data link control condition compared with the automatic and the manual conditions. The data link control condition had a credibly shorter duration than the automatic condition, M = 8.5, 95% HDI from 7.75 to 9.21, ES = 3.12, and the manual condition, M = 2.18, 95% HDI from 0.60 to 3.79, ES = 0.32, although there was a lot more variability in the data link duration data. Controllers provide weather advisories to pilots based on operational needs and workload levels. This means that the advisory content will vary from one weather situation to the next. In extreme situations, the controller might be unable to provide advisories due to workload. In intermediate situations, advisories are shorter and contain only precipitation intensities, distances, or azimuths. The safety

2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

(A)

Figure 5. (A) Production duration data for the manual and automatic conditions and the data link control condition. (B) The difference data for the advisory production time, the mean of the difference, the SD of the difference, and the effect size (ES ) of the difference. HDI = high density interval.

Original Article

(B)

aspect is paramount; controllersâ&#x20AC;&#x2122; first priority is to keep vertical and horizontal separation â&#x20AC;&#x201C; not to provide weather advisory information. Detailed time duration events for weather advisories are not readily available from ATC field operations. There is also no objective way to determine when a controller decides to provide a weather advisory to an aircraft, thereby initiating the production phase. Furthermore, there are no data available on how the advisory coverage compares with the correct precipitation coverage. All we can objectively measure from real-world data is the duration of the weather advisory (i.e., the time duration of the voice from PTT recordings). For the present study, we were interested in the content and time distribution of controller field advisories for comparison with our SME advisories. However, controller field 2015 Hogrefe Publishing

phraseology will not exactly follow the weather advisory examples given in FAA (2014) and as used in this study. Given the operational constraints from traffic load and controller workload levels, we predicted a larger SD of advisory durations compared with the advisories recorded in this simulation. Nevertheless, we were interested in a comparison because it can provide some additional clues to whether automated weather advisories could benefit ATC field operations. As part of an ERAM Post Implementation Review, FAA researchers received audio recordings from the busiest ATC sectors at a variety of air route traffic control centers. These tape recordings contained two channels: one for the audio (radio communications) and one for the interrange instrumentation group (IRIG) time code. From these audio recordings, we transcribed the audio data from two ATC Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18â&#x20AC;&#x201C;35

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Original Article

(A)

Figure 6. (A) Advisory voice duration data for the automatic and manual conditions and the data link control condition. (B) Voice duration difference data, the mean of the difference, the SD of the difference, and the effect size (ES ) of the difference. HDI = high density interval.

(B)

centers. The resulting transcripts were searched for the following keywords: precip, weather, extreme, moderate, and rain. The start and stop time for each air traffic controller transcription utterance was marked using in-house speech detection software that precisely determined when speech began and ended. We then used these time stamps to compute the duration of each communication that contained a keyword utterance. After a manual inspection of the communication contents, we retained 126 weather communications. Figure 7A shows the resulting advisory voice duration data (in seconds) and mean posterior durations from the air traffic controller group (N = 126) and the manual voice duration data from the SME group (N = 183). Figure 7B shows the air traffic controller (left) and SME (center) posterior SD and the difference of the SD (right). Figure 7C shows the difference of means between the air traffic

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18â&#x20AC;&#x201C;35

controller group and the SME group (left) and the ES of the difference (right). The mean voice duration was 8.43 s for air traffic controller advisories and 14.4 s for the SME advisories. Not surprisingly, the posterior mean difference of 5.98 was credibly different, as the value 0 is not included in the 95% HDI. Figure 7B also shows that there was a lot more variability in the air traffic controller advisories (SD mode = 4.05) than in the SME advisories (SD mode = 1.79). The right side of Figure 7C shows the ES for the difference of means, computed from credible combinations of means and standard deviations. The credible ES had a mode of 1.84 and the 95% HDI was completely outside the ROPE. As expected, the ATC field sample had a lot more voice duration variability compared with the SME data. Nevertheless, 25% of the air traffic controller advisories had longer durations than the shortest SME duration of 10.8 s.

2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

(A)

Figure 7. (A) Advisory voice duration (seconds) data and voice duration (seconds) means for the air traffic controller group and the subject matter expert (SME) group. HDI = high density interval. (B) The air traffic controller (ATC) group SD, the SME group SD, and the difference of SDs. (C) The difference of means between the air traffic controller group and the SME group and the effect size (ES ) of the difference.

Original Article

(B)

(C)

2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18â&#x20AC;&#x201C;35

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Original Article

Weather Advisory Performance Data The results indicated time benefits from using automated weather advisories compared with using manual weather advisories. First, the total advisory durations – from alert to release – for the automatic advisories were credibly shorter than the durations for manual advisories. Second, the production durations – from alert and push – were credibly shorter for the automatic advisories compared with the manual advisories. Third, the voice durations – from push to release – were also credibly shorter for the automatic advisories compared with the manual advisories. Another assessment concerned the comparison between automated and manual advisories regarding the accuracy of weather parameters for distance to weather and weather extent. The weather parameters in the automatic weather advisory were always correct, because the AIRWOLF algorithm derived these from known ATC system parameters. What we wanted to assess was the difference between the manual advisory parameters (as provided by the SMEs) and the correct advisory parameters (as provided by the AIRWOLF probe). Using the 183 PTT voice data files, we transcribed the weather advisories using in-house speech detection software. However, we were not able to transcribe all words in every PTT file, as some advisory utterances were inaudible. However, the resulting transcriptions used for comparison contained both the distance to weather and the weather extent similar to the advisory example in the Background section of this paper. During the traffic scenarios, the correct distance from a given aircraft to a precipitation area ranged from 33 nautical miles (nmi) to 65 nmi, and the precipitation coverage area ahead of the aircraft position was either 30 , 60 , or 90 . We compared each weather advisory distance and coverage with the correct distance and coverage. For each of the resulting comparisons (N = 74), we computed a difference score by subtracting the correct distance from the advisory distance. We also compared the actual area of coverage with the advisory coverage (e.g., between 11 o’clock and 12 o’clock). Figure 8A shows the data for the manual advisory distances (left) versus the correct distances (right). Figure 8B shows the difference data with a posterior predictive check (top right), the mean of the difference (top left), the SD of the difference (bottom left), and the ES of the difference (bottom right). There was a credible difference between the correct distances and the advisory distances, with a posterior mean difference of 20.2 nmi, implying that SMEs underestimated the distances to the precipitation areas. There was also a credible effect size, with a mode of 1.49 and the 95% HDI completely outside the ROPE. The data for the advisory coverage area are presented in Table 1. The shaded cells in the table indicate the correct coverages for the 30 , 60 , and 90 precipitation areas. As we can see in the table, there were very few correct precipitation area coverages in the weather advisories: 0% for the 30 areas, 8% for the 60 areas, and 10% for the 90 areas. The majority of the advisory area coverages were (erroneously) stated as 120 : 75% of the 30 areas, 79% Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

of the 60 areas, and 90% of the 90 areas. The SMEs, generally, overestimated the area coverages, with a decreasing advisory coverage error for increasing azimuths. To sum up, the distance and coverage data illustrate that SMEs underestimated the distances to the precipitation areas and overestimated the precipitation area coverage.

Cognitive Workload In previous research, we have used eye-movement activity and instantaneous workload ratings to assess air traffic controller workload (Ahlstrom & Friedman-Berg, 2006). However, neither of these methods seemed appropriate for an assessment of SME workload during weather advisory tasks. Although eye-movement activity correlates with cognitive workload, the method has some problematic issues related to intrusiveness and spatial resolution. Furthermore, eye-movement activity measures only work optimally for displays with clearly defined objects and where the operator interacts with or visually tracks these objects. Subjective workload ratings, on the other hand, are impossible to use without interrupting the operator’s task; and the frequency by which researchers collect workload ratings over time is low. Usually, subjective workload ratings are collected every 2–5 min during the task or at the end of a simulation run. In addition to eye-movement activity measures and subjective workload ratings, researchers have also used objective fNIR recordings as a measure of cognitive workload. In previous fNIR studies (Izzetoglu et al, 2007), researchers assessed prefrontal blood oxygenation levels in relation to cognitive effort by manipulating task load and task difficulty (i.e., changes in the number of airplanes) during a complex air warfare task. Using non-ATC participants, the results verified a positive correlation between the level of blood oxygenation and performance measures. In addition, researchers have also found that n-back tasks yield a positive correlation between increasing workload and oxygenation levels. Similarly, Ayaz et al. (2012) studied mental workloads of operators performing standardized n-back tasks and complex ATC tasks as well as the development of expertise during the practice of complex cognitive tasks. In the Ayaz et al. (2012) study, researchers used both certified professional air traffic controllers and college students. They found that fNIR measures were sensitive to mental task load and practice level. More importantly, however, the outcome showed that the average oxygenation levels (as measured by fNIR) for certified professional air traffic controllers increased monotonically with increased task difficulty. In a more recent study, Harrison et al. (2014) assessed the cognitive workload of air traffic control specialists using a conflict resolution advisory tool. This study, using both subjective (ratings) and objective (fNIR) workload measures, also found credible oxygenation differences between continuously increasing workload levels induced by an increased traffic load. The study indicated that the cognitive workload of air traffic controllers might have increased at a much faster rate than what was indicated by controllers’ subjective workload ratings. This was 2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

(A)

Figure 8. (A) Weather advisory distance data versus the correct distance data; nmi = nautical miles. (B) Difference data for the weather advisory distances, the mean of the difference, the SD of the difference, and the effect size (ES ) of the difference. HDI = high density interval.

Original Article

(B)

Table 1. Percentage advisory coverage and actual coverage Correct coverage Advisory coverage

30 60 90 120

0% 25% 75%

8% 13% 79%

10% 90%

Notes. Examples of the 30 , 60 , and 90 advisory coverages could be ‘‘between 11 o’clock and 12 o’clock,’’ ‘‘between 11 o’clock and 1 o’clock,’’ and ‘‘between 11 o’clock and 2 o’clock,’’ respectively. The shaded table areas show the correct weather coverages.

2015 Hogrefe Publishing

probably due to the reluctance of controllers to use the entire workload rating scale (1–10), leading to biased estimates of controller self-rated workload levels. Therefore, in this study, we used an objective measure of neural brain activity (i.e., fNIR) to assess whether providing weather advisories during the automatic and manual conditions yielded different levels of cognitive workload. Typically, the fNIR signal from neural activation is a decrease of deoxygenated hemoglobin accompanied by an increase of oxygenated hemoglobin. Because there is a production of weather advisories in the manual condition, we hypothesized that we would see a greater oxygenation change in the manual condition than in the automatic condition. The optical signals acquired by the fNIR system were converted into concentration changes of oxygenated

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

U. Ahlstrom: AIRWOLF Weather Advisory Tool

Original Article

(A)

Figure 9. (A) Oxygenation data for the automatic and the manual conditions and the data link control condition. (B) Oxygenation difference data for the automatic and manual conditions, the mean of the difference, the SD of the difference, and the effect size (ES ) of the difference. HDI = high density interval.

(B)

hemoglobin and expressed in micromolar (micromoles per liter) concentrations. We averaged the oxygenation values (over the 16 channels) for each minute of the scenario, yielding 60 oxygenation values per SME per advisory condition. Figure 9A shows the oxygenation data for the automatic and the manual conditions and the data link control condition. Figure 9B shows the data for the difference between the automatic and manual conditions (top right), the mean of the difference (top left), the SD of the difference (bottom left), and the ES of the difference (bottom right). There is a credible difference in oxygenation levels between the automatic and the manual conditions, with higher oxygenation levels for the manual condition. The mode of the difference was 1.21, and the 95% HDI excluded the value zero. The SD of the difference was 2.31, and the ES of the difference was 0.536, with a 95% HDI that was completely outside the ROPE. Therefore, our conclusion is that manual advisory tasks elicit more cognitive activity-related hemodynamic changes Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18â&#x20AC;&#x201C;35

than automatic advisory tasks. This can be explained by the production phase in the manual condition. It adds a cognitive load that is indicated by elevated oxygenation levels in the prefrontal cortex. We also explored the differences in oxygenation levels between the automatic, the manual, and the data link control conditions. There was no credible difference in oxygenation levels between the automatic condition and the data link control condition, M = 0.06, 95% HDI from 0.38 to 0.25, ES = 0.05, and 91% of the posterior distribution within the ROPE; however, there was a credible difference between the manual condition and the data link control condition, M = 0.89, 95% HDI from 1.12 to 0.63, ES = 0.68.

Discussion In this paper, we examine weather advisory procedures for en route air traffic controllers. In todayâ&#x20AC;&#x2122;s en route 2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

2015 Hogrefe Publishing

In the present study, the performance data for distance and coverage judgments illustrated that SMEs underestimated the distances to the precipitation areas and overestimated the precipitation area coverage. Automated weather advisories could reduce these advisory errors, as the AIRWOLF algorithm computes the correct distances and area coverage based on aircraft, weather, and route information already in the ATC system. The question is how advisory errors found in this study relate to ATC field operations. Granted, there are several differences between our SME participants and certified professional controllers in the field. First, ATC field controllers are experts on their sectors and, therefore, have an acute sense for distances and directions within their airspace. In the present study, the SMEs (who were not current) worked a generic sector, which was easy to learn and operate, but they never gained the same proficiency as controllers operating in the field. Therefore, it is likely that the SMEs used mental shortcuts or heuristics for deriving distances to weather areas, as indicated by the strong tendency to give 20-nmi and 30-nmi advisory distances. Second, we used a standard weather advisory example as part of the study design. In ATC field operations, controller weather advisory behavior is dictated by operational constraints relating to traffic load and controller workload. Nevertheless, automated weather advisories could alleviate much of the mental effort needed to derive distances and coverage areas in many situations, thereby reducing the potential for errors. The tendency to overestimate an angle between two reference lines (one vertical and one oblique – like the end points of a precipitation area) is well known and has been documented by previous research (Gentaz et al., 2001). This overestimation bias is stronger for oblique orientations compared with the vertical orientation and could, therefore, have affected the SMEs’ judgments because we used precipitation area locations that intersected oblique (diagonal) traffic routes. A similar finding was also reported in an early study by Maclean and Stacey (1971). They found an unstable effect on performance from angle judgments in oblique presentations and found a performance variation between vertical and horizontal judgments. In their study, Cleveland and McGill (1986) investigated the accuracy of six basic judgments of graphical perception. The study result showed that position judgments were the most accurate, followed by length (distance) and angle judgments. Area judgments were the least accurate of all basic judgment tasks. Using three different subject groups, the authors found that judgment errors were not affected by subject training or expertise. This is an important finding because it implies that judgment errors for distances and angles could be a basic perceptual phenomenon that persists even after training. In a future version of the AIRWOLF tool, we envision an opportunity to create a more efficient controller tool for enhanced weather avoidance operations. The current AIRWOLF tool enhances controller weather situation awareness and advisory function by providing automated weather advisories and a graphical display of hazard conflicts. However, to see additional benefits, such as increased

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Original Article

operations, the radar (R)-side controllers manually integrate weather information to produce weather advisories. These weather advisories are transmitted to pilots via radio. This procedure is not optimal from a human factors perspective, and it is inefficient from an operational perspective. The manual weather advisory process adds a cognitive load on the controller, and the procedure is time consuming. Radio communications also have a capacity limitation in the ATC system. Often, radio communications, such as controller instructions, controller weather advisories, and pilot deviation requests, compete for bandwidth and create radio-frequency congestion. Adverse weather conditions also add more complexity for the controller because pilots will either request deviations or inquire about alternate routings around weather. These deviation requests must be evaluated by the controller to verify that new headings and routes do not create additional weather and traffic conflicts. If the controller workload is too high, the controller may not be able to provide the weather advisory function at all. Safety comes first; the primary controller task is to provide aircraft separation. The results of the present simulation revealed positive effects of automatic weather advisories provided by the AIRWOLF tool. Automated advisories (a) eliminate the need for a manual advisory production, (b) reduce the production time and, (c) reduce the voice duration and yield a shorter duration overall. Automated advisories also reduce the cognitive workload associated with the dissemination of advisories. As illustrated by the results from our data link control condition, the air traffic controller benefits could, potentially, be even greater if the weather advisory function is provided by electronic communication. Although not elaborated in the present study, the AIRWOLF concept is not restricted to en route operations. The concept could also be applied to terminal air traffic controllers responsible for flights in and out of terminal airspace. In a future ATC environment with data link communication, the roles and responsibilities with regard to providing weather advisories could be more effective and flexible. In today’s en route ATC, the R-side controller is the sole weather advisory provider via radio. Using data link communication, the weather advisory function could be accessible to both the R-side and the data (D)-side controllers. In extreme situations, the AIRWOLF tool could automatically generate and transfer weather advisories to pilots upon initial contact. As soon as a pilot switched frequency upon entering a new sector, the tool could automatically send weather advisories to the pilot. This would eliminate the controller’s responsibility for providing weather advisories, resulting in reduced workload and increased safety by allowing controllers to focus on traffic separation. However, the controller must be notified about these automated transfers, because new deviation requests must be evaluated to verify that new headings and routes do not create additional weather and traffic conflicts. Controllers and pilots need to have access to the same information to enhance common weather situation awareness. Pilots and controllers also need to share information to avoid potential misunderstandings and thereby improve the safety of operations.

Original Article

U. Ahlstrom: AIRWOLF Weather Advisory Tool

traffic throughput and weather avoidance efficiency, we aim to enhance the AIRWOLF algorithm with an aircraftto-aircraft conflict probe. Controllers need tactical weather decision-support tools to make safe and efficient decisions that support the hands-on, moment-to-moment management of traffic within their airspace. Making sure that pilots are aware of, and can avoid, hazardous weather is one part of the equation. Another part that is equally important is to make sure that deviation requests and reroutes are free of potential traffic conflicts. By combining the probing for weather and traffic, the AIRWOLF tool could provide conflict-free route alternatives while avoiding weather. These alternatives could be displayed to the controller for review, and used for traffic control decisions by the controller or transmitted to the pilot. Because of the potential benefits found in this study, this information transfer would be most efficient using data link communication. This could be a more optimal way to support en route air traffic controllers for the safe, efficient, and strategic efforts required to handle adverse weather conditions in the future NAS.

Conclusions and Recommendations Reducing the impact of weather is an important goal for aviation and ATC. One future concept related to this goal introduces new automation systems with integrated weather information and decision-making tools. This could require the identification and implementation of new procedures for dissemination of weather information to pilots and air traffic controllers, and the integration of weather information into controller support tools. In this study, we evaluated an air traffic controller decision-support tool that detects conflicts between aircraft and hazardous weather and alerts the controller. Simultaneous with the conflict detection, the support tool generates automatic weather advisories that controllers can relay to pilots. We found that automated weather advisories reduced advisory production times, improved the accuracy of weather advisory parameters, and reduced air traffic controller workload. Comparing the advisory duration times and cognitive workload differences between the automatic and the manual advisory conditions, we found effect sizes in the range of 0.5 to 0.7. This illustrates consistent mediumto-large effects from the automatic support tool. In none of our comparisons did we find evidence for null effects – that is, a lack of a credible effect from the tool’s automatic advisory support. Based on these findings, we propose a few recommendations. If implemented in its current form, the AIRWOLF automated weather advisory function would reduce air traffic controller cognitive workload because there is no need for the mental production of weather advisories. For future experiments, the current evaluation methodology should be expanded to include more complex traffic scenarios using simulation ‘‘pilots’’ during experiments. Although the scenarios in this study were deliberately simplistic to allow a focus on the advisory production, in real-world scenarios, Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

air traffic controllers and pilots perform complex trajectory negotiation (e.g., when a pilot requests to use a different route than that issued by a controller). Pilot requests have to be evaluated by the controller, which takes some time, and there is likely an impact on the benefits of the tool with increasing traffic load and sector complexity. This also implies that the advisory tool needs to be evaluated with not only an aircraft-to-weather probe, but a combined aircraft-to-weather and aircraft-to-aircraft conflict probe.

Acknowledgments This work was sponsored by the Federal Aviation Administration (FAA), Human Factors Research and Engineering Group. The study sponsor had no role in the study design, data collection, analysis, or interpretation of data. In addition, the sponsor had no role either in the writing of the report or in the decision to submit the paper for publication. The study plan was reviewed and approved by the FAA Institutional Review Board (IRB).

References Ahlstrom, U. (2005). Work domain analysis for air traffic controller weather displays. Journal of Safety Research, 36, 159–169. Ahlstrom, U., & Friedman-Berg, F. J. (2006). Using eyemovement activity as a correlate of cognitive workload. International Journal of Industrial Ergonomics, 36, 623–636. Ahlstrom, U., & Jaggard, E. (2010). Automatic Identification of Risky Weather Objects in Line of Flight (AIRWOLF). Transportation Research Part C: Emerging Technologies, 18, 187–192. Ayaz, H., Shewokis, P. A., Bunce, S., Izzetoglu, K., Willems, B., & Onaral, B. (2012). Optical brain monitoring for operator training and mental workload assessment. NeuroImage, 59, 36–47. Civil Air Navigation Services Organisation. (2013). Exploring objective risk measurement [Version 1.0]. Hoofddorp, NL: Civil Air Navigation Services Organisation Safety Standing Committee. Cleveland, W. S., & McGill, R. (1986). An experiment in graphical perception. International Journal of Man-Machine Studies, 25, 491–500. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159. Dienes, Z. (2011). Bayesian versus orthodox statistics: Which side are you on? Psychological Science, 6, 274–290. Farley, T. C., Hansman, R. J., Endsley, M. R., Amonlirdviman, K., & Vigeant-Langlois, L. (1998). The effect of shared information on pilot/controller situation awareness and re-route negotiation. Proceedings of the 2nd International FAA/ EUROCONTROL Air Traffic Management R&D Seminar. Federal Aviation Administration. (2014). Air traffic control (DOT/FAA/Order JO 7110.65V). Washington, DC: Author. Gentaz, E., Luyat, M., Cian, C., Hatwell, Y., Barraud, P. A., & Raphel, C. (2001). The reproduction of vertical and oblique orientations in the visual, haptic, and somato-vestibular systems. The Quarterly Journal of Experimental Psychology, 54A, 513–526. Gigerenzer, G. (2004). Mindless statistics. The Journal of SocioEconomics, 33, 587–606. 2015 Hogrefe Publishing

U. Ahlstrom: AIRWOLF Weather Advisory Tool

2015 Hogrefe Publishing

Plummer, M. (2003). JAGS: A program for analysis of Bayesian graphical models using Gibbs sampling. Proceedings of the 3rd International Workshop on Distributed Statistical Computing. Retrieved from http://www.ci.tuwien.ac.at/ Conferences/DSC-2003/Drafts/Plummer.pdf Plummer, M. (2011). RJAGS: Bayesian graphical models using MCMC. R package version 3-5 [Computer software]. Retrieved from http://CRAN.R-project.org/package=rjagss R Development Core Team. (2011). R: A language and environment for statistical computing [Computer software manual]. Vienna: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org Rosnow, R. L., & Rosenthal, R. (2003). Effect sizes for experimenting psychologists. Canadian Journal of Experimental Psychology, 57(3), 221–237. Steiner, M., Bateman, R. E., Megenhardt, D. L., & Pinto, J. O. (2009). Evaluation of ensemble-based probabilistic weather information for air traffic management. In Aviation, Range and Aerospace Meteorology (ARAM) Special Symposium on Weather–Air Traffic Management Integration, 4.3. Phoenix, AZ: American Meteorological Society. Vigeant-Langlois, L., & Hansman, R. J. (2002). Trajectorybased performance assessment of aviation weather information. In Proceedings of the 10th AMS Conference on Aviation, Range, and Aerospace Meteorology. Portland, OR. Wagenmakers, E.-J. (2007). A practical solution to the pervasive problems of p values. Psychonomic Bulletin & Review, 14, 779–804. Wagenmakers, E.-J., Wetzels, R., Borsboom, D., & van der Maas, H. L. (2011). Why psychologists must change the way they analyze their data: the case of psi. Journal of Personal Social Psychology, 100, 426–432. Wong, D. K. Y., Pitfield, D. E., Caves, R. E., & Appleyard, A. J. (2006). Quantifying and characterizing aviation accident risk factors. Journal of Air Transport Management, 12, 352–357.

Received August 21, 2014 Revision received November 17, 2014 Accepted for publication December 17, 2014 Published online April 10, 2015

Ulf Ahlstrom received a PhD in psychology from Uppsala University, Uppsala, Sweden, in 1994. He is currently an engineering research psychologist with the FAA William J. Hughes Technical Center Human Factors Branch. He has been with the FAA since 1997 following a 2-year postdoctoral fellowship at Vanderbilt University. His current research interests include areas of cockpit weather information displays and operator workload. Correspondence Address Ulf Ahlstrom Federal Aviation Administration FAA Human Factors Branch FAA Technical Center, Bldg. 28 Atlantic City International Airport, NJ 08405 USA Tel. +1 609 485-8642 Fax +1 609 485-6218 E-mail Ulf.Ahlstrom@gmail.com Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):18–35

Original Article

Hansman, R. J., & Davison, H. J. (2000). The effect of shared information on pilot/controller and controller/controller interactions. In Proceedings of the 3rd International FAA/ EUROCONTROL Air Traffic Management R&D Seminar (paper no. 019), June, 2000. Napoli, Italy: FAA/ EUROCONTROL. Harrison, J., Izzetoglu, K., Ayaz, H., Willems, B., Hah, S., Ahlstrom, U., . . . Onaral, B. (2014). Cognitive workload and learning assessment during the implementation of a next generation air traffic control technology using functional near infrared spectroscopy. IEEE Transactions on HumanMachine Systems, 44, 429–440. Hoekstra, J. M., van Gent, R. N. H. W., & Ruigrok, R. C. J. (2002). Designing for safety: The ‘‘free flight’’ air traffic management concept. Reliability Engineering & System Safety. 75, 215–232. Izzetoglu, M., Bunce, S., Izzetoglu, K., Onaral, B., & Pourrezaei, K. (2007). Functional brain imaging using near-infrared technology: Assessing cognitive activity in real-life situations. IEEE Engineering in Medicine and Biology Magazine, 26, 38–46. Kirk, D. B., & Bolczak, R. (2003). Initial evaluation of URET enhancements to support TFM flow initiatives, severe weather avoidance and CPDLC. Proceedings of the 5th Eurocontrol/FAA ATM R&D Seminar, 1–14. Krozel, J., Mitchell, J. S. B., Polishchuk, V., & Prete, J. (2007). Capacity estimation for airspaces with convective weather constraints. In Proceedings of the AIAA Guidance, Navigation, and Control Conference. Hilton Head, SC: American Institute of Aeronautics and Astronautics. Kruschke, J. K. (2011). Doing Bayesian data analysis: A tutorial with R and BUGS. Burlington, MA: Academic Press/Elsevier. doi: 101016/jtics201005001 Kruschke, J. K. (2013). Bayesian estimation supersedes the t test. Journal of Experimental Psychology: General, 142, 573–603. Love, J. F., Chan, W. N., & Lee, C. H. (2009). Analysis of automated aircraft conflict resolution and weather avoidance. Proceedings of the 9th AIAA Aviation Technology. Integration, and Operations Conference (ATIO). Reston, VA: American Institute of Aeronautics and Astronautics. Maclean, I. E., & Stacey, B. G. (1971). Judgment of angle size: An experimental appraisal. Perception & Psychophysics, 9, 499–504. McCrea, M. V., Sherali, H. D., & Trani, A. A. (2008). A probabilistic framework for weather-based rerouting and delay estimations within an airspace planning model. Transportation Research Part C, 16, 410–431. Mitchell, J. S. B., Polishchuk, V., & Krozel, J. (2006). Airspace throughput analysis considering stochastic weather. In Proceedings of the AIAA Guidance, Navigation, and Control Conference. Keystone, CO: American Institute of Aeronautics and Astronautics. Moosakhanian, A., Higginbotham, J., & Stobie, J. (2005). NEXRAD mosaics for en route air traffic controllers. Poster presented at the 32nd Conference on Radar Meteorology. American Meteorological Society. National Transportation Safety Board. (2006, October). Thunderstorm encounters (Safety Alert SA-11). National Transportation Safety Board. Retrieved from http:// download.aopa.org/epilot/2006/061012ntsb-alert.pdf Nguyen, M.-H., Alam, S., Tang, J., & Abbass, H. (2007). Dynamic weather avoidance trajectories in a traffic constrained en route airspace. In Proceedings of the 6th Eurocontrol Innovative Research Workshop. Bretigny sur Orge, France. O’Keefe, D. J. (2003). Colloquy: Should familywise Alpha be adjusted? Against familywise Alpha adjustment. Human Communication Research, 29, 431–447.

Pilots Who Are Perceived as Unsociable Are Perceived as More Likely to Have a Mental Illness An Affective Perspective Scott R. Winter and Stephen Rice

Original Article

Florida Institute of Technology, Melbourne, FL, USA Abstract. The mental state of pilots involved in commercial airlines incidents has been the subject of much debate. The current study seeks to use affective theory to address public perceptions of pilot behaviors and likelihood of perceived mental illness. Participants from India and the United States were given hypothetical scenarios about pilots who were presented as either sociable or unsociable. They were asked to give ratings of affective measures and likelihood of mental illness. The results indicate that pilots who were presented as behaving in an unsociable manner were rated as more likely to have a perceived mental illness compared with those who were behaving sociably. Affect appeared to at least partially mediate the relationship between sociability and perceived likelihood of mental illness for both cultural groups. Keywords: sociability, aviation, pilot, affect, culture

In light of the recent incidents with JetBlue Flight 191 and Malaysia Airlines Flight MH 370, the news media has made much of the mental state of mind of the pilot and copilot, as well as their recent history of sociability. The purpose of the current study was to examine whether pilot sociability affects consumer perceptions of the pilot as possibly suffering from a mental illness and, more specifically, to determine if affect is a possible mediator in the relationship between sociability and perceived mental illness. In addition, we examined possible cultural differences between Indians and Americans.

Defining Sociability Sociability can be defined ‘‘as a tendency to affiliate with others and to prefer being with others to remaining alone’’ (Cheek & Buss, 1981, p. 330). Within a social context, persons may be thought of as introverts (those who avoid or are reserved in social settings) or extroverts (persons who are more outgoing when in social settings). However, like most descriptors, these terms may not be direct and encompassing terms for all individuals. Based on cues obtained through visual or aural means, individuals may form perceptions of a person’s sociability or mental state as biased by stigmas and affect.

Literature Review Attitudes toward mental health may be influenced by a number of variables such as cultural influence, availability of mental health professionals, treatment, rehabilitation regimens, and personal experiences. The literature reviewed for the current study has sought to examine the different stigmas that exist toward mental health in two countries: India and the United States. Additionally, differences were noted in the care given between the two countries and cultures. A review of affect and the affective domain was provided, along with perceptions of how mental health is viewed within the aviation community, especially regarding those who are members of the flight crew.

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44 DOI: 10.1027/2192-0923/a000071

Defining Stigma Stigmas may heavily influence a person’s perspective toward others. They can be defined as prejudices held toward individuals that are either part of, or perceived to be part of, certain groups (Crocker, Major, & Steele, 1998). These prejudices typically result in a diminishing of the person in the eyes of the other. Prior research has been completed on how stigmas influence opinions in other fields (Crocker, Major, & Steele, 1998; Link & Phelan, 2001; Mahjan et al., 2008); however, the current study examined how these stigmas influenced a sample when

2015 Hogrefe Publishing

S. R. Winter & S. Rice: Pilot Sociability

applied in an aviation setting. The findings of this study may be unique for two reasons. First, those in aviation are viewed as holding high responsibility jobs, especially flight crewmembers. Second, due to medical requirements to maintain flying privileges, flight crewmembers may be reluctant to report and seek help for mental health issues. This concern became noticeable in March 2012 when a JetBlue pilot experienced an alleged psychological breakdown midflight (Hunter & Patterson, 2012). The concerned first officer of the aircraft locked the pilot out of the flight deck once he stepped out to use the lavatory, and diverted the New York to Las Vegas flight to Amarillo where the pilot was removed from the aircraft. Stigma of Mental Illness in India

2015 Hogrefe Publishing

Stigma of Mental Illness in the United States Many in the United States have representations of those with mental illness as being violent, uncontrollable, and unable to assume productive roles within society, especially in media depictions (Alexander & Link, 2000). Stigmatization can be harmful because it limits or prevents those needing care from seeking help and treatment. Those perceived of having mental health issues may have difficulty finding jobs, developing social networks, and remaining independent (Link, 1987). After obtaining care, patients often have trouble receiving health benefits, earn less income, and are denied housing (Link, 1987). Additionally, some Americans affected by psychological disorders find themselves homeless, which enhances the negative connotations of mental illness (Goldman & Morrissey, 1985). Unlike in India, people in the United States typically seek the ability to live independently and take responsibility for themselves; those with mental illness are frequently stigmatized as being unable to achieve this status. Therefore, much of the rehabilitation process is centered on helping the patient reenter society and regain his or her personal independence (Stanhope, 2002). Mental illness has also been known to place stress on the individualâ&#x20AC;&#x2122;s family members. Conversely to Indian culture, American family members usually do not share the view of being involved as part of the treatment process, and the afflicted person may even be perceived as a burden (Griffin, 1988). These feelings may be perpetuated by the existing stereotype that persons with mental illness should be sent away for treatment and rehabilitation, further increasing the social distance between society and those with mental illness (Link, Phelan, Bresnahan, Stueve, & Pescolido, 1999). There has also been a social distance correlation demonstrated between perceived violence and the severity of the mental illness (Link et al., 1999). As an example, a person diagnosed with paranoid schizophrenia is stigmatized as being more violent than a person diagnosed with depression. Both India and the United States show notable stigmatization toward mental illness not only within society but also in the structural make-up of both countries. Affect Research on the role that affect has on evaluative processes has increased (Bodenhausen, 1993; Bower, 1991; Clore, Schwarz, & Conway, 1994; Forgas, 1995; Loewenstein, 1996; Schwarz & Clore, 1996; Zajonc, 1998). In particular, this research has focused on increasing the understanding of the emotional influence on decision making. Studies suggest that emotions assist when humans need to process multiple courses of information quickly and there is a need to coordinate physiological, behavioral, and experiential responses (Frijda, 1986; Levenson, 1994; Oatley & Johnson-Laird, 1996). For events that require deliberation, emotions direct attention, memory, and judgment, and these emotions may be influential enough to even interrupt

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36â&#x20AC;&#x201C;44

Original Article

Views of mental health in India vary greatly depending upon location (Jadhav et al., 2007). Those living in rural areas typically demonstrated a higher stigma toward people with mental illness than those who resided in urban areas. However, many Indian communities are of an agrarian nature so those with mental illness are frequently provided opportunities to work despite having psychological disabilities. Somewhat conversely, Indians that resided in urban areas held less stigmatizing views toward those with mental illness, but they were also less willing to work alongside those with psychological disabilities. There also appeared to be a sex difference, with women being more adverse than men and also more likely to be a victim of stigmatization (Thara & Srinivasan, 2000). Half of those receiving care in India were diagnosed with common mental illnesses; these were most frequently associated with poverty and female sex (Patel et al., 1998). Treatment of mental illness within India is completed with a familial focus during each step of the rehabilitation process. Each member of the family is expected to assume a role in the treatment and in the care of those afflicted individuals (Stanhope, 2002). However, while family members in Indian culture tend to be more accepting of mental illness, it can still inhibit personal opportunities such as getting married (Chakraborty, 1997). Mental illness in India is commonly perceived to be the result of karma, a curse or Godâ&#x20AC;&#x2122;s will. A notable difference between India and the United States is the Indian perspective to not place blame or require the individual to take personal responsibility for the illness; rather, focus is placed on creating an environment in which the person may still be able to contribute to the community (Stanhope, 2002). Those with mental illness may be discriminated against due to laws and policies restricting their privileges. For example, within the United States, the mentally ill have limited rights in terms of voting, marriage, and custody of children (Corrigan, Markowitz, & Watson, 2004). These individuals may also be limited in the types of jobs they can perform. Within India, severe limitations exist for providing resources to the mentally ill. Hospitals there have not been adapted to meet the needs of those needing psychological services (Stanhope, 2002).

Original Article

S. R. Winter & S. Rice: Pilot Sociability

cognitive processes (Johnson-Laird & Oatley, 1992; Lazarus, 1991; Schwarz, 1990; Simon, 1967; Tooby & Cosmides, 1990). Therefore, it is possible that stigmas toward those with mental illness are, at least in part, the result of negative emotional reactions (Pryor, Reeder, Yeadon, & Hesson-McInnis, 2004). It is important to highlight the fact that previous research has shown affect and cognition to be separate components (Trafimow & Sheeran, 1998, 2004; Trafimow et al., 2004); however, emotions have also been shown to play an integral role in evaluative judgments (Clore, Schwarz, & Conway, 1994; Schwarz, 1990; Schwarz & Clore, 1983, 1988, 1996). This is especially true in social settings, where persons use emotions to form their opinions. If stigmas produce negative emotions, it is possible that these negative thoughts may produce negative views toward mental health. This influence of affect could possibly explain the stigmatization of various groups of people. Affect is the result of a strong emotional response and therefore may not be the result of cognitive processes. This response could be strong enough to actually impact information processing and judgment (Zajonc, 1980), and it has been shown that affect completes mental models in those situations where cognitive complexity exceeds rational ability (Lee & See, 2004). Previous research with airline consumers (Remy, Winter, & Rice, 2014; Winter, Rice, & Mehta, 2014) has demonstrated that affect completely mediates the relationship between age, weight, and ethnicity with American and Indian participants and how much they trust their pilot. Given that affect has already been shown to be a significant mediator on trust outcomes in persons with a mental illness (Rice, Richardson, & Kraemer, in press), and in the public’s judgments about the need for mental health treatment (Rice & Richardson, in press), it seems plausible that affect would also be a mediator when considering the outcome of mental illness of pilots. In work completed by Alhakami and Slovic (1994), the affect heuristic was studied. This research determined that when persons respond emotionally, it is frequently quickly and unconsciously when determining if something is good or bad. By definition, heuristics offer a form of mental shortcut to quickly arrive at a decision. However, these shortcuts offer no guarantee that the correct conclusion is the one that is made. Additionally, Finucane, Alhakami, Slovic, and Johnson (2000) and Loewenstein, Weber, Hsee, and Welch (2001) discovered the affect heuristic to have a strong inverse relationship when looking at time pressure and emotional responses, respectively. The results of this research on affect seem to indicate that it is likely to have an impact on a person’s perceptions, and it is possible that this impact may influence a person’s view toward the mental health of an individual, particularly their pilot. Mental Health Within Aviation Members of the flight crew are only human, despite the position and responsibility that they hold when operating an aircraft and performing their duties. Therefore, they clearly are vulnerable to psychiatric disorders just as any Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

other members of the population are; however, due to the critical nature of their work, there is little room for error. These disorders can follow a wide spectrum that includes adjustment disorders, mood disorders, psychotic disorders, depression, bipolar disorder, obsessive compulsive disorder, and anxiety disorders, to name simply a few (Morse & Bor, 2006). One could easily imagine how these types of mental illnesses could detrimentally impact a pilot’s ability to concentrate and safely operate an aircraft. Of specific concern to aviation is the stigma and perceived consequences of pilots reporting mental illnesses. Fear of losing their career or medical certification may deter some pilots from seeking help for mental health issues (Bor & Hubbard, 2006). This creates concerns regarding the underreporting of mental illness for pilots. A unique aspect to the aviation world is the demanding travel schedule. While pilots’ schedules vary based on aircraft flown and seniority, inevitably there will be days spent on the road. These schedules may add to stress, not only in the workplace but also on the personal level. Research has shown (Cooper & Sloan, 1985; Raschmann, Patterson, & Schofield, 1990) that strong social support, especially at the spousal level may help act as a buffer against stress in the workplace. Unfortunately, when taken to an extreme, the concept of suicide by aircraft is not something that is unheard of within the aviation community. According to Cullen (1998), when reviewing general aviation accidents within the United Kingdom from 1970 to 1996, somewhere between 0.72% and 2.4% of accidents were attributed to pilot suicide. Within the United States between 1983 and 2003, 37 cases of pilot suicide were detected, and 25% of those were related back to alcohol use, while 14% were related to other illicit substances (Bills, Grabowski, & Li, 2005). However, this is not only limited to general aviation. It is suspected that the 1999 crash of Egypt Air Flight 990 shortly after departure from New York City and a 1997 Silk Air crash in Indonesia were both related to pilot suicide (Morse & Bor, 2006). Due to an absence of data, the actual rates of commercial pilot suicide are difficult to determine, although fortunately, it is considered a rare event (Morse & Bor, 2006).

Current Study The purpose of the current study was to examine how a pilot’s perceived sociability would influence participants’ views of the mental health of their pilot. Participants consisted of individuals from both India and the United States. Additionally, measures of affect were gathered to attempt to determine if affect mediates any of the relationship between sociability and perceived mental illness. Based on previous research, our hypotheses were the following: Hypothesis 1 (H1): Due to stigmas associated with mental illness, participants will rate pilots who are viewed as unsociable as more likely to have a perceived mental illness than their sociable counterparts. 2015 Hogrefe Publishing

S. R. Winter & S. Rice: Pilot Sociability

Hypothesis 2 (H2): American participants will be more extreme in their responses than participants from India. We base this prediction on previous research that has shown that Indians tend to take a less extreme view in these types of hypothetical scenarios compared with Americans (Remy, Winter, & Rice, 2014; Rice et al., 2014; Winter, Rice, & Mehta, 2014). Hypothesis 3 (H3): Affect will, at least partially, mediate the relationship between sociability and perceived mental health (Rice & Richardson, in press; Rice, Richardson, & Kraemer, in press).

Methods

One hundred and thirty-eight participants (51 women) from the United States and 145 participants (52 women) from India took part in the study. The mean age was 33.30 (SD = 12.52) for Americans and 32.41 (SD = 10.95) for Indians. Participants were recruited via a convenience sample using the Amazon Mechanical Turk website (MTurk). MTurk provides an online source of participants who are willing to complete human intelligence tasks in exchange for a small amount of compensation, typically 10–30 US cents. Previous research has shown that data from MTurk is as reliable as normal laboratory data (Buhrmester, Kwang, & Gosling, 2011; Germine et al., 2012). Amazon verifies that only participants from the selected country are allowed to participate in the study, as delimited by the researchers.

Materials and Stimuli The authors created the study instrument because no currently valid and reliable instrument was available for this population. Therefore a limitation to the study was the lack of the instrument being found valid and reliable. Participants first signed an electronic consent form indicating that they were at least 18 years old. Participants were then presented with the following scenario: Imagine that you are on a commercial airline flight from one major city to another. As you are preparing to board, you overhear one of the flight attendants telling the other that the pilot has recently been acting like his usual cheerful self. Please rate how you feel about this: . . . In a separate condition, they were told: Imagine that you are on a commercial airline flight from one major city to another. As you are preparing 2015 Hogrefe Publishing

to board, you overhear one of the flight attendants telling the other that the pilot has recently been acting strange and not like his usual self. He has lost his temper twice in the past two weeks. He has not been very communicative with his crew or friends. He has avoided social media. He has not posted to Facebook in the past month. He has been rude to his co-pilot on several occasions. Please rate how you feel about this: . . . Similar to previous research on affect (Rice, Richardson, & Kraemer, in press; Winter, Rice, & Mehta, 2014), participants were then asked to rate their feelings on a 7-point Likert-type scale from extremely negative ( 3) to extremely positive (+3) with a neutral zero option. Following this, participants were given the following statements and asked to rate how strongly they agreed or disagreed with the statements on a 7-point Likert-type scale from extremely disagree ( 3) to extremely agree (+3) with a neutral zero option. • This pilot is likely suffering from a mental illness. • This pilot would likely be classified as mentally ill if he were seen by a psychiatrist. Following this, participants were asked for basic demographic information and dismissed.

Design A two-way between-participants design was employed whereby Country of Origin and Sociability were the manipulated factors.

Results There were no noticeable sex differences in the data (all ps > .05); however, due to the large disparity in the number of female and male participants, sex analyses cannot be considered highly reliable in this study. All further analyses were conducted after combining the sex data.

Statistical Analysis The data were first analyzed using a two-way factorial ANOVA with Country and Sociability as betweenparticipants factors. For the Affect data, there was a marginally significant effect of Country, F(1,279) = 3.79, p = .053, g2 = .01, and Sociability, F(1,279) = 261.31, p < .001, g2 = .48. This was qualified by a significant interaction between Country and Sociability, F(1,279) = 37.29, p < .001, g2 = .12. These data can be found in Figure 1 and reveal that (a) there was a strong effect of Sociability on the ratings of affect and (b) US participants produced stronger affective responses for both ends of the spectrum, and particularly for the unsociable condition. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

Original Article

Participants

S. R. Winter & S. Rice: Pilot Sociability

Affect

0 -1

Original Article

India

0 -1

-2 2 -3

Mental Illness

Ratings

-2 Sociable 1.69

Unsociable -1.91

1.01

-0.61

-3

Sociable

Unsociable

-2.11

-0.03

India

-0.67

0.77

Figure 1. Affect data from the study. Standard error bars are included.

Figure 2. Mental illness ratings from the study. Standard error bars are included.

The data for the mental illness questions were subjected to Cronbach’s alpha test and found to be internally consistent (.88); therefore, we combined this data into one measure. The two-way ANOVA revealed a main effect of Country, F(1,279) = 45.52, p < .001, g2 = .13, and Sociability, F(1,279) = 105.11, p < .001, g2 = .27, with a marginally significant interaction between the two variables, F(1,279) = 3.49, p = .063, g2 = .01. These data can be found in Figure 2 and reveal that (a) there was a strong effect of Sociability on the ratings of mental illness likelihood and (b) Indian participants were, in general, more likely to assume the presence of a mental illness.

General Discussion

Mediation Analyses For the sake of brevity, the ‘‘ratings of likelihood of mental illness’’ was shortened to ‘‘mental illness’’ for these analyses. The paths for the mediation analyses conducted on Indian participants can be found in Figure 3. To conduct the mediation analysis, the correlation between Condition (sociable or unsociable) and Mental Illness was first found to be significant, r = .390, p < .001, showing that the initial variable correlated with the outcome variable. The standardized path coefficients were condition to affect (.447, p < .001); affect to mental illness (.279, p = .001); condition to mental illness controlling for affect (.266; p = .002). These data showed that Affect had some mediation effect on the relationship between Condition and Mental Illness. The paths for the mediation analyses conducted on US participants can be found in Figure 4. To conduct the mediation analysis, the correlation between Condition and Mental Illness was first found to be significant, r = .691, p < .001, showing that the initial variable correlated with the outcome variable. The standardized path coefficients were condition to affect (.880, p < .001); affect to mental illness (.481, p < .001); condition to mental illness controlling for affect (.268; p = .033). These data showed that Affect had a strong mediation effect on the relationship between Condition and Mental Illness. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

The purpose of the current study was to examine how a pilot’s perceived sociability level would affect participant’s ratings of the pilot’s perceived mental health. Participants from both India and the United States were sampled for this experiment. Additionally, data were gathered on Affect measures to determine if Affect had any mediation effect between the condition (sociable or unsociable) and the outcome variable, likelihood of mental illness. From the data, it appears that sociability level does have an influence on Affect and perceived mental illness ratings of the pilot, and that Affect does mediate, at least some, of the relationship between the condition and the outcome variable. The first hypothesis sought to determine how the level of pilot sociability would influence participant’s views toward the mental health of their pilot. The findings suggested that when a pilot was presented as being unsociable, participants rated the pilot as being more likely of having a mental illness than when the pilot was presented as sociable, which supports the hypothesis set forth by the researchers. A possible explanation for this outcome may be related to stigmas. There has been recent media attention given to pilot medical qualifications and perhaps a lack in mental health screenings for pilots. Given events such as the recent Malaysia Airlines missing aircraft and the case of a JetBlue flight that diverted after the pilot suffered an alleged mental breakdown in flight (Hunter & Patterson, 2012), participants may be more aware and perhaps more concerned over the mental health of their pilot. These reports may be exacerbated by the stigmas toward those with mental illness and influencing the decision making of consumers (Crocker, Major, & Steele, 1998; Link & Phelan, 2001; Mahjan et al., 2008). Pilots are typically held in high regard and hold positions of great responsibly when taking command of an aircraft for a flight. If their behavior has recently changed, it may be indicative of increased stress in their lives, which may have an influence on their mental health state. 2015 Hogrefe Publishing

S. R. Winter & S. Rice: Pilot Sociability

Figure 3. Path analysis for Indian participants.

Based on previous research (Remy, Winter, & Rice, 2014; Rice et al., 2014; Winter, Rice, & Mehta, 2014), the second hypothesis posed by the researchers expected American participants’ ratings of Affect and mental illness to be more extreme than the Indian participants’ ratings. This hypothesis was partially supported by the findings of the study. When providing ratings for Affect, Americans were more extreme in their responses based on the sociability level of their pilot. However, when rating mental illness, Americans were more extreme than Indians in the sociable condition, but less extreme than Indians in the unsociable condition. This finding is of particular interest because it is slightly different from the findings of previous work where Americans tended to have more extreme ratings in both conditions than Indian participants. It is possible that some of the cultural differences between the two countries may explain these differences. Americans tend to be a more individualistic culture, and Indians are more likely to be collectivist (Robbins & Judge, 2009). Those from collectivist cultures traditionally hold an interdependent view of themselves (Markus & Kitayama, 1991), and Hofstede (1980) found that these cultural tendencies influence a person’s propensity to trust. Research has found that those from collectivist cultures are more likely to trust without question compared with those from individualistic cultures. It is possible that these cultural influences, at least partially, explain the findings related to Affect ratings. These ratings of perceived mental illness are different from what was hypothesized by the researchers. In the sociable condition, Americans are more extreme in their ratings than Indian participants, but this is reversed in the unsociable condition, with Americans being mostly neutral and Indians being more suggestive that the pilot has a mental illness. However, cultural influences may also possibly explain these findings. A major difference between the current study and previous work relates to what participants

2015 Hogrefe Publishing

were rating. In earlier work, participants were rating inanimate objects, such as autopilots (Rice et al., 2014) or providing somewhat superficial ratings of trust based on sociodemographic factors (i.e., age, weight, ethnicity) (Remy, Winter, & Rice, 2014; Winter, Rice, & Mehta, 2014). However, in the current study, participants were asked to rate their feelings toward a person who may or may not be suffering from mental illness. Part of being from a collectivist culture is placing a value on concern for, and the well-being of, others, at times perhaps even placing the well-being of others before that of oneself (Markus & Kitayama, 1991). Additionally, those from Indian culture assume more responsibility toward the treatment of those suffering from mental illness when compared with Americans (Stanhope, 2002). Therefore, if presented with someone being unsociable, Indian participants may be responding with more concern for the individual than participants from the American sample. The final hypothesis predicted that Affect would at least partially mediate the relationship between the condition (sociable or unsociable) and the likelihood of mental illness of the pilot. The findings supported this hypothesis with some mediation effect between the condition and mental illness with the Indian participants, and a strong mediation effect between the condition and mental illness with American participants. Obviously, diagnosing something like the mental health of a person is a process that needs to be completed by a medical professional, and a small snippet of information overheard between two flight attendants may not provide the entire picture. However, it is possible that stigmas toward mental health are influencing participants’ ratings. Stigmas can have powerful influences over opinions and perceptions formed (Alexander & Link, 2000; Zajonc, 1980), especially given their appearance in sources like the media. Prior research has shown that even though cognitive and emotional responses are separate (Trafimow & Sheeran, 1998, 2004; Trafimow et al., 2004), emotions still play a role in the decision-making process (Clore, Schwarz, & Conway, 1994; Schwarz, 1990; Schwarz & Clore, 1983, 1988, 1996). Therefore, when persons have to make judgments in short periods of time or with little information, it is possible that emotions influence their responses. This type of effect has been seen before in the mental health literature (e.g., Rice & Richardson, in press; Rice, Richardson, & Kraemer, in press).

Practical Implications and Limitations The current study may have some practical implications. First, consumers seem sensitive to information regarding the mental health of their pilot. While their responses may have been influenced by cultural tendencies, both populations became more concerned regarding the mental health of their pilot based on changes to sociability. Secondly, Affect appears to play a role in the judgmentmaking process of participants. While these emotional

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

Original Article

Figure 4. Path analysis for American participants.

Original Article

S. R. Winter & S. Rice: Pilot Sociability

reactions are strong, it is possible that more information regarding mental health, and perhaps even on the mental health requirements for pilots, is needed. This factual information may help provide more accurate information to the populations, and reduce the impact of emotion on participant responses. There were some limitations to the current study. First, the study was limited to those persons who completed online human intelligence tasks. While this data has been shown to be as reliable as laboratory data (Buhrmester, Kwang, & Gosling, 2011; Germine et al., 2012), further research should be undertaken with an expanded population to increase the generalizability of these findings. Second, there was a large disparity in the sex data, with almost a 2:1 ratio of male to female participants. Future research should investigate sex differences using a larger and more balanced sample. Third, participants produced a response related to perceived mental illness, because they were only able to offer their opinions and not any type of formal diagnosis. Finally, a new instrument was created for this study by the authors. Further research can verify the validity and reliability of this instrument. This study was also being conducted in the aftermath of the Malaysia Airlines accident/missing aircraft incident. At the time of the study, little factual information has been presented, and the media has discussed a number of factors, such as pilot mental health. Due to the global nature of this event, it may be possible that these events have influenced or made participants more aware or concerned regarding the issue of pilot mental health. Replicating the study at a later time may help verify the accuracy of the current study.

Conclusions The current study sought to determine how the level of sociability of a pilot would influence participant’s views of his or her mental health across two populations of participants: Indians and Americans. Participants rated a sociable pilot as being less likely to have a mental illness than one that was describe as unsocial. Additionally, Affect was found to mediate at least some of the relationship between the condition and the ratings for mental illness. Cultural influences and views regarding stigmas of those with mental illness may offer possible explanations for some of the differences between the two populations.

References Alexander, L. A., & Link, B. G. (2003). The impact of contact on stigmatizing attitudes toward people with mental illness. Journal of Mental Health, 12, 271–289. Alhakami, A. S., & Slovic, P. (1994). A psychological study of the inverse relationship between perceived risk and perceived benefit. Risk Analysis, 14, 1085–1096. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

Bills, C., Grabowski, J., & Li, G. (2005). Suicide by aircraft: A comparative analysis. Aviation, Space, and Environmental Medicine, 76, 715–719. Bodenhausen, G. V. (1993). Emotions, arousal, and stereotypic judgments: A heuristic model of affect and stereotyping. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 13–37). San Diego, CA: Academic Press. Bor, R., & Hubbard, T. (2006). Aviation mental health: An introduction. In R. Bor & T. Huddard (Eds.), Aviation mental health (pp. 1–9). Burlington, VT: Ashgate. Bower, G. H. (1991). Mood congruity of social judgment. In J. Forgas (Ed.), Emotion and social judgment (pp. 31–54). Oxford, UK: Pergamon. Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet highquality data? Perspectives on Psychological Science, 6, 3–5. Chakraborty, A. (1997). Stigma of schizophrenia. Indian Journal of Social Psychiatry, 13, 93–97. Cheek, J. M., & Buss, A. H. (1981). Shyness and sociability. Journal of Personality and Social Psychology, 41, 330–339. Clore, G. L., Schwarz, N., & Conway, M. (1994). Cognitive causes and consequences of emotion. In R. S. Wyer Jr & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., pp. 323–417). Hillsdale, NJ: Erlbaum. Cooper, C., & Sloan, S. (1985). Occupational and psychological stress among commercial airline pilots. Aviation, Space, and Environmental Medicine, 56, 317–321. Corrigan, P. W., Markowitz, F. E., & Watson, A. C. (2004). Structural levels of mental illness stigma and discrimination. Schizophrenia Bulletin, 30, 481–491. Crocker, J., Major, B., & Steele, C. (1998). Social stigma. In D. T. Gilbert & S. T. Fiske (Eds.), The handbook of social psychology (pp. 504–553). New York: McGraw-Hill. Cullen, S. (1998). Aviation suicide: A review of general aviation accidents in the UK, 1970–96. Aviation, Space, and Environmental Medicine, 69, 696–698. Finucane, M. L., Alhakami, A., Slovic, P., & Johnson, S. M. (2000). The affect heuristic in judgments of risks and benefits. Journal of Behavioral Decision Making, 13, 1–17. Forgas, J. P. (1995). Mood and judgment: The affect infusion model (AIM). Psychological Bulletin, 117, 39–66. Frijda, N. H. (1986). The emotions. Cambridge, UK: Cambridge University Press. Germine, L., Nakayama, K., Duchaine, B.C., Chabris, C. F., Chatterjee, G., & Wilmer, J. B. (2012). Is the web as good as the lab? Comparable performance from web and lab in cognitive/perceptual experiments. Psychonomic Bulletin & Review, 19, 847–857. Goldman, H., & Morrissey, J. (1985). The alchemy of mental health policy: Homelessness and the fourth cycle of reform. American Journal of Public Health, 75, 728–731. Griffin, Q. (1988). Families’ perceptions of burden of care for chronic mentally ill relatives. Hospital and Community Psychiatry, 9, 1296–1300. Hofstede, G. (1980). Motivation, leadership and organization: Do American theories apply abroad? Organizational Dynamics, 9, 42–63. Hunter, M., & Patterson, T. (2012, March 3). Pilot breakdown draws attention to mental health standards. CNN.com. Retrieved from http://www.cnn.com/2012/03/28/travel/ airline-crew-mental-health/ Jadhav, S., Littlewood, R., Ryder, A. G., Chakraborty, A., Jain, S., & Barua, M. (2007). Stigmatization of severe mental illness in India: Against the simple industrialization hypothesis. Indian Journal of Psychiatry, 49(3), 189–194. Johnson-Laird, P. N., & Oatley, K. (1992). Basic emotions, rationality, and folk theory. Cognition and Emotion, 6, 201–223. 2015 Hogrefe Publishing

S. R. Winter & S. Rice: Pilot Sociability

2015 Hogrefe Publishing

Richardson, J., & Rice, S. (in press). The role of affect and cognition in the public’s judgments about the need for mental health treatment and willingness to help. International Journal of Mental Health. Robbins, S. P., & Judge, T. A. (2009). Organizational behavior (13th ed.). Upper Saddle River, NJ: Prentice Hall. Schwarz, N. (1990). Feelings as information: Informational and motivational functions of affective states. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 527–561). New York: Guilford Press. Schwarz, N., & Clore, G.L. (1983). Mood, misattribution and judgments of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology, 45, 513–523. Schwarz, N., & Clore, G. (1988). How do I feel about it? The information function of affective states. In K. Fiedler & J. P. Forgas (Eds.), Affect, cognition, and social behavior: New evidence and integrative attempts (pp. 44–63). Toronto: C. J. Hogrefe. Schwarz, N., & Clore, G. (1996). Feelings and phenomenal experiences. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 433–465). New York: Guilford Press. Simon, H. A. (1967). Motivational and emotional controls of cognition. Psychological Review, 74, 29–39. Stanhope, V. (2002). Culture, control, and family involvement: A comparison of psychosocial rehabilitation in India and the United States. Psychiatric Rehabilitation Journal, 25, 273–280. Thara, R., & Srinivasan, T. N. (2000). How stigmatising is schizophrenia in India? International Journal of Social Psychiatry, 46, 135–141. Tooby, J., & Cosmides, L. (1990). The past explains the present: Emotional adaptations and the structure of ancestral environments. Ethology and Sociobiology, 11, 375–424. Trafimow, D., & Sheeran, P. (1998). Some tests of the distinction between cognitive and affective beliefs. Journal of Experimental Social Psychology, 34, 378–397. Trafimow, D., & Sheeran, P. (2004). A theory about the translation of cognition into affect and behavior. In G. Maio & G. Haddock (Eds.), Contemporary perspectives in the psychology of attitudes: The Cardiff Symposium (pp. 57–76). London: Psychology Press. Trafimow, D., Sheeran, P., Lombardo, B., Finlay, K. A., Brown, J., & Armitage, C. J. (2004). Affective and cognitive control of persons and behaviors. British Journal of Social Psychology, 43, 207–224. Winter, S. R., Rice, S., & Mehta, R. (2014). Aviation consumers’ trust in pilots: A cognitive or emotional function. International Journal of Aviation, Aeronautics, and Aerospace, 1(1), 1–18. Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151–175. Zajonc, R. (1998). Emotions. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (Vol. 1, pp. 591–632). New York: Oxford University Press.

Received May 12, 2014 Revision received November 11, 2014 Accepted for publication December 18, 2014 Published online April 10, 2015

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

Original Article

Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press. Lee, J. D., & See, A. K. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46, 50–80. Levenson, R. (1994). Human emotion: a functional view. In P. Ekman & R. J. Davidson (Eds.), The nature of emotion (pp. 123–126). New York: Oxford University Press. Link, B. G. (1987). Understanding labeling effects in the area of mental disorders: An assessment of the effects of expectationfs of rejection. American Sociological Review, 52, 96– 112. Link, B. G., & Phelan, J. C. (2001). Conceptualizing stigma. Annual Review Sociology, 27, 363–385. Link, B. G., Phelan, J. C., Bresnahan, M., Stueve, A., & Pescolido, B. (1999). Public conceptions of mental illness: Labels, causes, dangerousness, and social distance. American Journal of Public Health, 89, 1328–1333. Loewenstein, G. (1996). Out of control: Visceral influences on behavior. Organizational Behavior and Human Decision Processes, 65, 272–292. Loewenstein, G., Weber, E., Hsee, C., & Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127, 267–286. Mahjan, A. P., Sayles, J. N., Patel, V. A., Remien, R. H., Ortiz, D., Szekeres, G., & Coates, T. J. (2008). Stigma in the HIV/ AIDS epidemic: A review of the literature and recommendations for the way forward. AIDS, 22, 67–79. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224–253. Morse, J. S., & Bor, R. (2006). Psychiatric disorders and syndromes among pilots. In R. Bor & T. Huddard (Eds.), Aviation mental health (pp. 107–125). Burlington, VT: Ashgate. Oatley, K., & Johnson-Laird, P. N. (1996). The communicative theory of emotions: Empirical tests, mental models, and implications for social interaction. In L. L. Martin & A. Tesser (Eds.), Striving and feeling: Interactions among goals, affect, and self-regulation. Hillsdale, NJ: Erlbaum. Patel, V., Pereira, J., Coutinho, L., Fernandes, R., Fernandes, J., & Mann, A. (1998). Poverty, psychological disorder and disability in primary care attenders in Goa, India. The British Journal of Psychiatry, 172, 533–536. Pryor, J. B., Reeder, G. D., Yeadon, C., & Hesson-McInnis, M. (2004). A dual-process model for reactions to perceived stigma. Journal of Personality and Social Psychology, 87, 436–452. Raschmann, J., Patterson, J., & Schofield, G. (1990). A retrospective study of material discord in pilots: The USAFSAM experience. Aviation, Space, and Environmental Medicine, 61, 1145–1148. Remy, B., Winter, S. R., & Rice, S. (2014, April). American aviation consumer’s trust in pilots. Presentation at the 7th annual Human Factors and Applied Psychology Student Conference, Daytona Beach, FL. Rice, S., Kraemer, K., Winter, S. R., Mehta, R., Dunbar, V., Rosser, T. G., & Moore, J. C. (2014). Passengers from India and the United States have differential opinions about autonomous auto-pilots for commercial flights. International Journal of Aviation, Aeronautics, and Aerospace, 1(1), 1–12. Rice, S., Richardson, J., & Kraemer, K. (in press). The emotional mediation of distrust of persons with a mental illness. International Journal of Mental Health.

S. R. Winter & S. Rice: Pilot Sociability

Scott R. Winter (PhD, ATP) is an assistant professor of aviation science at the Florida Institute of Technology. He completed his doctorate from Purdue University in 2013, where his dissertation research focused on pilot decision making in irreversible emergencies. He presently conducts research in three foundational areas: pilots’ transition and information processing in glass cockpit aircraft, training pilots in very light jet operations, and enhancement methods for pilot cognition and decision making.

Stephen Rice is an associate professor of human factors and the chair of the Graduate Program in the College of Aeronautics at the Florida Institute of Technology. He received his BA in psychology from Rollins College and an MA and PhD in experimental psychology from the University of Illinois at Urbana-Champaign (UIUC). Stephen Rice’s research focuses on the intersections between aviation human factors, technology, society, education, and aviation consumer perceptions.

Correspondence Address

Original Article

Scott R. Winter College of Aeronautics Florida Institute of Technology 150 West University Blvd. Melbourne, FL 32901 USA Tel. +1 218 269-9376 E-mail scott.winter@mac.com

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):36–44

2015 Hogrefe Publishing

A Request for Regulatory Revision Instructions for Passenger Bracing for Emergency Landings Kyung In Yoo1 and Jan M. Davies2 1

Aviation Service Department, Hotel and Tourism College, Wonkwang Health Science University, Iksan, Jeonbuk, Korea (R.O.K.), 2Department of Anesthesia, Cumming School of Medicine, University of Calgary, Foothills Medical Centre, Calgary, AB, Canada Abstract. Although aviation grows safer, accidents still occur, with unanticipated/unplanned and anticipated/planned emergency landings. The latter give passengers time to adopt the brace position before landing. Bracing minimizes the physical effects of multiple, sequential impacts in a crash. Passengers who do not brace are more likely to be injured and less likely to escape unaided from the aircraft, with risk of further injury or death. Passenger safety cards include brace position information, but most preflight passenger briefings omit information and demonstration about bracing. These are airline specific and, with a few exceptions, not required by national aviation regulations, because of a lack of International Civil Aviation Organization (ICAO) requirements. If ICAO regulations required brace position briefings, these would help improve international airline passenger safety. Keywords: aviation safety, cabin safety, brace position, preflight briefings, regulatory requirements

Factors Minimizing Fatalities and Injuries A number of different factors have contributed to air travel’s overall increasing safety and specifically to improved Ó 2015 Hogrefe Publishing

survivability of aircraft accidents. In the preaccident and postaccident stages, safety management systems, including investigations, have helped reduce the occurrence of accidents and incidents. Better aircraft designs now provide increased occupant protection in impact-survivable accidents. These improvements include changes to seats from some of the ‘‘lethal design features’’ described by Swearingen in his 1966 head and face injury impact studies (Swearingen, 1966). Other changes include more (but not completely) fire-retardant materials (Federal Aviation Administration Fire Safety, 2014) and seats designed to withstand 16g (Federal Aviation Administration, 2010). However, not all passengers fly in the most modern and well-equipped of aircraft. What is common and important to all passengers is the brace position. This is described as a precrash position in which the body is positioned ‘‘against whatever it is most likely to hit during the crash,’’ thus minimizing any secondary impact (Barthelmess, 1988). In 1979, the National Transportation Safety Board described reports in which passengers who were in a brace position sustained fewer fatal and significantly less severe injuries during accidents than did those not braced (National Transportation Safety Board, 1979). Since then, a number of scientific studies and crash investigations, as well as postmortem analyses, have been conducted, including those by Chandler (1988); White, Rowles, Mumford, and Firth (1990); Dulcavsky, Geller, and Iorio (1993); White, Firth, Rowles, and N.L.D.B. Study Group (1993); Brownson, Wallace, and Anton (1998); Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51 DOI: 10.1027/2192-0923/a000072

APAHF in Practice

Despite recent tragic events, air travel continues to become safer, with the global accident rate for every million departures decreasing from 4.4 in 2005 (ICAO, 2014) to 2.8 in 2013 (ICAO, 2014 ). Passenger and crew fatalities have also dropped, from 1,056 deaths in 2005 to 220 in 2013, although 675 died in 2014 (Airline Safety, 2015). However, many would consider these deaths, as well as the injuries suffered by numerous other passengers and crew, as potentially avoidable. One factor common to all passengers and crew to help reduce their deaths and injuries, no matter what type of aircraft they fly in, is the brace position. This paper provides an overview of factors minimizing fatalities and injuries – specifically, the brace position and its influence on passenger deaths and injuries in emergency landings, including when and how passengers are more or less likely to be killed or injured. Next, the role of passengers in emergency landings is discussed. This is followed by a description of the presence and absence of regulatory requirements for briefing passengers on the brace position. Finally, we present a draft recommendation to ICAO that information about the brace position be provided to passengers both in the preflight passenger briefing and on passenger safety cards.

APAHF in Practice

K. I. Yoo & J. M. Davies: Brace for Emergency Landings

Grierson and Jones (2001); and Cullen (2004). These reports have shown there are two primary reasons for bracing. The first is to reduce flailing of the body and lower limbs around the fixed point of the seat belt (Swearingen, 1966) at the time of the first impact, from which many injuries are sustained and deaths result. The second is to reduce effects of subsequent impact(s), as the body recoils and then restrikes parts of the aircraft’s interior. This secondary impact also leads to head injuries, which are very common in aviation accidents and may cause or contribute to death (Cullen, 2004; White et al., 1990). Despite positive statements about the brace position from these publications, two major points should be made. First, there is no single brace position (Sperber et al., 2010). Rather, the best position will depend on a number of factors, including the passenger’s age, height, weight, body mass distribution, and mobility/flexibility, as well as the presence of any physiological or psychological disabilities (Sperber et al., 2010). The airline class (economy vs. business or first) will determine the pitch of the seating. Although seat pitch is defined as the ‘‘seat to seat distance from one point on the seat to the same point on the seat immediately’’ in front or behind (Transport Canada, 1999), most passengers interpret the term as determining how much or little leg room there is, which will affect their ability to attain some versions of the brace position. In addition, airline class means that on certain airlines, passengers in economy class will have simple lap belts while some in business or first will have lap belts with built-in air bags. This difference will also determine which brace position should be adopted. Second, although as stated above, there have been various studies to support the adoption of the brace position, most of the available papers lack validity. The majority of publications in this area give anecdotal reports from accidents and investigations, although there are a few scientific studies, such as those by Swearingen (1966) and Brownson et al. (1998) and in Sperber et al. (2010). There is also variability in the details of certain positions. For example, should the feet be placed forward (Transport Canada, 2012a), beneath the knees (Johnson, 1998), or angled behind the knees (Koenig, 1995)? One publication shows an illustration of a ‘‘feet forward’’ position but with the caveat that ‘‘feet position may vary among States’’ (Cabin Safety Team, 2001). A recent study, using dynamic 16g tests, aircraft passenger seats, and commercial crash test dummies, recommended that passengers extend their legs and ‘‘if possible, place them flat against the rigid structure of the seat in front.’’ In addition, the document advised passengers to ‘‘put any luggage under the seat in front and push it up to the front’’ and to put their ‘‘feet against the piece of luggage’’ (Sperber et al., 2010). These results would seem to counter those of Brownson et al. (1998), whose recommendations were also based on similar 16g testing of crash test dummies. Their recommendations were for the feet to be positioned ‘‘slightly rearward of the knee,’’ so as to

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51

minimize the ‘‘tendency for the upper limbs to flail upward on impact.’’ Both studies used dummies that simulated adult males with a height and weight equivalent to the 50th percentile of the US adult male population. These results are therefore not directly applicable to other passengers such as children, smaller females, and the obese. These two publications show some of the difficulties in being able to choose and then recommend one brace position for one group of adult passengers sitting in one type of seat in one airline class in one type of aircraft, let alone all passengers. But despite the lack of studies that cover the entire range of passengers, seating arrangements, and airplanes, the existing information represents the ‘‘best available evidence’’ (Schunemann & Guyatt, 2014). Furthermore, many passengers can and do survive accidents. Calculations of the numbers of passengers who might be killed or injured were carried out by researchers sponsored by the US Department of Transport. Using statistics from 17 transport category aircraft accidents for 1993– 1996, they determined the percentages of 100 occupants who would die (20%) or suffer serious injuries (20%) or minor injuries (20%) at the time of the first impact. A subsequent fire would reduce the percentage of survivors from 80% to 70%, with 25% suffering serious injuries/burns and 45% with minor injuries, thus demonstrating the need to be able to evacuate the aircraft and hence the importance of the brace position (Cherry, Warren, & Chan, 2000).

Types of Emergency Landings Most fatalities and injuries occur on landing (apart from inflight breakup of an aircraft). Anticipated and planned emergency landings afford time to prepare for landing or ditching. Resulting cabin conditions and expediency of evacuation are determining factors in survivability – for example, as in United Airlines Flight 634 at Newark, New Jersey, in 2010. In that event, the crew were unable to extend the right landing gear on approach. A go-around was carried out, followed by failure to manually extend the gear and a decision to land before fuel ran low (National Transportation Safety Board, 2011). Just before landing, the captain gave the ‘‘Brace Brace Brace’’ command (Hutchinson, 2010); passengers assumed the position of ‘‘holding the ankles with head between the knees’’ (Sulzberger & Schweber, 2010). There were no fatalities among the 48 passengers, and only three suffered minor injuries, all sustained during the emergency evacuation. A second type of landing is one not anticipated to be an emergency until after the aircraft touches down – for example, Air France Flight 358 at Toronto, Ontario, in 2005 (Transportation Safety Board of Canada, 2007) and American Airlines Flight 331 at Kingston, Jamaica, in 2009 (Jamaica Civil Aviation Authority, 2014). In both accidents, the aircraft unexpectedly overran the runway and then suffered major damage, breaking up in the postlanding trajectory. In neither accident was the brace position command

Ó 2015 Hogrefe Publishing

K. I. Yoo & J. M. Davies: Brace for Emergency Landings

given. While there were no fatalities, passengers were injured in both accidents, some seriously.1

Phases of Casualty Occurrence

Role of Passengers in an Emergency Landing Thus, when an aircraft crashes, passengers’ safety relies on three consecutive, linked steps: protecting themselves during the crash, evacuating rapidly, and surviving the aftermath. Passengers need to know what these actions are and how to undertake them spontaneously and appropriately. It is quite clear, however, that passengers will not be able to undertake the actions required in the second and third phases if they are not able to minimize their injuries (or death) in the first phase. For example, passengers’ actions may prove life-saving during an emergency landing, notwithstanding the important and legally required role of flight attendants in 1

passenger safety (Muir & Thomas, 2004). Passengers represent the majority of cabin occupants, and as previous accidents have shown, passengers can understand and undertake a role in an emergency. For example, in 1984, Pacific Western Airlines Flight 501 developed a fuel line rupture 1,300 feet into the takeoff roll at Calgary, Alberta. Leaking fuel ignited instantly, engulfing the aircraft’s left wing and aft section. There was no general announcement of the evacuation made by either the captain or the flight attendants. . .. Passengers’ decisions to leave their seats and evacuate were based on their perceptions of the emergency. . .. The right over-wing exit was opened by the passenger seated next to it at the urging of several passengers seated nearby. (Tucker, 1989, pp. 4 f.) About 40 of the 114 passengers used this overwing exit. Numerous passengers had minor injuries, including singeing of hair and skin reddening from the heat. There were no fatalities. The fact that ‘‘most passengers were regular travelers familiar with the Boeing 737’’ was considered to contribute to the success of the evacuation (Tucker, 1989).

How Passengers Learn About the Brace Position On every flight, passengers normally learn what to do in an aircraft emergency from the preflight briefing delivered by the cabin crew sometimes supplemented or replaced by a video. There is often no information presented about the brace position, although some information is provided in the briefing card in each passenger’s seat pocket. However, many passengers rarely look at preflight safety demonstrations or videos, or read the passenger safety cards (Muir & Thomas, 2004). In addition to those passengers who are ‘‘to those passengers who are ‘‘non-attenders’’ (Johnson, 1979), other passengers will have difficulty with the comprehensibility of the safety information (Muir & Thomas, 2004) including that presented as pictorials (Caird et al., 1997; Chang, 2013; Corbett et al., 2008). When an in-flight emergency occurs, if time is available, cabin crew check to ensure passengers’ seat belts, seat backs, and tray tables are secured. The crew may (or may not) verbally describe and then demonstrate a brace position. Timing and content of this information will vary, depending on the time available (approximately more than 15–20 minutes) and the specific passenger air carrier. More information may be given when the aircraft reaches 1,500 feet, to ensure cabin preparation is as complete as possible for the emergency landing/ditching. At this altitude, the captain will usually order cabin crewmembers to be seated in their emergency station jump seats. At 30 s before

Both these types of landings are in contrast to unanticipated and unplanned landings, as in controlled flight into terrain (CFIT). With these, there is frequently time neither to correct the aircraft’s course nor for cabin crew or passengers to take any preparatory safety measures in the cabin. CFIT will not be discussed further.

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51

APAHF in Practice

Whether anticipated or not, most emergency landings may be associated with very large impact forces, starting with the aircraft’s initial contact with the terrain (ground or water), resulting in massive, sudden deceleration. There is then progression through the subsequent and varying destruction of the aircraft itself. It is these impact forces and related deceleration that contribute to serious injuries and fatalities. Thus the first phase of casualty occurrence is the initial crash impact and subsequent multiple, sequential impacts that occur until the aircraft comes to a final stop. The second phase is the emergency evacuation from the aircraft. In many accidents, postcrash fires are frequent and life-threatening to surviving passengers and crewmembers. For example, after the Air China International Flight 129 accident in Gimhae, Korea, in 2001, autopsies of 123 of the victims showed soot in the tracheas of 16, suggesting they might have been alive at the time of the fire (Korea Aviation and Accident Investigation Board, 2005). Rapid evacuation is essential but may require movement around, under, over, or through shattered parts of the aircraft, any or all of which could impede progress and/or inflict further injuries. Attempting to do this when injured may be impossible. The third phase is that of surviving in the postevacuation environment. Emergency aircraft landings or accidents occur in all types of marine and ground conditions and in all types of weather. Crash survivors may need to survive hostile conditions and injuries, until search and rescue teams arrive. For example, after the Air China accident, surviving passengers and crew waited in rain and fog for rescuers to reach them on a densely treed mountainside (Korea Aviation and Accident Investigation Board, 2005).

APAHF in Practice

K. I. Yoo & J. M. Davies: Brace for Emergency Landings

touchdown, if possible, the captain will order adoption of the brace position, stating ‘‘Brace! Brace! Brace!’’ through the overhead system and repeatedly turning on and off the Fasten Seat Belts signs (the brace for impact signal). Cabin crew will repeat the order, as loudly and clearly as possible for passengers to adopt and maintain the brace position. In contrast, for an unanticipated emergency, should cabin crew think that passengers need to adopt the brace position, the crew may shout commands such as ‘‘Bend over, grab ankles’’ or ‘‘Head down, stay low.’’ Unfortunately, these commands may be difficult, if not impossible, for passengers to comprehend and to react to immediately, to prepare for imminent impact (Garner & Blethrow, 1970). First, the cabin environment is likely to be loud and chaotic, with noise from disruption of various parts of the aircraft and passengers’ screams. (Even the relatively low levels of noise that occur during normal commercial aircraft operations have been shown to have a negative effect on cognitive performance; Molesworth, Burgess, Gunnell, & Duong, 2014.) Second, passenger behavior in emergencies may not be as desired. While some passengers may remain calm and follow instructions, others may be too frightened to brace when warned to do so (White et al., 1990), exhibiting ‘‘behavioural inaction’’ (Jorna et al., 1996). Passengers who have not been told of, nor read about, the brace position may not be primed to recognize sudden, shouted commands (White et al., 1993) or may be ‘‘confused or unaware of the proper position’’ (White et al., 1990). Third, passengers may not understand the language in which the commands are given, especially if those commands are shouted in a foreign language shortly before or during a crash (Korea Aviation and Accident Investigation Board, 2005), hence the importance of modeling desired behavior. For example, in 1991, SAS Flight 751 crash-landed in a field near Gottröra, Sweden, shortly after takeoff. Although no brace command was given by the captain, 20 s before impact, the purser shouted the instruction in English. This was then repeated in English and Swedish by cabin crew in the rear of the cabin. There were no fatalities, and provision of the brace instruction was considered to have reduced the number and severity of injuries suffered. However, one of the safety inquiry recommendations was that ‘‘safety information in aircraft operated in international traffic operated by Scandinavian airlines companies is also given in one of the Scandinavian languages’’ (Swedish Board of Accident Investigation, 1993). Implications of this recommendation are that delivery of instructions is not enough – instructions must be readily understood.

Regulatory Aspects Why, then, is there variability in provision of information about the brace position? The answer lies in regulations governing preflight passenger safety briefings and cards.

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51

Airline Companies Currently, almost all individual airlines determine the content/format of passengers’ safety-related information. With respect to the brace position in the passenger safety briefing, this may range from none, to a live or video demonstration. Information in the passenger safety cards is also variable, with pictorials of varying degrees of clarity. However, some carriers have voluntarily included brace position information in their video briefings, recognizing the importance of doing so. These carriers include Air New Zealand, Arab Emirates Airlines, British Airways, Cathay Pacific Airways, Japan Airlines, Qantas Airways, Qatar Airways, Singapore Airlines, Sri Lankan Airlines, and Thai Airways.

Countries The reason for this variation among airlines is related to the almost universal absence of national regulations mandating inclusion in the passenger preflight briefing or safety cards. For example, European Commission regulations stipulate that the aircraft commander ensures that ‘‘all passengers are briefed on the location of emergency exits and the location and use of relevant safety and emergency equipment’’ (European Union, 2008, p. L 10/9). Airline operators are required to ensure passengers receive a safety briefing, parts or all of which may be provided by an audiovisual presentation. Briefings must include demonstrations of the use of safety belts and oxygen equipment. Safety briefing cards are to include ‘‘picture type instructions’’ on emergency equipment and exits. The brace position is not mentioned (European Union, 2008). In 1979, a National Transportation Safety Board report about the brace position and passenger injuries included three recommendations made to the Federal Aviation Authority (FAA). Recommendation 3 was to require ‘‘principal operations inspectors to instruct their assigned air carriers to describe the appropriate emergency brace position on the passenger briefing card and to require that preflight briefings include a reference to the proper brace position’’ (National Transportation Safety Board, 1979). The most recent FAA Advisory Circular states ‘‘oral briefings must be supplemented with briefing cards’’ but information in the cards ‘‘should contain information about protective brace positions to be assumed by passengers. . .’’ (Federal Aviation Administration, 2003). In this Circular, must denotes a ‘‘regulatory requirement,’’ but should refers to ‘‘guidelines that are not regulatory requirements’’ and therefore do not have to be followed. Thus, in the United States, as in many other countries, there is no national (federal) requirement to include the brace position in passenger safety information. Three countries do have such a requirement: Canada, Australia, and the United Kingdom. Transport Canada requires passenger safety cards to include passenger brace positions, as appropriate for each type of seat and restraint

Ó 2015 Hogrefe Publishing

K. I. Yoo & J. M. Davies: Brace for Emergency Landings

system (Transport Canada, 2012b). In Australia, the Civil Aviation Safety Authority requires carriers to ensure passenger safety cards include the brace position for emergency landing or ditching (Australian Civil Aviation Safety Authority, 2004). In the United Kingdom, the Civil Aviation Authority requires passenger safety cards to show ‘‘detailed brace positions for all types of seat orientation and pitch’’ (Civil Aviation Safety Authority, 2011). However, none of these countries require brace position information to be given in the standard passenger safety briefing. In addition, not all airline companies require or teach their cabin crew to shout verbal commands in an unanticipated emergency, should they sense the necessity for passengers to adopt the brace position. This was described in the Air France Flight 358 crash investigation in Toronto (Transportation Safety Board of Canada, 2007).

ICAO Annex 6 to the Convention on International Civil Aviation of the International Civil Aviation Organization (ICAO) clearly stipulates the international standards and recommended practices about informing passengers of the location/use of seat belts, emergency exits, and cabin emergency equipment (ICAO, 2010). What is not described is any mention of the brace position. In effect, if passengers have not adopted the brace position, then their probability of being able to follow the emergency lighting to the emergency exits could be reduced, depending on the type of landing.

Discussion

Ó 2015 Hogrefe Publishing

The first step to increase the probability that passengers will survive an aircraft crash is for ICAO to mandate all states to require that passengers be provided with information about versions of the brace position both in the preflight passenger briefing and on safety cards. ICAO would need to develop and mandate new international standards to achieve compliance from all air carriers. A new provision for the brace position could be established as a paragraph (4.2.12.5) in Annex 6 to the Convention on International Civil Aviation, Part I, International Commercial Air Transport-Aeroplanes. This new provision could state: The operator shall ensure that passengers are given preflight instruction about how to adopt the brace position for an emergency landing or impact of the aircraft. The operator shall also ensure the same information about how to adopt the brace position will also be included in the passengers’ safety cards. These changes would need to be fully orchestrated, with contracting states’ national legislations adopting the same standards. However, each state and airline company could make these changes before being required to do so. In addition, passengers and members of the public could begin to ask for these changes, as well as to ‘‘practice bracing’’ before a ‘‘forced landing seems likely’’ (White et al., 1993). The anticipated results would be to give passengers access on every flight to brace position information, from the preflight passenger briefing and safety card, thus potentially contributing to improving their own safety. Further regulatory changes should include standardizing information about the passenger brace position. Many passengers do not fly on only one type of aircraft, with only one airline company, or in only one country. Putting the onus on passengers to review and understand the many differences in recommended brace positions means that passengers can easily be confused about what to do, especially in an emergency. Of course, standardizing information would require that regulators, airline companies and operators, cabin safety professionals, scientists, doctors, and other researchers all work together to minimize the variability in the information available about the various brace positions. However, policy changes to reduce the international and national variability in the content and format of information could also help, as well as improving communication with and comprehension by passengers. Where possible, they need to understand the importance of the brace position and how to attain the most appropriate position, and they need to be prepared to do so. But passengers need ICAO and national regulators to effect these policy changes to help make commercial aviation safer.

References Airline Safety and Losses. (2015). 2014 annual review (Special Report). London: FG Analytics. Retrieved from http:// www.flightglobal.com/news/articles/airline-safety-amp-lossesannual-review-2014-407835/

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51

APAHF in Practice

Although commercial air travel is growing safer, aircraft accidents still occur. Despite the need for more and better evidence, the brace position has been shown to reduce injuries and save lives. With greater adoption of the brace position, more passengers would be more likely to survive, with fewer, less severe injuries. But passengers lack access to brace position information in most preflight passenger briefings. This information is not provided because it is not required – by most airlines, by their national states, and ultimately by ICAO. Why ICAO does not include the brace position in Annex 6 is not clear. However, to date, ICAO and the aviation safety community have been necessarily occupied with other aspects of aviation safety, including aircraft design and maintenance, flight operations, and air traffic control. Current cabin crew training does not generally include ICAO Annex provisions with respect to cabin safety. There is also an absence of publications on this topic. Furthermore, the lack of standardized information and limited application of results of some studies – for example, a lack of results describing the ‘‘feet position’’ for passengers other than the ‘‘average male’’ – is obviously a potential impediment to regulatory reform.

APAHF in Practice

K. I. Yoo & J. M. Davies: Brace for Emergency Landings

Australian Civil Aviation Safety Authority. (2004). Passenger safety information: Guidelines on content and standard of safety information to be provided to passengers by aircraft operators (August 2004, CAAP 253-2(0)). Canberra, Australia: Civil Aviation Advisory Publication. Retrieved from http://www.casa.gov.au/wcmswr/_assets/main/download/ caaps/ops/253_2.pdf Barthelmess, S. (1988). Positions brace passengers for impact to reduce injuries and fatalities. Cabin Crew Safety, 23, 1–2. Brownson, P., Wallace, W. A., & Anton, D. J. (1998). A modified crash brace position for aircraft passengers. Aviation Space Environmental Medicine, 69, 975a978. Cabin Safety Team. Operator Safety Practices Working Group. (2001). Cabin safety compendium: A companion to the operatoror Safety Practices Worki. Alexandria, VA: Global Aviation Information Network (GAIN) Program. Flight Safety Foundation. Retrieved from http://flightsafety.org/ files/cabin_safety_compendium.pdf Caird, J. K., Wheat, B., McIntosh, K. R., & Dewar, R. E. (1997). The comprehensibility of airline safety card pictorials. In Proceedings of the 41st Human Factors and Ergonomics Society Annual Meeting (pp. 801–805). Santa Monica, CA: HFES. Retrieved from http://pro.sagepub.com/content/41/2/ 801 Chandler, R. F. (1988). Brace for impact positions. Protection and Survivability Laboratory. Civil Aeromedical Institute. Washington DC: Federal Aviation Authority. Retrieved from http://www.unitedafa.org/safety/training/docs/brace.pdf Chang, Y.-C. (2013). Exploring cabin safety services needs of elderly air passengers. Current Issues in Tourism, 16, 407–412. Civil Aviation Safety Authority. (2011). Requirements and guidance material for operators. Norwich, UK: Author. Retrieved from https://www.caa.co.uk/docs/33/CAP%20789.pdf Cherry, R., Warren, K., & Chan, A. (2000). A benefit analysis for aircraft 16-g dynamic seats. April, 2000 (Report No. DOT/FAA/AR-00/13). Washington, DC: US Department of Transportation, Federal Aviation Authority. Retrieved from http://www.fire.tc.faa.gov/pdf/00-13.pdf Corbett, C. L., McLean, G. A., & Cosper, D. K. (2008). Effective presentation media for passenger safety I: Comprehension of briefing card pictorials and pictograms (Report No: DOT/FAA/AM-08/20). Washington, DC: Office of Aerospace Medicine/Federal Aviation Authority. Retrieved from www.dtic.mil/cgi-bin/GetTRDoc?AD= ADA488828 Cullen, S. A. (2004). Injuries in fatal aircraft accidents. NATO RTO-EN-HFM-113, 3. Retrieved from http://ftp.rta.nato.int/ public/PubFullText/RTO/EN/RTO-EN-HFM-113/EN-HFM113-03.pdf Dulcavsky, S. A., Geller, E. R., & Iorio, D. A. (1993). Analysis of injuries following the crash of AVIANCA Flight 52. The Journal of Trauma, 34, 282–284. European Union. (2008). Commission Regulation (EC) No 8/2008 of 11 December 2007: Common technical requirements and administrative procedures applicable to commercial transportation by aircraft. OPS 1: Commercial Air Transportation (Aeroplanes). Official Journal of the European Union, 12.1. Retrieved from http://eur-lex.europa.eu/ legal-content/EN/ALL/?uri=CELEX:32008R0008 Federal Aviation Administration. (2003). Advisory Circular 12124C: Passenger safety information briefing and briefing cards. Washington, DC: Author. Retrieved from http:// www.faa.gov/regulations_policies/advisory_circulars/index. cfm/go/document.information/documentID/22488 Federal Aviation Administration. (2010). Improved Seat Rule 14 CFR 121.311(j). In Information for operators, InFO 10021. Washington, DC: Author. Retrieved from http://www.faa. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51

gov/other_visit/aviation_industry/airline_operators/airline_ safety/info/all_infos/media/2010/Info10021.pdf Federal Aviation Administration Fire Safety. (2014, January 15). Fire research. Washington, DC: Author. Retrieved from https://www.fire.tc.faa.gov/research/backgrnd.stm Garner, J. D., & Blethrow, J. G. (1970). December 1970 evacuation tests from an SST mock-up. Washington, DC: Office of Aviation Medicine, Federal Aviation Authority. Retrieved from http://libraryonline.erau.edu/online-full-text/ faa-aviation-medicine-reports/AM70-19.pdf Grierson, A. E., & Jones, L. E. (2001). Recommendations for injury prevention in transport aviation accidents (Paper No. 2001-01-2658). Warrendale, PA: Society of Automotive Engineers. Retrieved from http://www.cs.odu.edu/ mln/ ltrs-pdfs/NASA-2001-aasce-aeg.pdf Hutchinson, B. (2010, January 10). Emergency landing by United Airlines Flight 634 shuts down Newark Liberty International Airport. New York Daily News. Retrieved from http://www.nydailynews.com/new-york/emergency-landingunited-airlines-flight-634-shuts-newark-liberty-internationalairport-article-1.171195#ixzz31XigmIR7 International Civil Aviation Organization. (2010). Operation of aircraft. Annex 6 to the Convention on International Civil Aviation, Part I International Commercial Air Transport: Aeroplanes (9th ed). Montreal, Canada: Author. International Civil Aviation Organization. (2012). 2012 safety report. Montreal, Canada: Author. Retrieved from http:// www.icao.int/safety/documents/icao_sgas_2012_final.pdf International Civil Aviation Organization. (2014). State of global aviation safety. Montreal, Canada: Author. Retrieved from http://www.icao.int/safety/documents/icao_2014%20safety% 20report_final_02042014_web.pdf Jamaica Civil Aviation Authority. (2014). Aviation accident investigation report: Runway overrun on landing. American Airlines Flight AA331 Boeing 737-823 United States Registration N977AN. Norman Manley International Airport, Kingston, Jamaica (MKJP) (Report No. JA-2009-09). Kingston, Jamaica: JCAA. Retrieved from http://www.jcaa. gov.jm/NEWS_UPDATES/FINAL_REPORT_AA331/AA331 %20FINAL%20REPORT%2002% 20May%202014.pdf Johnson, D. (1998). Studies reveal passenger misconceptions about brace commands and brace positions. Cabin Crew Safety, 33, 1–6. Johnson, D. A. (1979). An investigation of factors affecting aircraft passenger attention to safety information presentations (Report No. IRC-79-1). Washington, DC: US Department of Transportation, Federal Aviation Authority. Retrieved from http://www.dtic.mil/dtic/tr/fulltext/u2/ a082358.pdf Jorna, P., Amalberti, R., McDonald, N., Piers, M., Taylor, F., Wegmann, H., & Muir, H. (1996). Increasing the survival rate in aircraft accidents. Impact protection, fire survivability and evacuation. Brussels, Belgium: European Transport Safety Council. Retrieved from http://archive.etsc.eu/ documents/survival.pdf Koenig, R. L (1995). U.K. studies find that ‘‘Legs-back’’ brace position is optimal for forward facing passengers. Cabin Crew Safety, 30, 1–4. Korea Aviation and Railway Accident Investigation Board. (2005). Aircraft accident report. Controlled flight into terrain Air China International Flight 129, B767-200ER, B2552. Mountain Dotdae, Gimhae. April 15, 2002. Seoul, Korea: Author. Retrieved from http://www.skybrary.aero/ bookshelf/books/549.pdf Molesworth, B. R. C., Burgess, M., Gunnell, B., & Duong, K. (2014). Examining the benefits of active noise control and supplementary pictorial information in improving the recall of information in a noisy environment. Aviation Psychology and Applied Human Factors, 4, 50–57. Ó 2015 Hogrefe Publishing

K. I. Yoo & J. M. Davies: Brace for Emergency Landings

Ó 2015 Hogrefe Publishing

White, B. D., Firth, J. L., Rowles, J. M., & N.L.D.B. Study Group (1993). The effects of brace position on injuries sustained in the M1 Boeing 737/400 Disaster, January 1989. Aviation Space and Environmental Medicine, 64, 103–109. White, B. D., Rowles, J. M., Mumford, C. J., & Firth, J. L. (1990). A clinical survey of head injuries sustained in the M1 Boeing 737 disaster: Recommendations to improve aircrash survival. British Journal of Neurosurgery, 4, 503–510.

Received August 19, 2014 Revision received January 21, 2015 Accepted for publication January 21, 2015 Published online April 10, 2015

Jan Davies (MSc, MD, FRCPC) is a professor of anesthesia, Cumming School of Medicine, and an adjunct professor of psychology, Faculty of Arts, at the University of Calgary. Since 1983 she has worked and undertaken research in system safety in health care and industry. Jan Davies is an associate member of the International Society of Air Safety Investigators (ISASI). She continues to refine human factors-based and system safetyoriented models of prospective and retrospective investigation, as well as performance review methodology. Her current research includes human factors analysis of medical devices and procedures, as well as the passenger brace position. Jenny Yoo (BA, LLM, DSc) is an Assistant Professor of Aviation Safety, Department of Aviation Service, Hotel and Tourism College, Wonkwang Health Science University, an appointed advisor to the Korea Aviation & Railway Accident Investigation Board (KARAIB), an appointed aviation safety analyst of the Korea Transportation Safety Authority (Cabin Safety), and the director general of the Korean Society of Air Safety Investigators (KSASI). Jenny Yoo began her aviation career as a cabin crew member, holding various positions, including cabin crew team executive member. In 2001 she gained a Certificate in Aircraft Accident Investigation & Prevention from the Southern California Safety Institute (SCSI). At Wonkwang Health Science University, her work focuses on cabin crew safety, including the passenger brace position.

Correspondence Address Jan M. Davies Department of Anesthesia Cumming School of Medicine University of Calgary Room C-229 Foothills Medical Centre Calgary, Alberta T2N 2T9 Canada Tel. +1 403 944-4707 E-mail jdavies@ucalgary.ca Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):45–51

APAHF in Practice

Muir, H. C., & Thomas, L. J. (2004). Passenger education: Past and future. Paper presented at the fourth Triennial International Aircraft Fire and Cabin Safety Research Conference, Lisbon, Portugal. Retrieved from http://www.fire.tc.faa. gov/2004Conference/files/op/H.Muir&L.Thomas_Passenger_ education_past&future.pdf National Transportation Safety Board. (1979, October 4). Safety recommendations A-79-76 through -78 [Letter to Honorable Lanhorne M. Bond, FAA, from James B. King, NTSB]. Washington, DC: Author. Retrieved from http://www.ntsb. gov/doclib/recletters/1979/A79_76_78.pdf National Transportation Safety Board. (2011). January 10, 2010 United Airlines DCA10IA021 [Full online narrative]. Washington, DC: Author. Retrieved from http://www.ntsb.gov/ aviationquery/brief2.aspx?ev_id=20100112X10853&ntsbno= DCA10IA021&akey=1 Schunemann, H. J., & Guyatt, G. (2014). Clinical epidemiology and evidence-based health care. In W. Ahrens & I. Pigeot (Eds.), Handbook of epidemiology (2nd ed, pp. 1815–1873). New York: Springer Science+Business Media. Retrieved from http://link.springer.com/referenceworkentry/10.1007% 2F978-0-387-09834-0_30#page-1 Sperber, M., Schäbe, H., Masling, D., Toth, D., Küting, J., Demary, M., & Wodli, G. (2010). Carriage by air of special categories of passengers. Cologne, Germany: European Aviation Safety Agency. Retrieved from http://easa. europa.eu/system/files/dfu/Review%20Group%20for%20 RMT.0269%20%26%20RMT.0270%20%28MDM.072% 28a%29%26%28b%29%29%20%E2%80%94%20Issue% 202.pdf Sulzberger, A. G., & Schweber, N. (2010, January 10). Jet makes emergency landing at Newark Airport. New York Daily News, A12. Retrieved from http://www.nytimes.com/ 2010/01/11/nyregion/11plane.html Swearingen, J. J. (1966). Evaluation of head and face injury potential of current airline seats during crash decelerations. Washington, DC: Office of Aviation Medicine, Federal Aviation Authority. Retrieved from http://www.dtic.mil/dtic/ tr/fulltext/u2/653869.pdf Swedish Board of Accident Investigation. (1993). Aircraft traffic accident on 27 December 1991 at Gottröra, AB County. Case L-124/9 (Report C 1993:57). Stockholm, Sweden: SBAI. Retrieved from http://www.havkom.se/virtupload/ content/101/C1993_57e.pdf Transportation Safety Board of Canada. (2007). Aviation investigation report, runway overrun and fire, Air France Airbus A340-313 (F-GLZQ), Toronto/Lester B. Pearson International Airport (Report No. A05H0002). Hull, Quebec: Author. Retrieved from http://www.tsb.gc.ca/eng/ rapports-reports/aviation/2005/a05h0002/a05h0002.pdf Transport Canada. (1999). Brace positions for impact (Commercial and Business Aviation Advisory Circular No. 0155. Advisory Circulars). Ottawa, ON: Author. Retrieved from https://www.tc.gc.ca/eng/civilaviation/standards/commercecirculars-ac0155-1633.htm Transport Canada. (2012a). Brace positions for impact. Ottawa, ON: Author. Retrieved from https://www.tc.gc.ca/eng/ civilaviation/standards/commerce-cabinsafety-bracepositions1250.htm Transport Canada. (2012b). Standard 725.44: Safety features card and supplemental briefing card. Part VII: Commercial Air Services, Aviation Regulations (CARS). Ottawa, ON: Author. Retrieved from http://www.tc.gc.ca/eng/civilaviation/ regserv/cars/part7-standards-725-2173.htm Tucker, W. T. (1989). Learning from past accidents. In Aircraft Fire Safety, AGARD Conference Proceedings No. 467 (pp. 41–4-10). Neuilly, France: AGARD. Retrieved from http:// ftp.rta.nato.int/public/PubFullText/AGARD/CP/AGARDCP-467/AGARD-CP-467.pdf

Toward Evidence-Based Decision Making in Aviation The Case of Mixed-Fleet Flying Timothy J. Mavin,1 Wolff-Michael Roth,1,2 Kassandra Soo,1 and Ian Munro3 1

Griffith Institute for Educational Research, Griffith University, Mt. Gravatt, QLD, Australia, 2 Faculty of Education, University of Victoria, BC, Canada, 3Mount Cook Airlines, Christchurch Airport, New Zealand Abstract. Academic institutions and airlines have always worked together to develop and conduct research studies. However, most often the expertise or areas of interest of the academics have driven these studies. In this paper, we illustrate the results of an industry–university collaboration that generated data that the airline could use to engage in evidence-based decision making. The example given regards issues emerging from mixed-fleet flying, generally related to reverse transition from glass to analogue cockpits. Keywords: mixed-fleet flying, industry–university collaboration, field-relevant research methods, collaborative research design

APAHF in Practice

You know, after doing my training on the 600, I now have my doubts about mixed-fleet flying. Same airframe, same engines, the lot, but the flight decks are different, maybe too different? (Training manager, ATR 72 operations) In this vignette, the training manager of an airline talks to a university-based researcher about the new aircraft his company has purchased. The company had been operating the ATR 72-500 for a decade and recently purchased the newer ATR 72-600, a ‘‘glass cockpit’’ variant of the former. Initial plans were for all pilots to fly both aircraft types concurrently, a possibility that the aircraft’s developer had actively promoted. However, after receiving his training on the new aircraft, the training manager had doubts about the regulations allowing pilots to fly both aircraft concurrently. For the airline, there was a compelling need to obtain evidence to assist in its decision-making process about moving to mixed-fleet flying (MFF). In this specific case of flying the ATR 72-500 and ATR 72-600 concurrently, there was no (independent) research evidence available. Although research evidence for many dimensions of the aviation industry is widely available, there are two possible problems facing airlines today. First, airlines need to find how something formulated at a general level (i.e., studies concerning MFF) is applicable in a concrete situation. This issue is known from other disciplines, like teaching, where teachers generally do not see how research findings are relevant in their settings (e.g., Roth, 1998). For example, using top-down thinking, centralized governments implement policies that in some cases have classroom Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61 DOI: 10.1027/2192-0923/a000069

teachers applying practices that appear to generate few improvements in learning. Likewise in aviation, current trends may appear difficult to understand and implement. For example, currently there is a ‘‘safety improvement initiative’’ by the International Air Transport Association (IATA) toward evidence-based training (EBT; IATA, 2013, p. 5). It is based on the idea that, with increasing amounts of data now available to the aviation community, training improvements can be made. However, one source of data within EBT is the line-orientated safety audit (LOSA), which has been criticized for its approach to obtaining evidence (e.g., Dekker, 2003). Further, another underlying requirement for valid data is interrater reliability training for pilots. However, once again, the viability of this has recently been questioned (Mavin, Roth, & Dekker, 2013; Roth & Mavin, 2013). Like the teaching community, aviation must continually address whether both research and new approaches to practice are worthy of possible implementation. The second issue an airline may face is that current problems have not been researched to a satisfactory level. This issue is related to establishing evidence rather then using it. For some issues, current evidence is lacking, weak, or questionable. However, how does an airline, with skills primary in airline operations – not research – develop evidence to assist with making informed decisions? Furthermore, how does an airline reach out to those who can help, when there is currently criticism of the lack of unified work between regulators, operators, and scientists? People bring value to a company. In recent times, this simple fact has led to successful companies working in Ó 2015 Hogrefe Publishing

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

partnership collaborations with other organizations. Collaboration here is viewed not simply as a contractual consulting affiliation with others outside the company, but as ‘‘a cooperative, inter-organizational relationship that relies on neither market nor hierarchical mechanisms of control but is instead negotiated in an ongoing communicative process’’ (Lawrence, Philips, & Hardy, 1999, p. 481). The aim of interorganizational collaborations is to (a) utilize existing resources or regions of strengths more effectively or efficiently, (b) learn from other organizations, and (c) develop expertise within one’s own organization with the assistance of the learning partners. The collaborative relationships can be arranged through either loose verbal agreements or elaborate contractual agreements, but they must be based on an element of trust (Roth, 1998). In this paper, we discuss an innovative interorganizational collaboration between an airline and a university that emphasizes the practical application of human factors directly into operational issues that arise in the field.

1-Problem Itentified by Airline

5-Jormal writing

4-Data collection & analysis

2-Airline & University research meeting

3-University refine methods & obtain Ethical Approval

Figure 1. Evidence based decision-making flow chart.

A Move Toward Evidence-Based Decision Making

Ó 2015 Hogrefe Publishing

The evidence-based decision-making cycle adopted was a continuous process, with multiple projects operating at differing stages (Figure 1). Stage 1 had the airline identifying issues impacting current business practice, or where possible improvements to practice might occur. The second stage found the airline and university team in initial discussions concerning the problems. Accordingly, the second stage teased out pilot versus airline issues, focusing on clearly identifying questions to be answered. Whereas this may appear formal, almost all discussions took place over months in semiformal meetings, with a majority occurring while other projects were actually in the data collection or journal-writing phase. The emphasis of Stage 2 was on problem statement and formulation, which must emanate from the airline. In Stage 3, the formal project was initiated and the research team, involving university-based researchers, the training manager, and, where applicable, other parties such as pilot unions, developed a robust research design. The design issues are central, for the research had to meet the dual goal of rigor in the academic context and high ecological validity to be applicable directly to the company context. Ethical clearance was also an integral aspect of the design, involving the airline, pilot/cabin crew unions, the regulator, and university-based ethics boards. It was important that pilots have the option of not participating in any research and that their nonparticipation would have no effect on their career within the company. The fourth stage of a project consisted of data collection and analysis. In Stage 5, the research team prepares manuscripts for peer-reviewed journals. As the airline specializes in pilot training and not research, the peer review process is a form of quality-assurance. The findings sections from the journal articles are then integrated into reports for the airline. The following case concerning the issue of MFF is offered as an example of this process. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

APAHF in Practice

Over the past 3 years, the airline, the university, and local regulator have been working together in an interorganizational collaboration. So far, this collaboration has produced studies into pilot assessment, classroom-based instructor training, flight examiner training, debriefing poor-performing pilots, and issues associated with MFF. For the airline, there was now a true business case for having a university involved in practice generally and, specifically, in the production of evidence on which to base decisions. Because of the airline’s need for evidence drove the research agenda – rather than the university asking for airline participation – evidence produced could be used directly to improve the airline’s operational decision making. For the universitybased researchers, opportunities arose for (a) being involved in research that has both applied and empirical foci, (b) broadening the university’s research focus, and (c) enabling researchers to publish timely research applicable to the aviation industry more broadly. A fundamental requirement for the collaboration was that all research must have a high degree of ecological validity, enabling the dual focus on empirical research that was directly applicable to the participating airline. Second, the airline requested that findings be publishable in scholarly journals. This was an unusual request, with previous experience demonstrating that most companies feel uneasy about publicizing company issues. However, the airline argued that the advantages of researchers maintaining high scholarly standards via academic peer review far outweighed the benefits of keeping data within the organization. The aim of the collaboration was not to replace professional judgment, but to augment it. This aim was an orientation toward an evidence-based approach that aimed at making best decisions supported by concretely relevant research.

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

A Company Problem: Moving to Mixed-Fleet Flying In many regions of the world, airlines find it necessary to have a mixture of aircraft types, such as the Boeing 777, Airbus 320, and ATR 72, to service the needs of customers. When a particular aircraft type is sufficiently similar in terms of systems and flying characteristics (termed ‘‘commonality’’) to another aircraft type, aircraft manufacturers apply to aviation authorities and regulators for a common aircraft type certificate or type rating. This enables pilots – with specific training – to simultaneously fly both versions of the aircraft. Such flying of multiple versions of the same aircraft by the same pilots is referred to as mixed-fleet flying (MFF). Aircraft where this occurs include the Boeing 777-200/300, Airbus A319/320/321, and regional aircraft such as the Dash-8-100/200/300 series. Such configurations improve operational efficiencies – for example, by reducing overall number of pilots required and increasing rostering efficiency.

ants). For the airline, having separate groups of pilots flying each variant made little commercial sense. Commercial forecasting suggested an extra 10% pilot workforce would be required to operate separate fleets, with a consequent increase in airline operating costs. Having pilots revert to the older variant of an aircraft is called reverse transition and an example is transferring from a ‘‘glass’’ environment to a traditional analogue cockpit. In our participant airline, while there were only minor issues with forward transition, there were certainly concerns from the pilots that the differences that existed between the flight decks of the two aircraft would make MFF problematic. At this point, the company was confronted with the issue of how to assure safety within its operation when pilots were asked to move back and forth between the two aircraft types. Even though the manufacturer and regulators around the world were overseeing and regulating MFF, there were no independent studies available to assist the airline in its decision making. This led to an initial review of potential problems.

Stage 2: Airline/University Discussions Stage 1: Identifying the Problem

APAHF in Practice

The practice of transitioning pilots from an older variant of an aircraft to a newer variant is referred to as forward transition. During the introduction of the ATR 72-600 for the airline, forward transition from the ATR 72-500 to the newer aircraft appeared to present only minor issues to the pilot group, and the transition program was deemed appropriate for pilots moving to fly the ATR 72-600 (Figure 2 shows photographs of the flight decks of the vari-

During initial planning (including phone calls, e-mail, Skype calls, and face-to-face discussions) between the airline and university-based researchers – and given the lack of current studies into MFF relating to the ATR 72 – a two-prong design emerged. The first study was a survey deployed with pilots from the airline who had flown both aircraft types. Its aim was to obtain a broad perspective from these pilots and seek (a) their experiences during transitions from the 500 to the 600 and (b) their views on and concerns regarding MFF. A second study evolved exploring the issue of MFF in greater depth. While collecting data for another study, the training manager and university-based researchers worked on a whiteboard to develop a research design that was feasible within the airline context and would meet the rigor expected of scholarly research (Figure 3). This was to be a 2-day simulator study with company pilots from the ATR 72-600 flying the ATR 72500 simulators, using (a) the think-aloud protocol typical for identifying knowledge and problem-solving strategies of experts and novices alike and (b) stimulated recall sessions often used in technology environments subsequent to the simulator exercises. Whereas the survey would provide examples of forward transition issues, the purpose of the second study was to obtain concrete examples of the type of difficulties pilots might face when transitioning back to the ATR 72-500.

Stage 3: University Refines Method and Obtains Ethical Clearances Figure 2. Photo of ATR72/500 (top) depicting older style instruments. ATR72/600 (bottom) depicting newer flight deck setup including primary flight display (PFD), multi function display (MFD) and engine and warning display (EWD). Each pilot (captain – left and first officer – right) has his or her own PFD and MFD. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

Survey The first stage of the study was planned as a survey of pilots’ views on the transition to the ATR 72-600 and on MFF. The survey questions were developed on the basis of previous studies of the glass cockpit (e.g., Wiener, Chute, Ó 2015 Hogrefe Publishing

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

& Moses, 1999). Part 1 used a 5-point Likert scale (1 = strongly disagree, to 5 = strongly agree) and was separated into eight categories: overall perception of transition training, avionic suite, instruments, flight management system (FMS), standard operating procedures (SOPs), electronic checklist (ECL), workload, and crew resource management. Part 2 included open-ended questions relating to MFF, specifically investigating pilots’ opinions on transition training, glass cockpits, and MFF in general. The other prime focus of the survey study was to identify potential problems in relation to MFF.

Simulator Study A 2-day simulator trial using think-aloud protocols in a full flight simulator was initially planned (Figure 3). The benefits of think-aloud protocols have been well established in studies of expertise. However, it was anticipated that the yield of data from traditional think-aloud protocols might diminish during high workload situations; therefore a retrospective, video-stimulated debriefing session after each simulation was also conducted to allow pilots to articulate as much as they could about relevant situations. This retrospective debriefing was facilitated using the simulator debriefing tool – an integrated built-in fixed camera located in the simulator that records and transfers visual, audio, and flight instrument records of various aspects of the flight into the debriefing room (Figure 4). During the debriefing, the entire simulation would be played back to the pilots, who were instructed to pause the video at any time to talk about difficulties they had experienced. All debriefing sessions were video recorded.

Figure 3. The training manager and university-based researchers brainstormed on a whiteboard to design the research to be conducted in the simulator associated with debriefing sessions. Ethical Clearances The Griffith University Ethics Committee approved the study. Participants were informed about the purpose of the study and were assured that no job repercussions will occur if they chose to not participate or to opt out of the study, whereby participants’ related data would be deleted. This information was provided in the form of an

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

APAHF in Practice

Figure 4. Pilots (inner two participants) and research team (outer two participants) discuss the simulator exercises with the aid of the debriefing tool in the background. (All persons in the photograph have given their written permission to be shown.)

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

Table 1. Survey responses for Part 1 from current pilots flying the ATR 72-600 that were deemed significant Level of significance Agreement More experience on the ATR 72-600 was associated with strong agreement with the statement ( p < .01). More experience on the ATR 72-600 was associated with a moderate agreement with these statements ( p < .05).

Disagreement Pilots more experienced on the ATR 72-600 are more likely to strongly disagree with these statement ( p < .01). Pilots more experienced on the ATR 72-500 are more likely to disagree with these statements ( p < .05).

Response I found the transition from 72-500 to 72-600 complex/difficult. The 72-600 SOPs are very different from the 72-500 SOPs. Learning 72-600 SOPs during transition line training was very challenging. Compared with before flying the 72-600, I find that I have to increase verbal communication with crewmembers to maintain situational awareness of self and others.

I found the transitional line training very helpful, and it could not have been done any better. My flight performances have improved since flying the 72-600. The integrated functions of FMS is a great idea. The ECL is more helpful than the paper checklist during nonnormal operations. The glass cockpit has reduced my workload during normal operations. I notice abnormalities faster now, than I did back in the 72-500 cockpit. The transition line training prepared me to fly confidently during my first 72-600 flight.

Notes. ECL = electronic checklists; FMS = flight management system; SOPs = standard operating procedures.

information sheet (survey study) and an information and consent form (simulator study).

Stage 4: Data Collection and Analysis

situational awareness. Whereas these issues may appear worrying, they are not uncommon for pilots initially undergoing forward transition. Studies in the 1990s showed that transitioning to a glass cockpit is initially difficult, with issues primarily occurring in breakdown of communication and coordination with flight deck automation.

APAHF in Practice

Survey Study Of the 29 pilots with experience on both aircraft types, 19 agreed to participate in the survey. From the remaining 10 pilots, one pilot had not met the participation criteria. We suspect that the remaining pilots were busy during the period of study and were therefore unable to complete the survey within the set time frame. Table 1 presents issues that were found to be significant after analyzing Part 1 of the survey. The issues identified in the survey related to (a) presentation and layout of flight instruments, (b) transitioning between two different flight management computers (FMCs), (c) aircraft-systems instruments and their functionality, and (d) differences in SOPs. It was determined that the simulator study should investigate these issues further. Part 2 of the survey – open-ended questions – showed that there were concerns among the more experienced ATR 72500 pilots transitioning onto the glass cockpit version. These pilots found the transition more complex or difficult than the less experienced pilots. Issues related to a belief that their performance had decreased, difficulties learning new SOPs, the FMS, ECL, and an overall reduction in the speed at which they could notice abnormalities. Other issues arose in relation to the glass cockpit not reducing workload. An interesting finding was the increased amount of communication required between crewmembers to assist in maintaining Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

Simulator Study: Think-Aloud Protocol and Stimulated Recall Three experienced company pilots participated voluntarily in the simulator study: a flight examiner (FE) to operate and run the simulator and a captain and first officer (FO) to fly the simulator. Based on their flight log book, the captain had approximately 7,000 hours on the ATR 72-500 prior to his transition to the ATR 72-600 (13,000 total flight hr), and the FO had approximately 3,760 hours of experience (7,700 total flight hr). Neither pilot had flown the ATR 72500 since their transition to the ATR 72-600, 6 months earlier. The FO had performed a 2-hr preparation (i.e., reviewing ATR 72-500 manuals) prior to this study, whereas the captain had not engaged in any special preparation. Two 4-hr simulator sessions were rostered on consecutive days in a CAE ATR 72-500 full flight simulator with outside view. The motion feature of the simulator was turned off to better accommodate video recordings. Four simulations were designed to investigate issues arising from reverse transition, and were conducted in relation to operations and flight schedule of the airline: one normal day-light flight (Day 1), one normal night flight with a simple nonnormal during the cruise (Day 2), unprepared night rejected Ó 2015 Hogrefe Publishing

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

take-off (Day 2), and a pre-briefed night engine failure after take-off (Day 2). Initial planning was for two sessions on Day 1 (Figure 3). However, the reflective session was extended due to the large amount of information recalled by pilots during the debriefing. To ensure flight performance deviations during simulation were due to cockpit differences rather than to terrain or route unfamiliarity, the pilots selected each simulated flight route. Scenario 1: Normal Flight (Day) The two pilots flew one leg from Auckland to New Plymouth lasting approximately 70 minutes. The current weather was used for the simulation – day operation with thick cloud base starting from 3,000 ft to 11,000 ft. Destination weather cloud base was reduced to 1,300 ft requiring the pilots to conduct an instrument approach. The captain was the pilot flying (PF; pilot controlling the aircraft flight path and making decisions – though captain always had ultimate authority when required) for the flight to New Plymouth, and the FO was the pilot monitoring (PM; monitoring flight path, reading checklists, and actioning aircraft systems when required by SOPs or at direction of PF). Scenario 2: Normal Flight (Night) With Minor Malfunction (Flight Director Bars Failure)

Scenario 3: Rejected Take-Off (Night) Due to Engine Failure During Scenarios 1 and 2, both pilots had discussed some difficulty in reading the airspeed indicator at low speeds. This was immediately addressed by modifying the research design by introducing an unprepared rejected take-off. This maneuver required the pilots to recognize a malfunction and to determine if the speed was appropriate for either conducting a rejected take-off or continuing the take-off at speeds close to V1 (latest speed at which the decision to reject a take-off can be made). The simulator instructor and FE assessed the success or failure of emergency detection. Scenario 4: Engine Failure After Take-Off (Night) This scenario began immediately following the rejected take-off scenario. The instructor restarted the flight scenario and introduced an engine failure after take-off. Ó 2015 Hogrefe Publishing

Data Collection and Analysis All video-recorded debriefing sessions and the simulation session were transcribed word for word. The videos were watched in their entirety. Transcripts were analyzed following a data reduction method that involves identifying all issues arising for the pilots – verbalized in the simulator and during the debriefing session – which constituted the complete set of basic level categories. Comments were placed in a table, and a narrative written on the issue. These issues where then categorized as relating to a specific area, such as flight instruments or automation. Results From the Simulator Study The aim of the simulator study was to investigate potential problems with reverse transition from the newer glass version ATR 72-600 to the ATR 72-500. Along with the survey, this would give the company independent evidence allowing insight into potential problems of conducting MFF. Detailed analysis of the videos and transcripts revealed six categories of potential problems pertaining to reverse transition, including (a) flight instruments, (b) presentation, position, and functionality of secondary instruments, switches, and dials, (c) automation, (d) FMC, (e) ECL, and (f) general (Table 2). To assist readers in understanding how these categories emerged, the category of presentation, position, and functionality of secondary instruments, switches, and dials will be discussed. Over the 2 days, both pilots identified issues associated with the presentation, position, and functionality of secondary instruments, switches, and dials. For instance, the ATR72-600 had a newer flight management system (FMS) that encompassed: the primary flight display (PFD), which displays primary flight instruments; the multi function display (MFD) that can present navigation and engine start information for instance; and engine and warning display (EWD), an advanced centralised system relaying important information to the pilot for the current phase of flight. These advanced systems differ from the older ATR72-500 instruments that are fixed (see Figure 2 and Figure 5). It also assists the pilots in identifying malfunctions and identifying the correct procedures. However, the difference between the ATR 72-500 and the ATR 72-600 in functionality and positioning of instruments created problems. For example, the captain noted that, compared with the newer aircraft, the ATR 72-500 had a significantly greater number of engine instruments: ‘‘I was looking across, going ‘oh there’s a whole stack of them’’’ (Captain, Day 1). Additionally, due to the positioning of the engine instruments, the pilots had difficulty identifying correct engine instrument during engine start – often referred to as start scan. This issue continued into Day 2. By the end of the study, the problems with the engine instruments were improving, though slight issues still remained. At the end of Day 2, the FO and captain discuss the engine start procedures they had conducted: FO: After four engine starts, two yesterday and two today, on the fourth one was the first time my eyes went directly to where they needed to go. Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

APAHF in Practice

The flight was from Christchurch to Wellington. Like the previous flight, it was conducted in real time. The flight was conducted in simulated night conditions with no failures initiated by the instructor until the cruise phase, where a minor failure was introduced. The failure was a flight instrument failure of the flight director (FD) systems. The purpose of inducing the FD bar failure was to increase the workload of the pilots during flight and, specifically, during the missed approach – which was planned – at the destination airport.

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

Table 2. Typical Issues faced by pilots initially transitioning back from the ATR 72/600 to the ATR 72/500 Area Flight instruments

Type of issue General comments

Size of the artificial horizon Air speed indicator (ASI)

Instrument scan Skid-ball appearance Navigation display Minimum descent altitude (MDA) Terrain button Ground speed indication Presentation, position and functionality of secondary instruments, switches and dials

EFIS switches

Brightness controls Clock Third QNH setting Instrument display

APAHF in Practice

Data card Position of trim indicator Engine instruments

Communication & navigation switches Starter cut-out Flap Automation

Autopilot engaged light Speed modes Vertical speed control Speed confusion

Specific example First take-off required high concentration due to positioning of primary flight instruments. e.g. ASI, VSI. A lack of information overall is observed. The artificial horizon appeared small. ASI appearance is small. Difficulty reading ASI at low speed call of 70 knots with ‘‘little needle movement until just before 70 knots.’’ No other issue found apart from slow movement around 70 knots. Instrument scan is different. Crew looking in ‘‘other places subconsciously’’ for instruments. Skid-ball is more apparent on 500. The 600 skid-ball is more ‘‘incorporated in the overall picture.’’ Select of preferred navigation presentation lead to orientation issues. An expectancy that MDA appears on each pilots screen after one pilot selects MDA. When a pilot selects MDA on the 600 it appears on both pilots’ instruments. Difficulty finding switch for terrain to be presented on navigation display. Initial difficulty finding ground speed indication on navigation display. Pilots have difficulty finding switches. Pilots have difficulty remembering functionality of some switches. Captain having some issues with the brightness control for the EFIS. Positioning of clock for FO was found to create a minor issue of confusion (Captain: no issue identified). Difficulty finding dial to change QNH on standby altimeter. Captain comments on lack of color and bland appearance of analogue instruments. There was initial difficulty setting up the flight data card. On more then one occasion, captain mentioned difficulty finding trim indicator. Initial surprise at number of instrument. . . ‘‘oh, there’s a whole stack of them.’’ During engine start, pilots have initial difficulty finding instruments to follow. Captain feels more comfortable with radio and navigation switching on the 500. Lack of an aural ‘‘click’’ when engine-start switch disengages makes FO miss starter cut-out. Difficulty finding flap indicator due to its being in different position. Difficulty in initially finding autopilot engaged light on artificial horizon. During take-off, speed/bank (automatic on the 600) confused. Vertical speed control on autopilot difficult to find in dark. During practice engine failure after take-off, pilots use 600 procedures. However, at clean-up altitude, with autopilot engaged, aircraft begins a descent rather then maintaining level-off altitude and accelerating. (Continued on next page)

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

Ó 2015 Hogrefe Publishing

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

Table 2. (Continued) Area Flight management computer (FMC)

Type of issue Programming sequence Different fuel scales FMC difference change flow of briefing

Electronic check list (ECL)

General use Take-off configuration warning

General

Personal preferences

Specific example Major difficulty finding trigger to commence programming FMC. Having fuel in one FMC that was in 1,000 kg versus 1.0. The FO discusses that because data are first put into the FMC on the 500 compared with the 600, it changes flow of the briefing. No issue identified converting back to paper checklist from ECL. Some comments on change of personal flow. Slight confusion about take-off configuration warning difference between both aircraft. The 600 has a visual warning where the ECL ‘‘approves,’’ whereas the 500 only alerts if there is a problem. Outside of standard operating procedures, there are personal preferences. Both pilots describe struggling to get a flow going when they were in the 500.

Notes. ASI = air speed indicator; EFIS = electronic flight instrument system; FO = first officer; QNH = pressure setting allowing altimeter to read height above sea level allowing separation from terrain and other aircraft; VSI = vertical speed indicator.

Researcher: Is that right? FO: It has taken four starts. So when we started engine 1 [fourth start], I went straight to it. Captain: I was the same, I’ve been aiming a little too high initially, and then I’ve brought my eyes down to the right area [the captain is talking about his visual aim]. I guess it’s a little bit higher perhaps in the 600, I’m not sure. In both cases – for the captain and the FO – the differences in the position of the engine instruments created issues in relation to monitoring. Other issues relating to Ó 2015 Hogrefe Publishing

the EWD included pilots having difficulty finding flap indications, and the position of the trim indicator. As can be seen in Figure 5, the positions of many of the indicators vary between the two aircraft types.

Stage 5: Writing Reports and Texts for Publication in Academic Journals The final stage of the MFF study required the university team to write scholarly articles for review. The findings sections and the version submitted for peer review were the basis of the reports for the airline, which it could use in making decisions. In relation to the data already obtained Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

APAHF in Practice

Figure 5. Differing layouts and position of secondary instruments for ATR 72-500 (centre) and ATR 72-600 (left and right). The ATR72-500 center console display (center) is what pilots see all the time. For the ATR72-600 the flight management systems (FMS) displays the same instruments in different locations depending on phase of flight. a) When starting, pilots refer to high-pressure compressor speed (A); inter turbine temperature (B); and oil pressure (C). During engine start on the ATR72-600 these engine instruments automatically display on the multi function display (MFD). On completion of start sequence, the MFD revert back to modes each pilot had selected prior to engine start, for example navigation. b) During normal operations the instruments like engine torque (F), flap position indicator (G), and trim and indications (H) are displayed on the engine and warning display (EWD). Other instrument on the ATR72-500 (center) is the standby airspeed indicator (D) and standby altimeter (E).

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

in previous studies, the team wrote about MFF, not only in articles in applied journals, but also in broader theoretical papers concerning cognitive issues of flying an aircraft and moving between two versions. These journals, which sometimes include senior pilots as authors, have fed findings back to the airline to use for evidence-based decisions.

APAHF in Practice

Company Decision and Recommendation For the airline company, obtaining evidence that enabled relevant personnel to make informed decision about MFF was paramount. There existed numerous studies into pilots transitioning onto more advanced aircraft types – such as from analogue to glass cockpits – but there were few, if any, studies related to reverse transition generally. Furthermore, there were no studies on MFF for the ATR 72-500/ 600. The airline continued to take on recommendations from manufacturers, but was seeking greater clarity on the types of problems that might occur during the implementation of MFF. The survey study clearly showed that many pilots had concerns about MFF. However, the pilots in the think-aloud study were surprised with regard to the ease of the reverse transition. As the captain noted, ‘‘I came into yesterday thinking it was going to be a lot harder than what it has been. I’d probably be sitting at about 80%, 85% comfortable.’’ However, the cognitive study showed, for example, delayed actions and missing or misunderstood calls that might become more serious than the captain and first officer realized. Accordingly, the company made the decision to move cautiously to the next step of reviewing the forward and reverse transition course based on additional areas identified in the two studies. Issues that the company is still addressing include: 1. The potential of implementing a closer alignment of the SOPs of each type of aircraft; 2. The need to identify recommendations on consolidation time. That is, an appropriate time for new pilots to familiarize on one type (the ATR 72-500 and ATR 72-600 variant) prior to conducting MFF; 3. The need to identifying recency requirements relating to (a) whether each aircraft type could be flown on the same day, and (b) time allowed between flying particular variants; and 4. The need for an appropriate recurrent simulator training program.

Conclusion Modern airlines, like many industries, are faced with increased levels of complexity. Senior managers at the frontline of operational decision making are either faced with clear and easy-to-make decisions, or decisions that rely on less-than-obvious evidence. One issue relates to

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

dealing with evidence based on results at a general level and transferring it to practice. This paper shows how an airline can work closely with university-based researchers to conduct applied human factors research, with an aim of generating data for assisting in evidence-based decision making. Whereas many researchers may have their own areas of expertise and interest, and airlines may feel obliged to assist in generating new knowledge for the broader industry, there is a compelling argument for the airline to drive the agenda. In this paper, we describe one possible solution to this problem, which lies in collaboration. Because of the disciplinary demands and rigor in studies to be published, the airline obtains high-quality evidence, gathered under conditions that accord with scientific rigor. This dual perspective therefore meets the needs for both the airline and the university.

References Dekker, S. W. A. (2003). Illusions of explanation: A critical essay on error classification. International Journal of Aviation Psychology, 13, 95–106. International Air Transport Association. (2013). Evidence-based training implementation guide July 2013. Retrieved from http://www.iata.org/whatwedo/ops-infra/itqi/Documents/ ebt-implementation-guide.pdf Lawrence, T., Philips, N., & Hardy, C. (1999). Watching whale watching: Exploring the discursive foundations of collaborative relationships. Journal of Applied Behavioural Science, 35, 479–502. Mavin, T. J., Roth, W.-M., & Dekker, S. W. A. (2013). Understanding variance in pilot performance ratings: Two studies of flight examiners, captains and first officers assessing the performance of peers. Aviation Psychology and Applied Human Factors, 3, 53–62. doi: 10.1027/2192-0923/a000041 Roth, W.-M. (1998). Science teaching as knowledgeability: A case study of knowing and learning during coteaching. Science Education, 82, 357–377. Roth, W.-M., & Mavin, T. J. (2013). Assessment of nontechnical skills: From measurement to categorization modeled by fuzzy logic. Aviation Psychology and Applied Human Factors, 3, 73–82. doi: 10.1027/2192-0923/a000045 Wiener, E., Chute, R., & Moses, J. (1999). Transition to glass: Pilot training for high-technology transport aircraft. Retrieved from http://hdl.handle.net/2060/20000032979

Received March 13, 2014 Revision received October 23, 2014 Accepted for publication November 11, 2014 Published online April 10, 2015 Dr Timothy J. Mavin (EdD) is an associate professor at the Griffith Institute for Educational Research (Brisbane) where he studies skills assessment and learning. He has 10,000 hr of flight experience, including 7,000 hr of jet time. He continues to conduct type-rating endorsements on the Boeing 737. He is also a qualified high school teacher.

Ó 2015 Hogrefe Publishing

T. J. Mavin et al.: Evidence-Based Decision Making in Aviation

Wolff-Michael Roth (PhD, 1987) is Lansdowne Professor of Applied Cognitive Science at the University of Victoria, Canada. His research is in the domain of knowledge and learning across the lifespan from interdisciplinary perspectives. His work includes Graphing and uncertainty in the discovery sciences: With implications for STEM education (Dordrecht, 2014), and Geometry as objective science in elementary classrooms: Mathematics in the flesh (Routledge, 2011). Kassandra Soo (BSc) is a PhD candidate at Griffith University, Brisbane. She majored in psychology, completing her Honors thesis that investigated the effects of mixed fleet flying on pilot performance. She is currently investigating the learning and development of pilots.

Captain Ian Munro is the Airline Training Manager for Mount Cook Airlines. Previously he had spent 16 years in the military environment including the Royal New Zealand Air Force (RNZAF) as a fighter pilot. His senior military roles included senior training officer for fighter conversion for the RNZAF and as a flight commander of ‘‘Weapons Systems Operators’’ training for the Royal Saudi Air Force.

Correspondence Address Timothy J. Mavin Griffith Institute for Educational Research M15 3.18 Griffith University Mt Gravatt QLD 4122 Australia Tel. +61 7 3735-5604 E-mail t.mavin@griffith.edu.au

APAHF in Practice

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):52–61

Book Reviews Why Don’t Women Fly? Monica Martinussen

Book Reviews

Donna Bridges, Jane Neal-Smith, and Albert J. Mills (Eds.) Absent Aviators: Gender Issues in Aviation Ashgate, 2014 ISBN 978-1-4724-3338-1 Prices £ 75.00/$ 129.95 In 1928 a seaplane landed in Narvik, a small town on the coast of Northern Norway. It was a Swedish rescue team en route to Svalbard to search for the missing airship ‘‘Italia’’ used in an expedition led by Umberto Nobile. This was a big event in the town, and many people came to look at the seaplane and the pilots. One of the spectators was Gidsken Jacobsen, a 19-year-old girl who became fascinated by this event, both with the plane, the pilots and their mission. Later in life she often referred to this event as the starting point for her interest in aviation (Gynnild, 2007). Gidsken was the second woman in Norway to hold a pilot’s license and the first to be involved in commercial aviation including starting an airline (Gynnild, 2007). Her life and adventures make fascinating reading and one can only speculate what made it possible for a young woman in 1928 to make such untraditional choices. Today, almost 90 years later both the society and possibilities for young people have changed in Norway, but still only between 3 and 4% of commercial pilots are women. Why are so few women attracted to this profession? Is it a lack of abilities, motivation, attitudes, or is the answer to be found in the industry itself ? The book Absent Aviators addresses this question in addition to many other gender related topics. The book is an edited volume with a total of 13 chapters, and 24 different authors from Europe, Australia, South Africa, Canada, and the United States with experience from both civil and military aviation. The chapters include interesting historical material, theoretical perspectives, new research findings and reviews of previously published material. The first part of the book labeled ‘‘Identifying gender issues in piloting’’ includes an interesting historical analysis

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):62–63 DOI: 10.1027/2192-0923/a000073

of the employment and trading practices of Pan American Airways. The authors analyze different types of discrimination using sociological theory that explains how different categories (gender, ethnicity, and nationality) interact and contribute to injustice. The second chapter analyses the lack of women in both civil and military aviation by exploring why women do not see aviation as a possible career option. In addition, women pilots are interviewed to explore factors that have inspired them to fly. The third chapter includes a comprehensive review of research examining physiological differences between men and women related to flying which has often been listed as an argument for why women are not suitable for flying especially in a military context. Finally, this part of the book explores male pilots’ attitudes to women pilots through questionnaire data and a literature review. In general the findings indicated that male pilots were more negative towards women pilots than women pilots were. The second part of the book, ‘‘Barriers in military and civil aviation’’, includes four chapters examining different barriers women pilots face based on findings from various qualitative and quantitative studies. This includes findings from interviews with women pilots in the Australian Air Force and also an analysis of the Australian Defence Force exploring why both aviation and the military do not attract or retain women. This part of the book also reports the results from a comprehensive survey of gender related attitudes in three countries where the pilots’ comments to the survey were analyzed using qualitative methods. Many of the comments convey stereotypical perceptions of female pilots consistent with the quantitative part of the study. The final chapter in this part of the book presents the results from interviews with UK women pilots in order to identify their perceptions of barriers to employment in the aviation industry. The third part of the book, ‘‘Technical Issues’’, includes a number of chapters on new technologies being implemented in aviation. Two chapters seek to evaluate the introduction of the class cockpit, and different aspects such as attitudes and preferences for the class cockpit are explored. A third chapter examines this topic in relation to accidents and gender related strengths and weaknesses. An interesting

Ó 2015 Hogrefe Publishing

Book Reviews

experimental study was conducted in a flight simulator where the aim was to optimize the design for men and women respectively. The final part of the book, ‘‘Taking action’’, reports results from different attempts to attract more women to aviation, including efforts taken by the Royal Australian Air Force to increase the number of women pilots. Also the US project on teaching women to fly is described where the aim was to learn more about women’s motivation to become pilots, and also positive and negative experiences through flight training. Based on the results from this survey, a list of 10 barriers was identified but also 10 ways to promote women’s success in general aviation. The book presents the most comprehensive overview of gender-related issues in aviation to date. It summarizes relevant research and also provides helpful advice to encourage more women to seek a career as a pilot. Various sociological theoretical perspectives were presented that may be helpful when analyzing the problem, however a broader theoretical foundation or possible some introduction may have been helpful to the reader without a background in sociology. The book also includes many empirical findings even though is very difficult to conduct research on gender differences when the number of female pilots is so small compared to male pilots. This constitutes both a practical and a methodological problem. Hopefully this book may encourage other researchers to examine gender issues even though the possible sample size may be small. Only then will it be possible to learn more about

gender differences, and hopefully increase the number of female pilots, and avoid the loss of talent. Many of the chapters acknowledge that there are indeed gender differences, and that these may be utilized instead of viewed as obstacles. The book will undoubtedly be a valuable resource for students, researchers, and other professionals involved in aviation. Hopefully it may also encourage young women to follow the example set by Gidsken Jacobsen and overcome the barriers to becoming a pilot.

Reference Gynnild, O. (2007). Seilas i storm. Et portrett av flypioneren Gidsken Jacobsen [Stormy Sailing. A portrait of the aviation pioneer Gidsken Jacobsen]. Kabelvåg, Norway: Orkana Publisher.

Correspondence Address Monica Martinussen RKBU-Nord Faculty of Health Sciences UiT The Arctic University N-9037 Tromsø Norway Tel. +47 77 645-881 E-mail monica.martinussen@uit.no

Book Reviews

Ó 2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):62–63

News and Announcements The Best Paper Award Aviation Psychology and Applied Human Factors (APAHF) ISSN-Print 2192-0923 ISSN-Online 2192-0931

in the respective year. The winner will be selected by the Editor-in-Chief and the Associate Editors of APAHF. The award will be formally conferred at a special session at the bi-annual EAAP Conference. The winning paper will be published in the APAHF journal and the first author will be invited to present the paper in a Keynote Speech at the EAAP Conference.

Eligibility

News and Announcements

Aviation Psychology and Applied Human Factors (APAHF) is the peer-reviewed scientific journal of the European Association for Aviation Psychology (EAAP), in cooperation with the Australian Aviation Psychology Association (AAvPA) and published by Hogrefe Publishing, Göttingen, Germany. The journal APAHF invites innovative, original, high quality papers from researchers and practitioners addressing all psychological/human factors aspects of the aerospace domain. Aviation Psychology and Applied Human Factors is abstracted/indexed in PsycINFO, PSYNDEX, and Academic Index. The new annual APAHF Best Paper Award, introduced in 2015, recognizes excellence in aviation psychology and human factors research of a scientific paper, research note, or practitioner paper submitted to Aviation Psychology and Applied Human Factors which was accepted for publication

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):64 DOI: 10.1027/2192-0923/a000074

Any researcher is eligible to submit relevant work. Membership in EAAP or AAvPA is not required. Each manuscript accepted for publication in APAHF in 2015 will be considered for the Best Paper Award 2015. Submissions must cover original, unpublished research and comply with the manuscript submission instructions. Manuscripts should be submitted via the journal e-mail address: journal@eaap.net. Ioana Koglbauer Editor-in-Chief, Aviation Psychology and Applied Human Factors Contact: journal@eaap.net André Droog EAAP President Contact: president@eaap.net

Ó 2015 Hogrefe Publishing

News and Announcements

Meetings and Congresses ISAP – 18th International Symposium on Aviation Psychology, Dayton, OH, USA May 4–7, 2015 Contact: John Flach, Pamela Tsang,Wright State University, Dayton, OH, E-mail isap2015@isap.wright.edu, Web http:// isap.wright.edu/conferences/2015 The International Symposium on Aviation Psychology is convened for the purposes of: • presenting the latest research on human performance problems and opportunities within aviation systems; • envisioning design solutions that best utilize human capabilities for creating safe and efficient aviation systems; • and bringing together scientists, research sponsors, and operators in an effort to bridge the gap between research and application. Although the symposium is aerospace-oriented, we welcome anyone with basic or applied interests in any domain to the extent that generalizations from or to the aviation domain are relevant. ATM2015 – 11th USA/Europe Air Traffic Management Research and Development Seminar, Lisbon, Portugal June 23–26, 2015 Contact: Colin Meckiff, EUROCONTROL, E-mail colin. meckiff@eurocontrol.int, Eric Neiderman, FAA, E-mail eric.neiderman@faa.gov, Web http://www.atmseminar.org Since 1997 the Federal Aviation Administration (FAA) and the EUROCONTROL have jointly organized an international seminar series for air traffic management research and development. These conferences are held biannually, alternating between Europe and the USA, and have become the top event for ATM researchers.

6th International Conference on Applied Human Factors and Ergonomics, Las Vegas, NV, USA July 26–30, 2015 Contact: AHFE, Conference Administrator, E-mail ahfeadmin@twc.com, Web http://www.ahfe2015.org

2015 Hogrefe Publishing

The conference objective is to provide an international forum for the dissemination and exchange of scientific information on theoretical, generic, and applied areas of ergonomics, including, physical ergonomics, cognitive and neuroergonomics, social and occupational ergonomics, cross-cultural aspects of decision making, ergonomics modeling and usability evaluation, human digital modeling, healthcare and special populations, human factors in oil, gas and nuclear energy industries, human factors in unmanned systems, safety management and human factors, ergonomics in design, affective and pleasurable design, human factors, software, and systems engineering, transportation (road and rail, maritime and aviation), training and human performance, occupational safety management, and the human side of service engineering. This will be accomplished through the following six modes of communication: keynote presentation, parallel sessions, demonstration and poster sessions, tutorials, exhibitions, and meetings of special interest groups. The 5-day conference will start with tutorials. The tutorials will be held on July 26–27, 2015. Tutorials will be offered at introductory, intermediate, and advanced levels covering the entire spectrum of the conference. Please visit the website for further information about submission of abstracts and papers. Second African Human Factors and Aviation Symposium, Cape Town, South Africa September 5–11, 2015 Contact: Paper submission enquiries: E-mail Andrew.Thatcher@wits.ac.za. Attendance & sponsorship enquiries: Jaco vd Westhuizen, ATNS Head Office, Johannesburg, South Africa, E-mail jacovdw@atns.co.za and Matita Tshabalala, E-mail matitat@atns.co.za The main aim of the symposium is to create an opportunity for the industry to network with academia and explore how mutual efforts in combining research rigor with operational experience can improve safety performance in the African context of the aviation industry. Furthermore, the symposium provides an opportunity for the aviation industry to obtain first-hand information on current system safety related research. The symposium also aims to create an interest amongst young researchers to become involved in aviation as a research field and junior operational staff to become accustomed to research techniques. Therefore, both operational and academic papers are accepted and the symposium will also include expert panel discussions on critical human performance issues within the industry. HFES 2015 International Annual Meeting, Los Angeles, CA, USA October 26–30, 2015 Contact: HFES/Annual Meeting, Santa Monica, CA, USA, E-mail lois@hfes.org, Web https://www.hfes.org// Web/HFESMeetings/meetings.html HFES Annual Meetings are important events for the Society’s members and others who are interested in the latest developments in the field. The Society’s mission is to promote the discovery and exchange of knowledge concerning

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):65–66 DOI: 10.1027/2192-0923/a000075

News and Announcements

ISAP’15 – 7th International Summer School on Aviation Psychology, Graz, Austria July 6–10, 2015 Contact: Wolfgang Kallus, University of Graz, Austria, E-mail isap@uni-graz.at, Web http://www.uni-graz.at/ isap15/ The 7th International Summer School is organized by the University of Graz in collaboration with the Austrian Aviation Psychology Association (AAPA). As such the ISAP 2015 is recognized as special course in the ‘‘Dissertation Program’’ as well as the ‘‘Top Performance Program’’ of the University of Graz. Terms and conditions for EAAP membership and certification/registration can be found on the EAAP Web-site: http://www.eaap.net.

News and Announcements

the characteristics of human beings that are applicable to the design of systems and devices of all kinds. Human Factors in Flight Safety – SMS, Risk Management & Safety Investigation. An International Course May 10–14, 2015, Dubai, UAE May 18–22, 2015, Barcelona, Spain May 26–30, 2015 (advanced), Barcelona, Spain Contact: Brent Hayward, E-mail bhayward@dedale.net, Web http://www.eaap.net/courses.html The European Association for Aviation Psychology (EAAP) is proud to announce the 2015 Human Factors in Flight Safety Courses dedicated to ‘‘Safety Management Systems, Risk Management and Safety Investigation.’’ This course has been conducted on behalf of the EAAP since 1999 by Rob Lee, Kristina Pollack, and Brent Hayward. A complete overview on all aspects of the courses, including objectives, content, participants, registration, location and much more, is available from the above website.

News and Announcements

6th International Conference on Traffic and Transport Psychology, Brisbane, Australia August 2–5, 2016 Contact: QUT Conferences, Event Manager – Alanna Hankey, E-mail icttp2016@qut.edu.au, Web http://icttp2016.com

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):65–66

With a theme of ‘‘Taking Traffic and Transport Psychology to the World’’, this conference will feature a strong program of keynote speakers, oral and poster presentations, workshops and symposia. The Conference will be a global forum at which all those involved in traffic and transport psychology, human factors, cognition and behaviours, road safety research, policy, education, enforcement and injury prevention, can meet with researchers, academics, and professionals to discuss and present on the latest work being undertaken in these areas.

32nd EAAP Conference September 26–30, 2016 Contact: European Association for Aviation Psychology (EAAP), E-mail secretarygeneral@eaap.net, Web http:// conference.eaap.net EAAP organizes conferences biannually to encourage EAAP members and guests to share their latest findings, data and experience in academia, research and applied aviation psychology. For details please visit the above website, news will follow soon.

2015 Hogrefe Publishing

News and Announcements

Aviation Human Factors Related Industry News1 FAA Ready for 2015

For further information, see: http://www.faa.gov/news/press_releases/news_story.cfm? newsId=15795 http://www.faa.gov/nextgen http://www.faa.gov/news/press_releases/news_story.cfm? newsId=16875

Qantas Named World’s Safest Airline for 2015 AirlineRatings.com, the world’s only safety and product rating website, has announced its top ten safest airlines and top ten safest low cost airlines for 2015 from the 449 it monitors. Top of the list again is Qantas which has a fatality free record in the jet era. Making up the remainder of the top ten in alphabetical order are: Air New Zealand, Cathay Paciﬁc Airways, British Airways, Emirates, Etihad Airways, EVA Air, Finnair, Lufthansa, and Singapore Airlines. AirlineRatings.com’s rating system takes into account a variety of factors related to audits from aviation’s governing bodies and lead associations as well as government audits and the airlines’ fatality records. AirlineRating.com’s editorial team, one of the world’s most awarded and experienced, also examined airlines’ operational histories, incident records and operational excellence to arrive at its top ten safest airlines. According to AirlineRatings.com editor, Geoffrey Thomas, ‘‘our top ten safest airlines are always at the

Parts of this section are compiled from ‘‘Aviaion Human Factors Industry News’’ and reproduced with permission of Roger Hughes.

2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):67–69 DOI: 10.1027/2192-0923/a000077

News and Announcements

The Federal Aviation Administration’s enduring mission is to ensure that the United States has the safest, most efficient aerospace system in the world. In 2014, the FAA has taken many steps to accomplish this result. Here are highlighted three particular efforts. First, the FAA issued a rule that requires helicopter operators, including air ambulances, to have stricter flight rules and procedures, improved communications, training, and additional on-board safety equipment. This rule will help reduce safety risk involved in helicopter operations and help pilots make good safety decisions through the use of better training, procedures, and equipment. The rule represents the most significant improvements to helicopter safety in decades and responds to government and industry concerns over continued risk in helicopter operations. Second, the FAA continues to modernize the airspace system by implementing NextGen – our major initiative to make flying more efficient and greener, while ensuring that all safety needs are met. NextGen includes our Metroplex initiative, an effort to reduce air traffic congestion in the nation’s busiest metropolitan areas. As part of our Metroplex initiative, we implemented scores of new satellite-based air traffic procedures in several major cities, including Houston, North Texas and Washington, D.C. These procedures are helping to increase on-time arrival, and reduce fuel consumption and emissions, results that are benefiting the airlines, the passenger, and the environment. For instance, in the North Texas area, we implemented 80 new satellitebased air traffic procedures. We project that airspace users will save 4.1 million gallons of fuel each year, resulting in a savings of 41,000 metric tons of carbon dioxide emissions and $10.3 million dollars. Third, the FAA continues to make strides toward safely integrating unmanned aircraft into the nation’s airspace system. Unmanned aircraft are a burgeoning technology, and the application of the technology is limited only by our imagination. The FAA has chosen six unmanned aircraft test sites that are now operational. The research conducted at these sites will help inform the FAA as we move forward with integration. In addition, the FAA has begun a process for approving commercial unmanned aircraft operations in

low-risk situations such as moviemaking, agricultural research and utility surveying. Unmanned aircraft often provides a safer alternative to work that can be very high risk in manned aircraft. The agency hopes to issue a rule for public comment very soon on small unmanned aircraft. These three accomplishments are just a few of the many projects we’ve been focused on in 2014. In the coming year, we look forward to continuing these efforts to ensure the safety and efﬁciency of aviation for our nation.

News and Announcements

forefront of safety innovation and launching new planes.’’ ‘‘These airlines are a byword for excellence,’’ he added. ‘‘There is no doubt that Qantas is a standout in safety enhancements and an industry benchmark for best practice,’’ said Mr Thomas. Over its 94-year history Qantas has amassed an extraordinary record of ﬁrsts in safety and operations and is now accepted as the world’s most experienced airline. ‘‘Qantas has been the lead airline in virtually every major advancement in airline safety over the past 60 years,’’ said Mr Thomas. Responding to public interest, the AirlineRatings.com editors also identiﬁed their top ten safest low cost airlines. These are in alphabetical order: Aer Lingus, Alaska Airlines, Icelandair, Jetstar, Jetblue, Kulula.com, Monarch Airlines, Thomas Cook, TUI Fly, and Westjet. ‘‘Unlike a number of low cost carriers these airlines have all passed the stringent International Air Transport Association Operational Safety Audit (IOSA) audit and have excellent safety records,’’ said Mr Thomas. ‘‘Low cost does not mean low safety.’’ Of the 449 airlines surveyed 149 have the top seven-star safety ranking, but almost 50 have just three stars or less. Five airlines only achieved one star for safety from AirlineRatings.com. These are: Agni Air, Kam Air, Nepal Airlines, Scat, and Tara Air.

For several years, the FAA has used the results of an airport capacity tool called the runwaySimulator model, developed by MITRE. The FAA and MITRE are now making the tool publicly available for aviation applications, including airport planning. The tool is designed to assess an airport’s existing capacity, as well as capacity improvements such as new infrastructure or ﬂight procedures. The tool replaces the FAA’s Airﬁeld Capacity Model (ACM), which is now dated. You can request access to the tool and related training at http://www.faa.gov/airports/planning_capacity/runwaysimulator/ in order to • Request access to the runwaySimulator tool, license, and training • Submit a software bug report

Sleeping on the job is a necessary reality for many night shift workers, but a new study suggests that instead of providing an energy boost, a night-time nap might put workers at risk. Research on sleep inertia – the state you are in when you first wake up – by the University of South Australia’s Centre for Sleep Research PhD candidate Cassie Hilditch has particular relevance for night shift workers in safety-critical industries such as healthcare or transport, who have to return from breaks and operate at full capacity. ‘‘Sleep inertia is the groggy feeling most people experience when waking up, and is characterized by slow reaction times, poor decision-making, and reduced information processing,’’ Hilditch says. ‘‘This doesn’t matter for people getting dressed in the morning, but for workers in industries such as aviation, petrochemicals, transport, and health, post-nap alertness is critical for workplace safety.’’ Hilditch’s study found a 30-minute nap during a night shift produced long-lasting sleep inertia, with recovery times of up to 45 minutes. A 10-minute nap during a night shift, however, helped stabilize performance during the hour after waking, with little-to-no sleep inertia. Hilditch says her research shows the importance of workers allowing time between a nap and the recommencement of work. ‘‘Our research suggests that if you have a 30-minute break in a shift at night, it’s better to take a 10-minute nap at the start of your break. Don’t take a 30-minute nap if you need to return to work straight away,’’ Hilditch says. ‘‘Our participants were well-rested before the study, so these are likely to be best-case figures, as shift workers may already have cumulative fatigue, which could prolong recovery from sleep inertia. In the real world, people are carrying a lot of sleep debt.’’ Cognitive tests also revealed participants tended to overestimate their abilities after a nap, with the gap between perception and reality producing further risk. ‘‘If sleep inertia persists beyond your break, and you think you’re more alert than you actually are while, say, operating heavy machinery, then there is a clear safety risk,’’ Hilditch says. ‘‘One of the challenges is getting people to recognize their limitations. Shift workers might think that since they’ve been doing shift work for 6 years, they are fine, but they might not be – many studies support this.’’ Hilditch’s findings may also have an implication for desk-based jobs. ‘‘Lawyers or people in finance might work

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):67–69

2015 Hogrefe Publishing

Adapted with permission from http://www.airlineratings. com/news.php?s&id=425

FAA runwaySimulator Airport Capacity Model

News and Announcements

For Night Shift Workers, Sleep Inertia Adds Risk to Naps

News and Announcements

super-long hours, and their decision-making is just as impaired as the next person’s; it’s just that the risk is ﬁnancial,’’ she says. Prior to her PhD, Hilditch spent 5 years working for a research consultancy in London, undertaking fatigue-related research in safety-critical industries. In this role, she developed fatigue risk management systems in settings ranging from the Libyan desert to the Swiss Alps.

‘‘Trying to give people advice on how to schedule shifts made me realize we don’t know enough to provide all the details,’’ Hilditch says.

2015 Hogrefe Publishing

Aviation Psychology and Applied Human Factors 2015; Vol. 5(1):67–69

Adapted with permission from: Peter Krieg, ‘‘Asleep on the job: new study explores napping on nightshift,’’ URL http://www.unisa.edu.au/Research/Centre-for-Sleep-Research. by University of South Australia, Adelaide, SA.

News and Announcements

The 18th International Symposium on Aviation Psychology

May 4-7, 2015 Dayton, Ohio U.S.A. isap.wright.edu

Instructions to Authors – Aviation Psychology and Applied Human Factors Aims and Scope: Aviation Psychology and Applied Human Factors (APAHF) publishes innovative, original, high-quality applied research covering all aspects of the aerospace domain. In order to make the journal accessible to both practitioners and scientific researchers, the contents are broadly divided into original scientific research articles and papers for practitioners. The fully peer-reviewed Original Articles cover a variety of methodological approaches, ranging from experimental surveys to ethnographic and observational research, from those psychological and human factors disciplines relevant to the field, including social psychology, cognitive psychology, and ergonomics. High-quality critical review articles and meta-analyses cover particular topics of current scientific interest. Shorter studies are published as Research Notes. APAHF in Practice consists of shorter, less technical, but still fully peer-reviewed articles covering a wide range of topics, such as comments on incidents and accidents, innovative applications of aviation psychology, and reviews of best practices in industry. Book Reviews present recently published books of particular interest. Finally, the journal’s News and Announcements section features information about past and upcoming events around the world, association news, interviews, and similar. Please read the following information carefully before submitting a document to APAHF: Address for Submissions: Submissions for any section of APAHF must be sent to the journal’s e-mail address journal@eaap.net. Submission Letter: Manuscripts must be accompanied by a formal letter of submission addressed to the Editor in Chief confirming that: (a) the manuscript has not been published previously (and is not under consideration for another journal); (b) the manuscript is being submitted with the agreement of all authors; and (c) all participants and their data were treated in accordance with appropriate ethical guidelines. Enquiries and Help: Should you have any technical queries regarding this process, please contact Michaela Schwarz (admin@eaap.net, Tel. +43 664 350-7497). Informal enquiries concerning the content and format of papers should be addressed to Ioana Koglbauer (PhD, Editor-in-Chief), Graz University of Technology, Institute of Mechanics, Kopernikusgasse 24/IV, 8010 Graz, Austria, journal@eaap.net) Types and Lengths of Manuscripts Original Articles are scientific papers with a maximum of 6,000 words (including references), based on any methodological approach (e.g., experimental surveys; ethnographic research; observational research) and from any psychological/human factors discipline (e.g., social psychology; cognitive psychology; ergonomics). High-quality critical review articles or meta-analyses covering a particular topic of current scientific interest are of particular interest. Research Notes are scientific papers with a maximum of 3,000 words (including references), and no more than 1 figure, 2 tables, and 12 references. APAHF in Practice papers are shorter than scientific papers (maximum of 4,000 words including references), and may cover a range of topics, including (but not limited to): comment on incidents and accidents; innovative applications of aviation psychology; reviews of best practice in industry. They should be written in a more accessible (less technical) style and may be more liberally illustrated with diagrams and pictures. Original Articles, Research Notes, and APAHF in Practice papers are all subject to double-blind peer review. News and Announcements pieces are brief and informal. General: All manuscripts should be prepared in accordance with the Publication Manual of the American Psychological Association 6th ed. (APA Manual). A free tutorial about APA style is available at http:// www.apastyle.org/learn/tutorials/basics-tutorial.aspx. Anonymization: Original Articles, Research Notes, and APAHF in Practice papers will undergo double-blind peer review. Authors should therefore remove all potentially identifying information from the submitted manuscript, replacing names and any indication of the institution where a study was conducted by neutral placeholders. Upon acceptance, this information must of course be reinstated. Title Page: To facilitate blind reviewing, the title page of the submitted manuscript should include only the paper’s title and running head. A second title page including all author information should be submitted as a separate document. This should include the title; author name(s); affiliation(s); and an address for correspondence (including the name of the corresponding author with e-mail and phone numbers). Abstract / Keywords: An abstract (maximum length 120 words) should be printed on a separate sheet for Original Articles, Research Notes, and APAHF in Practice papers. A maximum of 5 keywords should be given after the abstract. References: Citations in the text and in the reference list proper should follow conventions listed in the APA Manual. Tables / Figures: Table and figures should be numbered separately using Arabic numerals and must all be cited in the text (e.g., ‘‘As shown in Table 1,. . .’’ – ‘‘As can be seen in Figure 1,. . .’’). Each figure and each table

should be printed on a separate sheet. Each table should have a brief descriptive title; this should then be followed by the body of the table. Longer explanations, if required, should be included in a footnote to the table. Figures must be accompanied by a legend. Figures may be reproduced in color in the online journal, but will normally be reproduced in black/white or grayscale only in the printed journal, which authors should take into consideration when designing figures. Scientific Nomenclature and Style: Authors should follow the guidelines of the APA Manual regarding style and nomenclature. Authors should avoid using masculine generic forms in their manuscripts. General statements about groups of people should be written in gender-neutral form (see APA Manual, pp. 73–74); when presenting examples, authors may alternate between female and male forms throughout their text. Language: It is recommended that authors who are not native speakers of English have their papers checked and corrected by a native-speaker colleague before submission. Standard US American spelling and punctuation as given in Webster’s New Collegiate Dictionary should be followed. Proofs: PDF proofs will be sent to the corresponding author. Changes of content or stylistic changes may only be made in exceptional cases in the proofs. Corrections that exceed 5% of the typesetting costs may be invoiced to the authors. Free Copies: Hogrefe will send the corresponding author of each accepted paper free of charge an e-offprint (PDF) of the published version of the paper when it is first released online. This e-offprint is provided for the author’s personal use, including for sharing with coauthors (see also ‘‘Online Rights for Journal Articles’’ in the Advice for Authors on the journal’s web page at www.hogrefe.com). For each article, the corresponding author will receive a total of 5 free copies of the issue in which his/her article appears, to be shared with any coauthors. Additional copies may be requested against payment; such requests should be made, at the latest, when returning the corrected proofs. Copyright Agreement: By submitting an article, the author confirms and guarantees on behalf of him-/herself and any co-authors that the manuscript has not been submitted or published elsewhere, and that he or she holds all copyright in and titles to the submitted contribution, including any figures, photographs, line drawings, plans, maps, sketches, and tables, and that the article and its contents do not infringe in any way on the rights of third parties. The author agrees, upon acceptance of the article for publication, to transfer to the publisher the exclusive right to reproduce and distribute the article and its contents, both physically and in nonphysical, electronic, or other form, in the journal to which it has been submitted and in other independent publications, with no limitations on the number of copies or on the form or the extent of distribution. These rights are transferred for the duration of copyright as defined by international law. Furthermore, the author transfers to the publisher the following exclusive rights to the article and its contents: 1. The rights to produce advance copies, reprints, or offprints of the article, in full or in part, to undertake or allow translations into other languages, to distribute other forms or modified versions of the article, and to produce and distribute summaries or abstracts. 2. The rights to microfilm and microfiche editions or similar, to the use of the article and its contents in videotext, teletext, and similar systems, to recordings or reproduction using other media, digital or analog, including electronic, magnetic, and optical media, and in multimedia form, as well as for public broadcasting in radio, television, or other forms of broadcast. 3. The rights to store the article and its content in machine-readable or electronic form on all media (such as computer disks, compact disks, magnetic tape), to store the article and its contents in online databases belonging to the publisher or third parties for viewing or downloading by third parties, and to present or reproduce the article or its contents on visual display screens, monitors, and similar devices, either directly or via data transmission. 4. The rights to reproduce and distribute the article and its contents by all other means, including photomechanical and similar processes (such as photocopying or facsimile), and as part of so-called document delivery services. 5. The right to transfer any or all rights mentioned in this agreement, as well as rights retained by the relevant copyright clearing centers, including royalty rights to third parties. Hogrefe OpenMind: Information about the open access publishing program Hogrefe OpenMind, including the article processing fee and the Creative Commons license under which the article will then be published, are given at www.hogrefe.com/openmind. Online Rights for Journal Articles: Guidelines on authors’ rights to archive electronic versions of their manuscripts online are given in the Advice for Authors on the journal’s web page at www.hogrefe.com.

March 1, 2015. Ó Hogrefe Publishing