Pi Sigma Alpha Undergraduate Journal of Politics Spring 2018 by psajournalou

Vol. XVIII No. 1

Spring 2018

PI SIGMA ALPHA

Undergraduate Journal of Politics

Oakland University

Pi Sigma Alpha Undergraduate Journal of Politics

The Pi Sigma Alpha Undergraduate Journal of Politics (ISSN 1556-2034) is published biannually by the Nu Omega Chapter of Pi Sigma Alpha, Oakland University, Department of Political Science, Varner Hall, Room 418, 371 Varner Drive, Rochester, MI 48309-4485. The Journal is funded by Pi Sigma Alpha, the National Political Science Honor Society, 1527 New Hampshire Avenue, NW, Washington, DC 20036, http://www. pisigmaalpha.org. The Pi Sigma Alpha Undergraduate Journal of Politics was founded in the spring of 2001 by Delta Omega Chapter of Pi Sigma Alpha at Purdue University, under the name The American Undergraduate Journal of Politics and Government. With the sponsorship of Pi Sigma Alpha, the National Political Science Honor Society, the name of the Journal was changed to The Pi Sigma Alpha Undergraduate Journal of Politics as of the Fall 2004 edition. Electronic editions of the Journal are available online at http://www.psajournal. org. For further information, please contact Dr. Terri Towner at Oakland University (towner@ oakland.edu). All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the editors and faculty advisors of The Pi Sigma Alpha Undergraduate Journal of Politics. The Pi Sigma Alpha Undergraduate Journal of Politics and content appearing there-in is copyrighted by Pi Sigma Alpha. While holding these rights, Pi Sigma Alpha does not exert editorial or other control over the content of the Journal or the decisions or actions of its staff in the course of normal business operations. As such, Pi Sigma Alpha neither asserts nor accepts responsibility for the content or actions of staff of the publication in the normal course of business as the customs and usages of the law allow. All assertions of fact and statements of opinion are solely those of the authors. They do not necessarily represent the views of Pi Sigma Alpha, the National Political Science Honor Society, the Editorial Board, the Advisory Board, the Faculty Advisors, Oakland University, or its faculty and administration. COPYRIGHT ÂŠ 2018 PI SIGMA ALPHA. ALL RIGHTS RESERVED

ÂŠ Pi Sigma Alpha 2018

Grievance Theory and the Causation of the Arab Spring

The Pi Sigma Alpha Undergraduate Journal of Politics Spring 2018 Volume XVIII

Number 1 Thirty-Fifth Edition

Abigail Brown Meredith Kate Hubbell Jana Al-Jarrah Tanir-Vefa Avci Dr. Terri L. Towner

Outreach Editor Co-Scheduling/Content Editor Co-Scheduling/Content Editor Cover Designer Faculty Advisor and Editor

Editorial Board Jessica Abdo Jacob Adams Jana Al-Jarrah Tanir-Vefa Avci Alex Bertges Maria Boersma Abigail Brown Ryan Dempsey

Rolland Giberson Ajaypal Gill Hannah Gorosh Meredith Kate Hubbell Eleanor Lauer Christina Walker Brian Quinn

Advisory Board Dr. Robert Alexander Dr. Nicole Asmussen Mathew Dr. Cristian Cantir Dr. Rosalee Clawson Dr. Matthew Dabros Dr. David Dulio Dr. Natasha Duncan Dr. Cody Eldredge Dr. Alan Epstein Dr. Megan Hershey Dr. Dwaine Jengelley Dr. John Klemanski

Dr. Jeanine Kraybill Dr. Paulette Kurzer Dr. Laura Landolt Dr. Anthnoy Nowns Dr. Daniel O’Neill Dr. Zoe Oxley Dr. Ronald Rapoport Dr. Jo Reger Dr. Jaime Settle Dr. Harry “Neil” Strine Dr. Peter Trumbore Dr. Kali Wright-Smith

Pi Sigma Alpha Undergraduate Journal of Politics

Editorâ&#x20AC;&#x2122;s Preface to the Spring Edition The Pi Sigma Alpha Journal of Undergraduate Politics would first and foremost like to acknowledge all those individuals and institutions which make the publication of this Journal possible semester after semester and year after year. The Journal has continued to grow in terms of submissions, quality, and prestige. Admissions to the Spring 2018 edition were both vast in number and constituted a diverse array of topics. We greatly appreciate all those who have submitted their work to the Journal in hopes of being published. The articles published herein exemplify a high quality sample of the types of undergraduate research being conducted across the country. Although the publication is a completely student-run endeavor, the efforts of the student Editorial Board are guided and supported by a number of individuals and institutions which we would like to thank. First, we would like to thank the Pi Sigma Alpha Executive Council and Executive Committee whose vision and financial support has maintained the quality and direction of the Journal. Second, we would like to thank the Faculty Advisory board: the thorough and constructive reviews provided by the members of this board have ensured the articles published herein meet a consistent standard of quality. Finally, we extend tremendous thanks to Editorial Board Faculty Advisor Terri Towner, who has clocked countless hours to ensure the integrity of the Journal continues to exceed the standards of excellence set by the editors of its previous editions. The Editorial Board at Oakland University is proud to present the Spring Edition which contains a wellrounded set of articles with varied methodological approaches and topical matter. The publishing process for the Spring Edition followed a relatively smooth path from submission to publication, and the Nu Omega Chapter and Oakland University wish the readers of this edition a similarly enjoyable time. Best, The Editors

ÂŠ Pi Sigma Alpha 2018

Grievance Theory and the Causation of the Arab Spring

Submission of Manuscripts The Journal accepts manuscripts from undergraduates of any class and major. Members of Pi Sigma Alpha are especially encouraged to enter their work. We strive to publish papers of the highest quality in all areas of political science. Generally, selected manuscripts have been well-written works with a fully developed thesis and strong argumentation stemming from original analysis. Authors may be asked to revise their work before being accepted for publication. Submission deadlines are October 1st for the Fall edition and February 1st for the Spring edition. Manuscripts are accepted on a rolling basis; therefore early submissions are strongly encouraged. To submit your work, please email psajournalou@gmail.com with an attached Word document of the manuscript. Please include your name, university and contact details (mailing address, email address, and phone number). If possible, include how you heard about the Journal. Submitted manuscripts must include a short abstract (approximately 150 words), citations and references that follow the APSA Style Manual for Political Science. Please do not exceed the maximum page length of 35 double-spaced pages. The Journal is a student-run enterprise with editors and an Editorial Board that are undergraduate students and Pi Sigma Alpha members at Oakland University. The Editorial Board relies heavily on the help of our Advisory Board consisting of political science faculty from across the nation, including members of the Pi Sigma Alpha Executive Council. Due to the time committed to the manuscript review process, we would like to remind students to submit only one manuscript at a time. Please direct any questions about submissions or the Journalâ&#x20AC;&#x2122;s upcoming editions to our editors at psajournalou@gmail.com.

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

Contents An Experiment in Race and Voter Turnout in the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Chenoa Yorgason, University of Pittsburgh The Determinants of Death Penalty Policy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Christina Martin, Susquehanna University The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Kerry Dugandzic, Connecticut College Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level . . . . . . . . . . . . . . . . . 45 Zoe Bouras, Illinois Wesleyan University

An Experiment in Race and Voter Turnout in the United States

An Experiment in Race and Voter Turnout in the United States Chenoa Yorgason, University of Pittsburgh Recent presidential elections have brought the race and gender of candidates to the forefront of discussion. In this paper, how the race of an individual and the race of a candidate impact voting patterns are examined using results from an online experiment. In the experiment, it is tested how descriptive representation is desired when neither candidate explicitly provides any substantive representation. It is hypothesized that all else held equal, turnout will be higher when there is a candidate of the same race as the potential voter, that voters will prefer the candidate of the same race, and that more marginalized groups will prefer candidates of their own race at higher levels than less marginalized groups. There is no evidence that race impacts voter turnout in any racial group. African-Americans tend to strongly prefer the African-American candidate once at the polls, but no other groups preferred the candidate of the same race. INTRODUCTION AND MOTIVATION

n Barack Obama’s first presidential race in 2008, Black voter turnout was nearly 10% higher than it was in 2004 (McDonald 2016). In 2016, women overwhelmingly voted for Hillary Clinton and gave the election among the widest gender gaps in terms of candidate preference (Tyson and Maniam 2016). Heartfelt stories about how the nation’s first Black president or the prospect of the nation’s first female president encouraged historically underrepresented demographics that their own political aspirations were within reach ran rampant in the media of last decade. These stories have materialized, as the United States is now experiencing an increase in female candidates for political office an election cycle after the 2016 presidential election (Tackett 2017). This prompts the following questions: to what extent does sharing descriptive characteristics with a candidate encourage political participation? Furthermore, to what extent do individuals prefer the candidate with the shared characteristic? When considering political representation, two types of representation are generally discussed. The first is substantive representation, in which the electorate votes for candidates who advocate for policies in line with the desires of the constituency. This ensures that the most agreed-upon ideologies are represented in government. The latter is descriptive representation, in which the electorate votes for candidates with similar descriptive attributes, such as race, employment, birthplace, or gender. Given that the electorate and the candidate likely have shared experiences because of these attributes, it is expected that the candidate understands the needs and wants of the electorate and will pursue those when elected. It is important to note that descriptive and substantive representation are not mutually exclusive: an elected official can be one, both, or neither. It is difficult to separate descriptive and substantive representation among real-life candidates

because variables commonly associated with descriptive representation, particularly race, gender, and geographic region, are strongly correlated with political party and ideology. The effectiveness of descriptive representation is wellstudied, and the results point to some inconclusion in its usefulness. Haider-Markel (2007) studied the impact of LGBT legislators in 1992-2002 and found that the election of LGBT legislators results in substantive representation, but comes with an increase in the introduction of anti-LGBT legislation. Meanwhile, women’s groups within government and general women’s movements are associated with more effective womenspecific policy than do the presence of female legislators (Weldon 2002). In some contexts, the election of additional women legislators results in diminishing marginal benefits in their effectiveness (Kanthak and Krause 2012). Overwhelming evidence suggests, however, that increased political involvement of citizens occurs once there is descriptive representation. Mansbridge (1999, 642) states, “Black representatives…are likely to be contacted by Blacks “throughout the region” and not just in their own districts.” Mariani, Marshall, and Mathews-Schultz (2015, 726) find that female Democratic candidates, such as House Speaker Nancy Pelosi and presidential candidate Hillary Clinton, “… enhance young women’s anticipated political involvement [in ways] that cannot be explained by ideological congruence.” Schwindt-Bayer and Mishler (2005) use data from 31 countries and find that both male and female confidence in legislators increases when the gender distribution of the legislature is more proportional. Political efficacy (Merolla, Sellers, and Fowler 2013) and positive evaluations of Congress (Brunell, Anderson, and Cremona 2008) from African-Americans increase when there is a descriptive representative in a national office. LatinoAmericans are less politically alienated when Latino-American representatives are in state-level and U.S. Congressional seats

Pi Sigma Alpha Undergraduate Journal of Politics

(Pantoja and Segura 2003). Additional studies indicate that this behavior also manifests itself in electoral decision-making. Other factors aside, men prefer male representatives and women prefer female representatives (Rosenthal 1995). Female officeholders running alongside other female candidates results in increased political interest from women (Reingold and Harrell 2010). However, desires for descriptive representation do not necessarily translate to votes for the candidate with underrepresented attributes: in the 2016 general election, women of color that expressed desire in having more women in office overwhelmingly voted for Clinton, although this trend was absent among white women (Phillips 2018). Do these preferences fare well when descriptive representation is based on race instead of gender? To answer the latter question, Washington’s (2006) work revealed that as the presence of Black Democratic candidates on the ballot increases, turnout among Blacks also increases (although no such relationship is found with Black Republican candidates). Other works reveal that voter engagement (Clark 2008) and turnout (Rocha, Tolbert, Bowen, and Clark 2010) is higher among Blacks in districts represented by Blacks. Cho (1999) finds that Asian-Americans tend to donate to other AsianAmerican candidates, regardless of their competitiveness. Cumulatively, it is clear that descriptive factors of representatives have an impact on American political efficacy among traditionally underrepresented populations. These effects are widespread and not constrained to a single group or to particular types of political manifestations. Each of these studies makes good use of available election-related data, but are not able to completely separate the effects of descriptive representation from substantive representation. Washington (2006), Phillips (2018), and Mariani, Marshall, and MathewsSchultz’s (2015) results suggest that the actual desires for descriptive representation are conditional on additional factors such as party. That is, responsiveness to descriptive factors seems to rely on at least some degree of substantive representation already being achieved.

Preliminary Examination in House of Representatives Elections

Using data from the American National Election Studies (2010, 2014), in the 2008 and 2012 elections, the presence of a nonwhite candidate (versus only white candidates) in House of Representatives races yielded increases in turnout among Blacks by 2-4%. For all other races, turnout in the presence of a nonwhite candidate decreased in 2008 but increased in 2012 (except for Latino-Americans, who had a less than 1% decrease). Data from the Cooperative Congressional Election Study (Ansolabehere 2006, 2010) further complicate the picture. In 2006, nonwhite voter turnout decreased by about 6% with the presence of a nonwhite House candidate, relative to elections with only white candidates. White voter turnout increased by about 6% when a nonwhite candidate was running in a House election. In 2010, nonwhite voter turnout 8

again decreased by about 6% with the presence of a nonwhite House candidate, while white voter turnout decreased by 1.3%. These results leave an inconclusive explanation. Perhaps the patterns shown are congruent with Reingold and Harrell’s (2010) finding with regards to women running for office: individual women running do not generate excitement towards ongoing elections, but multiple women running do increase excitement. This explains the slight increases in voter turnout among Black voters during the presidential election years, as this turnout is associated with multiple, higher-office nonwhites on the ticket, as President Obama was on the ballot in the two elections examined. The lack of increase in voter turnout in midterm elections, when many of the House candidates were among the highest-office nonwhite candidates, could be because House elections typically generate less excitement and can involve more costly information relative to presidential elections (DeSilver 2014). However, these are all suppositions. Large variations in candidate quality exist from election-toelection, and existing samples from real elections are unable to account for this. Neither are they able to measure which candidate is preferred, an important measure as previous studies have shown some forms of blowback when underrepresented candidates were running (Haider-Markel 2007; Washington 2006). Subsequently, an experiment that equalizes the desirability of each candidate is ideal to assess whether voter turnout and candidate preference are dependent on the race of the candidate and the respondent.

EXPECTATIONS Because strategic groups prefer representatives with policy preferences near their own, it is logical to argue that, when holding all other factors constant, an elected official who can provide descriptive representation will be preferred to an official who cannot, and individuals will expend effort to elect such a candidate. Furthermore, groups where widely understood contemporary race-based issues exist will likely have a heightened desire for racially similar representatives than groups without. From this, the following relationships are hypothesized.

Turnout Hypotheses: White Turnout Hypothesis: White respondents will report the highest likelihood of voting when two white candidates are on the ballot, followed by when one white candidate is on the ballot. White respondents will be least likely to turn out when two non-white candidates are on the ballot. Black Turnout Hypothesis: The presence of a Black candidate will result in Black individuals reporting a higher likelihood of turning out than when both candidates are white. Asian-American Turnout Hypothesis: The presence of an

An Experiment in Race and Voter Turnout in the United States

Asian-American candidate will result in Asian-American individuals reporting a higher likelihood of turning out than when both candidates are white. Hispanic Turnout Hypothesis: The presence of a Hispanic candidate will result in Hispanic individuals reporting a higher likelihood of turning out than when both candidates are white. Preference Hypothesis: The candidate that is the same race as an individual will be preferred to a candidate that is not the same race as the individual. Severity Hypothesis: Groups with distinct and notable racebased issues (African-Americans and to a lesser extent, Latino-Americans) will prefer the candidate of the same race at higher rates than individuals from groups without notable race-based issues (Asian-Americans and whites).

RESEARCH DESIGN AND METHODS Although an observational design can better measure voter turnout (especially because people tend to inflate their voting behavior in surveys due to social desirability bias (Holbrook and Krosnick 2009)), the logistics of an observational design are time-consuming relative to a survey due to candidate demographics and extraneous variables. Of primary importance is to consider the types of races that may encourage voter turnout due to prospective descriptive representation. It is known that constituencies are more aware of politicians in higher office (for example, the results of McKenzie et al. (2016) on local judges compared to Gibson and Caldeira (2009) on Supreme Court justices indicate that Americans have much more knowledge about national justices than local judges); therefore, desires for descriptive representation should be stronger for candidates running for higher offices. Candidates and election winners are predominantly white, especially for higher office (Hardy-Fanta, Sierra, Lien, Pinderhughes, and Davis. 2005; Lien 2015); therefore, gathering a statistically valid sample is challenging and could take years as so few races for higher office feature non-white candidates. Furthermore, an observational design utilizing real elections would need to attempt to separate the effects of additional factors where the needed variation may not exist such as party (women, as well as racial minorities, lean heavily Democratic (Pew Research Center 2016)) and candidate viability (Barreto, Villarreal, and Woods (2005) found higher voter turnout for co-ethnic viable candidates), in addition to the candidateâ&#x20AC;&#x2122;s descriptive factors. Because these barriers to conducting a good observational study exist, the use of a survey experiment, where such factors are easily controlled for, is ideal. An experimental design can control for candidate qualification levels, ideology, viability, office level, and backgrounds, and ensure that minority candidates appear more frequently on the ballot. A controlled experiment guarantees that all differences are the result of intentional treatment,

and determines whether individuals prefer descriptive representation while controlling for other candidate attributes. Respondents were recruited through Amazon.comâ&#x20AC;&#x2122;s Mechanical Turk (MTurk) from March 15 to 30, 2017.1 MTurk is a website that enables the public to earn money for completing online tasks, including participating in survey research. The usage of MTurk is advantageous due to its ability to gain high-quality data inexpensively and is regularly used in social science research (Buhrmester, Kwang, and Gosling 2011). Though the usage of MTurk as a recruitment tool may raise questions about the representativeness of the sample relative to the population, MTurk samples have been found to be often more representative than in-person convenience samples (Berinsky, Huber, and Lenz 2012), which are frequently used in political science research. Relative to the Cooperative Congressional Election Study, both old and young MTurk respondents tend to be more liberal (Huff and Tingley 2015), and relative to the American National Election Studies, the age of MTurk respondents are much younger (Berinsky, Huber, and Lenz 2012). Party and age of the MTurk sample used for this paper will be discussed further in the Group Demographics section below. To have the option of completing the survey, MTurk users needed to be over 18 years of age, to have at least a 90% approval rating, and to have registered for MTurk in the United States. To complete the survey, respondents needed to state that they were U.S. citizens, verify their age, and identify with one of the four main racial groups examined within the survey. Because MTurk respondents were overwhelmingly white, only non-white respondents were able to complete the survey in the second week of recruitment. Respondents were paid $0.20 to complete the one- to two-minute survey. Two hundred and eighty-one respondents completed the survey. After disqualifying respondents who failed the attention check (See item 9 in the Appendix), as well as excluding respondents who changed their race after initially not qualifying for the survey, 221 respondents remained in the experiment. Respondents were asked a short series of demographic questions to ensure that they qualified for the survey. Following the demographic questions, they were presented with the following question: In this scenario, Election Day is exactly as it is today, meaning most schools and offices are open. The two contenders for governor are [Candidate A] and [Candidate B]. [Candidate A] is dedicated to creating a strong economy. [Candidate B] promises to invest for the future. Given this information, on a scale of 0 (absolutely would not vote) to 10 (absolutely would vote), how likely would you be to vote in this election? Responses are on measured on an integer scale ranging from 0 to 10 a. Based on this information, which candidate would you prefer to vote for? Responses are [Candidate A] or [Candidate B]

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

Each non-white survey taker received the question with two hypothetical candidates: one with a white sounding name (Andrew Johnson) and another with a race-specific name (James Thompson (white), Jon Chen (Asian-American), Tyrone Jackson (African-American), or Diego Lopez (LatinoAmerican). For the race-varied candidate, respondents were given either the white name (James Thompson) or a racematched name (Jon Chen for an Asian-American respondent, for example). Pictures were included alongside these names to ensure that the race of the candidates was apparent, and are included in the Appendix. White respondents were given either two African-American names (Tyrone Jackson and Malik Jefferson), two white names (Andrew Johnson and James Thompson), or one of each (Andrew Johnson and Tyrone Jackson). Each name was placed in slot A and B at equal rates to ensure that order of the names presented or the difference between the vague promises of a good economy or investing for the future did not affect the outcome. Lists of most common race-specific last names were found at the Mongabay.com website which utilized 2000 Census data to determine common race-specific last names as well as the proportion of people with that last name that belonged to a particular racial group. Names were selected primarily based on these two criteria. For example, the Jefferson surname only belonged to 50,000 individuals, but over 75% of them were Black, which made it a good candidate for a “typical” Black name. Young’s (2016) piece also provided sway over which names to choose for the Black candidate. Two Black first names were chosen from a Freakonomics (Dubner and Conley 2013) podcast that highlights how having a “Black” first name is associated with one’s résumé receiving

less attention. Diego was chosen based on its position (fourthmost popular among males) on BabyCenter’s (2016) 100 Most Popular Hispanic Baby Names of 2011 list. (BabyCenter primarily has a U.S. user base). Chen was chosen as Chen is not only a very common last name among Asian-Americans but also because it is less associated with specific Asian ethnicities than other common names, such as Nguyen or Patel, thus signifying that the holder is Asian but not revealing a specific ethnicity. A generic English first name was chosen for the Asian candidate due to the tendency of many of those with Asian first names to also identify by an English name (Hsu 2009). Photos of supposed candidates appeared alongside the questions. See the Appendix for photos. These photos were chosen from the Chicago Face Database (Ma, Correll, and Wittenbrink 2015), a database of nearly 600 highquality pictures of faces in addition to ratings of these faces based on a host of objective criteria. The database included rankings of individuals based on age, perceived attractiveness, masculinity/femininity, predicted race, and other factors. Research in psychology has shown that masculinity is associated with being perceived as more competent in governmental tasks (Rosenwasser and Dean 1989), physically attractive hypothetical political candidates are thought to be more competent and trustworthy (Surawski and Ossoff 2006), and that in peacetime, “trustworthy faces” are valued for candidates (Little, Roberts, Jones, and DeBruine 2012), so it is imperative that these factors be similar for each of the presented candidates in this study Ratings on the perceived demographic information of each of the faces are elaborated on in Ma, Correll, and Wittenbrink (2015). In short, the researchers utilized a convenience sample of 1,087 respondents, and asked respondents Table 1: Summary Statistics on Respondent Demographics to estimate the age of a given face, to categorize each face’s gender and Proportion Party Education Age race (categories included Asian, Female Identification Black, Hispanic/Latino, White, and White other), and to rate each face based Johnson/Thompson 0.438+ 3.969 3.656 39.900 on a myriad of attributes including Jackson/Jefferson 0.594 3.250+ 3.688 37.375 attractiveness, trustworthiness, Johnson/Jackson 0.636+ 4.090+ 3.667 38.879 and masculinity. Each respondent only ranked 10-15 faces to prevent Latino respondent fatigue. The database Johnson/Thompson 0.400 3.800 3.733 35.400 grew from near-200 faces when the Johnson/Lopez 0.667 2.938 3.750 31.375 piece was initially published, and now contains data on 597 faces in Black 14 dimensions. All non-categorical Johnson/Thompson 0.750 3.625 3.710 40.25 ratings are based on a 1-7 Likert Johnson/Jackson 0.703 2.852 3.630 38.040 scale (7=extremely). Asian For this paper, an estimated Johnson/Thompson 0.520 3.895 3.895 32.840 age of at least 35 (range 36.8-56.4) was chosen as the survey asked Johnson/Chen .429 3.428 4.050 34.238 respondents to imagine that the +designates p<0.15. *designates p<.05 for within-race differences, two-sided Welch t-tests faces presented were gubernatorial

An Experiment in Race and Voter Turnout in the United States

candidates, and individuals younger than 35 are unlikely to be viable gubernatorial candidates. On the 1-7 scale, each face was chosen to be similar in attractiveness, masculinity, and trustworthiness. Of the chosen faces, the average attractiveness rating is around 3 (2.96-3.09), the average masculinity rating is near 5 (4.07-5.84), and the average trustworthiness rating is near 3.5 (2.89-4.3). Apart from the Latino face, respondents had a very high accuracy (at or near 100%) in determining the race associated with the face, and relative to other Latino faces within the database, the Latino face did receive high accuracy in race determination. Because the names provide additional cues as to the face’s race, this inaccuracy should not be of concern.

GROUP DEMOGRAPHICS

difference between the means of any of these proportions. Two pairs of differences in the white group (rows 2 and 3, and rows 2 and 4) approached significance and were notable at the 85% confidence level. The gender of all candidates in this experiment was male, thus it was unlikely that an abundance or lack of women would swing results in one direction or another. As previously noted, party identification is associated with varied turnout in elections with candidates of varied race (Washington 2006), but the differences within-race in this sample are all within one point on a seven-point scale, which is substantively tolerable. Indeed, Table 1 reports some visible differences between groups, but randomization succeeded in creating roughly equal values within each group. Relative to the 2010 Census’ measurement of Americans by race and ethnicity (U.S. Census Bureau 2016), the MTurk sample leaned young. Because this survey was not open to individuals under the age of 18, the Census-sourced age distribution was modified to include only individuals aged 18 and above. Figure 1 shows that in the sample, the 18-29 age group was significantly overrepresented and the 30-39 age group was slightly overrepresented. While this sample was slightly older on average than the MTurk sample discussed in Berinsky, Huber, and Lenz (2012), this sample was still much younger than face-to-face samples used for the American National Election Studies. Similar to Berinksy, Huber, and Lenz’s (2012) results, the MTurk respondent has slightly less than a college degree, while face-to-face samples yielded

The overall sample was 18.6% Asian-American, 23.5% AfricanAmerican, 14.03% Latino-American, and 43.89% white, had a mean education of slightly under a college degree, was on average 37 years old, was 56.6% male, and leaned slightly liberal. Within-group summary statistics are provided in Table 1. A respondent’s assignment to treatment groups was based on the respondent’s race, and then random assignment. For example, an Asian-American respondent would be randomly assigned to one of two groups comprised exclusively of AsianAmericans. As random assignment to treatment groups was employed, the averages of respondent characteristics, such as age, gender, and education, within each treatment group Figure 1: Population Distribution against Sample Age Distribution are controlled. As a result, the variation within responses and subsequent results should be a consequence of the treatment, and not because of group-specific attributes. To confirm the latter, summary statistics among samples within each racial group are reported in Table 1. (Explanations of party identification and education values can be found in the Appendix items 4 and 6.) Although several of these differences among the same racial groups appear somewhat large, such as the difference in party identification among LatinoAmericans in the Thompson vs. Lopez group (rows 6 and 7), Welch (unequal variance) t-tests with 95% confidence revealed no significant © Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

about 13 years of education. Consistent with a Pew Research Center (2015) analysis, Blacks and Hispanics tilted towards the Democratic Party, while whites were in the middle of the Democratic-Republican spectrum. Asians within the MTurk sample tended to be in the middle of the spectrum, only slightly more Democratic than whites, though the Pew Research Center (2015) measured a 42-point Democratic advantage among Asians.

RESULTS For the first two hypotheses, an unpaired difference of means with unequal variances t-test was employed to test for statistically significant differences in voting propensity and preference. To test the severity hypothesis, binomial proportion confidence intervals were constructed. The threshold for significance was a standard 95% confidence level, although approaching-significance confidence levels are also noted. Below, the results for the turnout hypothesis for white, Latino, Black, and Asian-American voters are discussed, followed by the results for the preference hypothesis for each racial group, and lastly the severity hypothesis.

Voter Turnout Hypothesis White Voters

Hypothesis: White voters will report the highest likelihood of voting when there are two white candidates (Johnson/Thompson), followed by when there is one white and one non-white (in this case, African-American) candidate (Johnson/Jackson), and lastly when there are two Black candidates (Jackson/Jefferson). As Table 2 reports, none of the groups were statistically significant from any of the other groups at the 95% confidence level. Therefore, the hypothesis that white voters are more inclined to vote when there is a white candidate in the election

cannot be verified. Though none of these levels were statistically significant from each other (possibly due to large variances in response), the ordering and differences are somewhat interesting. Voting likelihood was highest when there were two Black candidates or two white candidates (the difference of .03 on an 11-point scale is negligible). The differences between both the two white-only and Black-only candidate groups against the two-race-candidate group were both significant at the 85% level, suggesting that perhaps something about the mixed group made the election less worthwhile.

Latino-American Voters

Hypothesis: Latino-American voters will report higher likelihoods of voting when presented with a Latino-American candidate (Johnson/Lopez) relative to when presented with two white candidates (Johnson/Thompson). Like white voters (see Table 2), neither of these results were significant at the 95% confidence level, and the hypothesis that Latino-Americans report higher likelihoods of voting when presented with a Latino-American candidate cannot be accepted. In fact, similar to white voters (Table 2), usage of an 85% confidence level suggests a result contrary to what was hypothesized: that Latino-Americans were less likely to vote when presented with a Latino-American candidate. The LatinoAmerican cohort was the smallest within this study, and a larger future Latino-American sample would help indicate whether this apparent difference holds true.

African-American Voters

Hypothesis: African-American voters will report higher likelihoods of voting when presented with an African-American candidate (Johnson/Jackson), as opposed to being presented with two white candidates (Johnson/Thompson). The hypothesis that African-American voters are more likely to turn out when an African-American candidate is running for office also cannot be verified. Unlike white and Latino-American voters (Tables 2 and 3), the differences

Table 2: Voting Likelihood Differences for White voters White

Mean Voting Likelihood

Difference from Johnson/Thompson

Difference from Jackson/Jefferson

Difference from Johnson/Jackson

Johnson/Thompson

7.719

0.031

-0.932+

Jackson/Jefferson

7.750

-0.031

-0.900+

Johnson/Jackson

6.818

0.900+

0.932+

+designates p<0.15. *designates p<0.05

Table 3: Voting Likelihood Differences for Latino voters Latino

Mean Voting Likelihood

Difference from Johnson/ Thompson

Difference from Johnson/Lopez

Johnson/Thompson

7.600

-1.350+

Johnson/Lopez

6.250

1.350+

+designates p<0.15. *designates p<0.05

ÂŠ Pi Sigma Alpha 2018

An Experiment in Race and Voter Turnout in the United States Table 4: Voting Likelihood Differences for African-American voters African-American

Mean Voting Likelihood

Difference from Johnson/ Thompson

Difference from Johnson/Jackson

Johnson/Thompson

7.458

-0.351

Johnson/Jackson

7.107

0.351

+designates p<0.15. *designates p<0.05

Table 5: Voting Likelihood Differences for Asian-American voters Asian

Mean Voting Likelihood

Difference from Johnson/ Thompson

Difference from Johnson/Chen

Johnson/Thompson

7.578

-1.150+

Johnson/Chen

6.428

1.150+

+designates p<0.15. *designates p<0.05

between the Jackson and Thompson races are did not approach significance and were relatively small (0.35 on an 11-point scale). Even though this difference was small, it still occurred opposite the hypothesized direction: on average AfricanAmerican voters reported slightly higher likelihoods of voting when there were two white candidates in the race than when one white and one African-American candidate were up for election.

Asian-American Voters

Hypothesis: Asian-American voters will report a higher likelihood of voting when presented with an Asian-American candidate, as opposed to being presented with two white candidates. Similar to the above results in Tables 2-4, the hypothesis that Asian-American voters are more likely to turn out when there is an Asian-American in the race cannot be accepted. Like the white and Latino-American results (Tables 2 and 3), the mean difference between the all-white and mixed-race elections was relatively sizable, though it only approached significance. Again, the results seem to be contrary to what was expected: higher likelihoods of voting were reported when there were two white candidates in the race than when there was a white and an Asian-American candidate in the race. In sum, there is no evidence that individuals are more likely to turn out to vote when there is a candidate of the same

racial background running in the race when all other factors that are typically present in elections (party, campaigning, platforms, etc.) are absent.

Voter Preference Hypothesis

The second part of this analysis examines if respondents prefer the candidate of the same racial background. To test this, the results from the samples of individuals that were asked to choose from one white and one non-white candidate were tallied, and the differences in between vote shares were calculated. T-tests were conducted with the premise that the white and the non-white candidate were equally preferred. For African-American, Latino-American, and Asian-American respondents, the differences (the rightmost column in Table 6) were expected to be negative. For white respondents, the difference value was expected to be positive. The Asian-American, Latino-American, and white respondents had little difference in preference between the white and non-white candidates. The results for AsianAmericans were in the expected direction, though the results for Latino-Americans indicated no preference, and the results for whites were slightly in the unexpected direction. It can be concluded that Asian-Americans, Latino-Americans, and whites have little desire for a candidate of the same racial background when no substantive representation is offered.

Table 6: Candidate Preferences White Candidate Vote Share (%)

Non-White Candidate Vote Share (%)

Difference (White -Non-White) (%)

White (Johnson/Jackson)

48.48

51.52

-3.04

African-American (Johnson/ Jackson

25.0

75.0

-50**

Latino (Johnson/Lopez)

50.0

Asian (Johnson/Chen)

47.62

52.38

-4.76

+designates p<0.15. *designates p<0.05. **designates p<0.01

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

The African-American sample produced a very significant, 50% difference in between voting for Andrew Johnson (the white candidate) and Tyrone Jackson (the African-American candidate). It can be concluded that for African-Americans, race is not significant in the decision to turn out, but is significant in determining candidate preference when all attributes of the candidates but race are held equal.

SUMMARY AND CONCLUSION

Groups that are classified as “historically underrepresented” are still significantly underrepresented in government (Krogstad 2015), especially at national and top state levels. Previous research suggests that African-Americans are more involved and more satisfied in the democratic process when their representatives are also African-American (Brunell, Anderson, and Cremona 2008; Merolla, Sellers, and Fowler Severity Hypothesis 2013), but there is little information directly comparing different The results have suggested that race-based descriptive racial groups and their desires for descriptive representation. representation does not manifest itself on voter turnout, and This paper examines a survey administered via Amazon.com’s subsequently, the severity for voter turnout hypothesis cannot be Mechanical Turk to test three hypotheses. It was hypothesized correct. However, once one is at the polls, the severity hypothesis that individuals would be more interested in voting when for candidate preference may be suggested. African-Americans, there was an opportunity to obtain descriptive representation. the group with perhaps the most disadvantaged history as well Furthermore, it was hypothesized that individuals would as most disadvantaged current conditions, showed an incredible prefer to vote for the candidate that would provide descriptive preference for the African-American candidate over the white representation, and that these effects would be most pronounced among groups that were more disadvantaged. candidate, while other groups showed virtually no preference. After conducting a series of t-tests, no significant Figure 2 shows the vote share of the same-race candidates differences were found in reported voting likelihood when along with two-sided 95% confidence intervals. While Africanpresented with the opportunity for descriptive representation. American respondents preferred the African-American candidate In addition, no significant differences in candidate preference more than other-race respondents preferred their respective for the same-race candidate (as opposed to a different-race same-race candidate, the differences are not statistically candidate) were found for Latino-Americans, Asian-Americans, significant. This is likely due to the large variances within each or white respondents. African-American respondents greatly group due to small sample sizes. Only approximately half the respondents received a survey in which a same-race candidate was preferred the African-American candidate relative to the white pitted against a different-race candidate, and this half was further candidate. Although African-Americans greatly preferred the black candidate while each other group was relatively broken down by race. Subsequently, each of these separate ambivalent about their same-race candidate, no statistical groups averaged 25 respondents. Substantively, the differences in African-American preferences relative to other-race preferences difference was found in between the rate at which various racial do appear to be noteworthy, though it is impossible to pinpoint groups preferred their same-race candidate. whether such a result is genuine instead of a statistical fluke That only African-Americans significantly preferred the without an increased sample size. same-race candidate was perhaps the most interesting of the results. Junn and Masuoka (2008) Figure 2: Two-Sided 95% Confidence Intervals for Proportion Preferring the Same-Race Candidate find that the racial consciousness of Asian-Americans is weaker than that of African-Americans, so it is logical that these results manifest themselves as they do. Though no racial category has completely uniform upbringings and life experiences, the backgrounds of African-Americans share more commonalities than those of Latino-Americans and AsianAmericans. Nearly 80% of Blacks in the United States are descendants of slaves, while Latino and Asian immigration have been steadier and more recent. One of the effects of American slavery was that slaves were stripped of their heritage while Americans of Asian and 14

An Experiment in Race and Voter Turnout in the United States

Latino descent were better able to maintain their own cultural traditions when they immigrated to the United States. This results in fewer shared experiences within a group: any two Asian-Americans could very well differ by their generational status in the United States, linguistic background, and skin tone, making the relatability of a vague Asian-American candidate a poor one. While not as severe for Latinos due to more inter-country homogeneity and relatively less historical inter-country animosity in Latin America, the same issues are applicable. Little exists in the way of uniquely Asian-American issues, and uniquely Latino-American issues lie mostly in immigration, though this is nowhere near the top issue for Latino-Americans (Krogstad 2014). Recent news concerning police brutality and restrictive voting laws has brought AfricanAmerican identity to the forefront of discussion, highlighting how the African-American experience with the government is unlike that of others. Because of these factors, it is possible that African-Americans have formed a more cohesive identity and see a need for those in power to share similar life experiences than do Asian-Americans or Latino-Americans. This experiment tested voter behavior when just the race and names of the candidates were altered and the results (especially within the voter turnout hypothesis) are inconsistent with real-life election expectations; after all, it is a fact that African-American voter turnout soared when President Obama was running for office (Krogstad and Lopez 2017). In actual electoral situations, candidates hoping to capitalize on being a descriptive representative can make efforts to connect with the community that they are trying to court. While not at the forefront of his platform, it is impossible to deny that race (in both ways that supported and opposed Obama) played a part in Obamaâ&#x20AC;&#x2122;s election. In contrast, the experimental candidates were devoid of political affiliation, any actual policy proposals, and personality, and in no way attempted to court voters. They were descriptive representatives by virtue of being the same race but offered no evidence of substantive representation. Perhaps descriptive representation is less of a motivating factor when there is no actual evidence that the candidate shares ideology or party, but can become important when citizens see that there may be substantive representation to go along with the descriptive representation. A study that examines whether descriptive representation plays a role when levels of substantive representation vary, for example, one that studies the response elicited with an African-American Republican candidate among African-American voters, who lean heavily Democratic, would be able to explain how substantive and descriptive representation interact. The weaknesses in this analysis are largely concentrated on two issues. First, recruitment of non-white respondents was difficult, and small samples sizes brought about largerthan-desirable variances into the t-tests. When variances are large, even seemingly significant differences can be statistically insignificant. To mollify this issue in the future, the recruitment process could be lengthened and higher incentives should

be offered to further recruit more high-quality respondents. Second, this experiment was unable to prescribe the costs of voting onto the respondents. Respondents to election surveys frequently overestimate their propensity to show up and vote as well as wrongly reporting that they recently voted due to social desirability bias. While it is easy to click a button indicating a high likelihood of voting, getting time off from work, learning about all the candidates on the ballot, or standing in cold weather for hours can make the physical act of voting more costly. n

ABOUT THE AUTHOR Chenoa Yorgason graduated summa cum laude from the University of Pittsburgh in the spring of 2018 with a BS in Economics-Statistics and a BA in Political Science with honors. While at Pitt, she was a Brackenridge Fellow in the summer of 2016, where she conducted preliminary research for this paper. She has completed internships with Senator Bob Casey in Pittsburgh and Mercer in Washington, D.C.. In her sophomore, junior, and senior years, she worked as a math tutor and teaching assistant through the universityâ&#x20AC;&#x2122;s Math Assistance Center. She will join the Seattle health practice of Milliman, an actuarial consulting firm, as an Actuarial Analyst in August 2018.

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

APPENDIX Survey Question Text

Choices

1. Are you a United States Citizen? +

Yes/No

2. What is your age? +

10-100

3. What is your race? If you are multiracial, please choose the group that you most identify with.**

Asian Black Hispanic/Latino White Other

4. On a 7 point scale, from Strong Democrat (1) to Strong Republican (7), where do you stand?

1-7

5. What is your gender?

Male/Female/Other

6. What is your education level?

Less than high school

(Appendix note: less than high school graduate is coded as 1, graduate education is coded as 5.)

High school graduate Some College College Graduate Graduate School

7. You will be asked to evaluate an election under either the current system or under a system where Election Day were a federal holiday (meaning most schools and many businesses and offices would be closed).

0-10

In this scenario, election day is exactly as it is today, meaning most schools and offices are open. The two contenders for governor are [A] and [B]. [A] is dedicated to creating a strong economy. [B] promises to invest for the future. Given this information, on a scale of 0 (absolutely would not vote) to 10 (absolutely would vote), how likely would you be to vote in this election?** 8. Based on this information, which candidate would you prefer to vote for?

[A], [B]

9. Now imagine if Election Day were a federal holiday, meaning that schools and many places of employment are closed. There are multiple candidates in this race. Susan Green wants to improve transportation infrastructure. This is a verification test to ensure that you are reading the questions. Please select the candidate that appeared in the last question. Jane Garcia wants to overhaul education. Malik Jefferson wants to reform healthcare. Andrew Johnson promises to focus on economic issues.

Andrew Johnson Susan Green Malik Jefferson Jane Garcia (pictures were again included)

[Attention Check] + The candidate must be 18+ and a U.S. citizen to continue. Otherwise, the survey will end. **Candidates A and B will be chosen based on their response to the race question, and the same candidates will stay with the respondent for the entire survey. The selection criteria for A and B is: 1. For non-white respondents, 50% will receive two white candidates, Andrew Johnson and James Thompson. 50% will receive one white candidate, Andrew Johnson, and one racially matched candidate (Jon Chen for Asian-American, Tyrone Jackson for Black, Diego Lopez for Latino-American). Each Candidate will be candidate A 50% of the time, and each candidate will be candidate B 50% of the time. 16

An Experiment in Race and Voter Turnout in the United States Jon Chen

Tyrone Jackson

Malik Jefferson

Andrew Johnson

Diego Lopez

James Thompson

2. For white respondents, the process is a little different: White candidates have a 33.3% chance of encountering two white candidates (Andrew Johnson and James Thompson), 33.3% chance of a white and Black candidate (Andrew Johnson and Tyrone Jackson), and a 33.3% chance of two Black candidates (Tyrone Jackson and Malik Jefferson). Again, for each situation, each candidate will be candidate A 50% of the time and candidate B 50% of the time. These pictures were taken from the Chicago Face Database. Their target codes (to find these faces within the Chicago Face Database) are AM-215, LM-236, WM-223, WM-225, BM-244, and BM-245.

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

REFERENCES American National Election Studies (www.electionstudies.org) 2008 Direct Democracy Study [dataset]. Stanford University and the University of Michigan [producers and distributors], 2008.

Holbrook, Allyson L, and Jon A. Krosnick. 2009. “Social Desirability Bias in Voter Turnout Reports: Tests Using the Item Count Technique.” Public Opinion Quarterly 74(1): 37-67.

American National Election Studies (www.electionstudies.org) 2012 Direct Democracy Study [dataset]. Stanford University and the University of Michigan [producers and distributors], 2012.

Hsu, Huan. 2009. “The Name’s Du Xiao Hua, But Call Me Steve.” Slate, April 27. http://www.slate.com/articles/arts/ culturebox/2009/04/the_names_du_xiao_hua_but_call_me_steve. html (Accessed May 29, 2018)

Ansolabehere, Stephen. “Cooperative Congressional Election Study, common content, 2006.” Harvard Dataverse, V4 (2010). https:// cces.gov.harvard.edu/data Ansolabehere, Stephen. “Cooperative Congressional Election Study, common content, 2010.” Harvard Dataverse, V3 (2010). https:// cces.gov.harvard.edu/data Barreto, Matt A., Mario Villarreal, and Nathan D. Woods. 2005. “Metropolitan Latino Political Behavior: Voter Turnout and Candidate Preferences in Los Angeles.” Journal of Urban Behavior 27(1): 71-91. Berinsky, Adam J., Gregory A. Huber, and Gabriel S. Lenz. 2012. “Evaluating Online Labor Markets for Experimental Research: Amazon.com’s Mechanical Turk.” Political Analysis 20(3): 351-368. Brunell, Thomas L., Christopher J. Anderson, and Rachel K. Cremona. 2008. “Descriptive Representation, District Demography, and Attitudes toward Congress among African Americans.” Legislative Studies Quarterly 33(2): 223-244. Buhrmester, Michael, Tracy Kwang, and Samuel D. Gosling. 2011. “Amazon’s Mechanical Turk: A New Source of Inexpensive, Yet High-Quality Data?” Perspectives on Psychological Science 6(1): 3-5 Clark, Christopher J. 2014. “Collective Descriptive Representation and Black Voter Mobilization in 2008.” Political Behavior 36(2): 315-333. Cho, Wendy K. Tam. 1999. “Naturalization, Socialization, Participation: Immigrants and (Non-) Voting.” The Journal of Politics 61(4): 1140-1155.

Huff, Connor, and Dustin Tingley. 2015. ““Who Are These People?” Evaluating the Demographic Characteristics and Political Preferences of MTurk Survey Respondents.” Research & Politics 2(3): 1-12 Junn, Jane, and Natalie Masuoka. 2008. “Asian American Identity: Shared Racial Status and Political Context.” Perspectives on Politics 6(4): 729-740. Kanthak, Kristin, and George A. Krause. 2012. The Diversity Paradox: Political Parties, Legislatures, and the Organizational Foundations of Representation in America. New York: Oxford University Press. Krogstad, Jens Manuel. 2015. “114th Congress is Most Diverse Ever.” Pew Research Center, January 12. http://www.pewresearch.org/ fact-tank/2015/01/12/114th-congress-is-most-diverse-ever/ (Accessed May 29, 2018). Krogstad, Jens Manuel. 2014. “Top issue for Hispanics? Hint: It’s not immigration.” Pew Research Center, June 2. http://www. pewresearch.org/fact-tank/2014/06/02/top-issue-for-hispanicshint-its-not-immigration/ (Accessed May 29, 2018). Krogstad, Jens Manuel and Mark H. Lopez. “Black Voter Turnout Fell in 2016, Even as a Record Number of Americans Cast Ballots.” 2017. Pew Research Center¸ May 12. http://www.pewresearch.org/ fact-tank/2017/05/12/black-voter-turnout-fell-in-2016-even-as-arecord-number-of-americans-cast-ballots/ (Accessed May 29, 2018). Lien, Pei-te. 2015. “Reassessing Descriptive Representation by Women and Men of Color: New Evidence at the Subnational Level.” Urban Affairs Review 51(2): 239-262.

DeSilver, Drew. 2014. “Voter Turnout Always Drops Off For Midterm Elections, but Why?” Pew Research Center. http://www. pewresearch.org/fact-tank/2014/07/24/voter-turnout-alwaysdrops-off-for-midterm-elections-but-why/ (Accessed July 10, 2018)

Little, Anthony C., S. Craig Roberts, Benedict C. Jones, and Lisa M. DeBruine. 2012. “The Perception of Attractiveness and Trustworthiness in Male Faces Affects Hypothetical Voting Decisions Differently in Wartime and Peacetime Scenarios.” Quarterly Journal of Experimental Psychology 65(10): 2018-2032.

Dubner, Stephen J. and Dalton Conley. 2013. “How Much Does Your Name Matter?” Freakonomics, April 8. http://freakonomics. com/2013/04/08/how-much-does-your-name-matter-fulltranscript/ (Accessed May 29, 2018)

Ma, Debbie S., Joshua Correll, and Bernd Wittenbrink. 2015. “The Chicago Face Database: A Free Stimulus Set of Faces and Norming Data.” Behavior Research Methods 47(4): 1122-1135.

Gibson, James L., and Gregory Caldeira. 2009. “Knowing the Supreme Court? A Reconsideration of Public Ignorance of the High Court.” The Journal of Politics 71(2): 429-441 Haider-Markel, Donald P. 2007. “Representation and Backlash: The Positive and Negative Influence of Descriptive Representation.” Legislative Studies Quarterly 32(1): 107-33. Hardy-Fanta, Carol, Christine M. Sierra, Pei-te Lien, Dianne M. Pinderhughes, and Wartyna L. Davis. 2005. “Race, Gender, and Descriptive Representation: An Exploratory View of Multicultural Elected Leadership in the United States.” Presented at the Annual Meeting of the American Political Science Association, Washington, D.C. http://www.gmcl.org/pdf/APSA9-05-05.pdf (Accessed July 10, 2018). 18

Mansbridge, Jane. 1999. “Should Blacks Represent Blacks and Women Represent Women? A Contingent ‘Yes’” The Journal of Politics 61(3): 628-57. Mariani, Mack, Bryan W. Marshall, and A. Lanethea MathewsSchultz. 2015. “See Hillary Clinton, Nancy Pelosi, and Sarah Palin run? Party, Ideology, and the Influence of Female Role Models on Young Women.” Political Research Quarterly 68(4): 716-731. McDonald, Michael P. 2016. “Voter Turnout Demographics.” United States Elections Project http://www.electproject.org/home/voterturnout/demographics (Accessed May 9, 2018). McKenzie, Mark J., Cynthis R. Rugeley, Daniel B. Bailey, and Seth C. McKee. 2016. “Americans’ Knowledge of Their Local Judges.” Political Behavior 39(2): 259-227

An Experiment in Race and Voter Turnout in the United States Merolla, Jennifer L., Abbylin H. Sellers, and Derek J. Fowler. 2013. “Descriptive Representation, Political Efficacy, and African Americans in the 2008 Presidential Election.” Political Psychology 34(6): 863-875. Mongabay.com. 2016. “Most Common First Names and Last Names in the U.S.” https://names.mongabay.com/ (Accessed May 29, 2018). Pantoja, Adrian D., and Gary M. Segura. 2003. “Does Ethnicity Matter? Descriptive Representation in Legislatures and Political Alienation among Latinos.” Social Science Quarterly 84(2): 441460. Pew Research Center. 2015. “A Deep Dive Into Party Affiliation: Sharp Differences by Race, Gender, Generation, Education.” http://www.people-press.org/2015/04/07/a-deep-dive-into-partyaffiliation/ (Accessed May 29, 2018). Pew Research Center. 2016. “The Parties on the Eve of the 2016 Election: Two Coalitions, Moving Further Apart: Trends in Voter Party Identification 1992-2016.” http://www.people-press. org/2016/09/13/the-parties-on-the-eve-of-the-2016-election-twocoalitions-moving-further-apart/ (Accessed May 29, 2018).

table/1.0/en/PEP/2016/PEPALL5N/0100000US (Accessed May 29, 2018). Washington, Ebonya. 2006. “How Black Candidates Affect Voter Turnout.” The Quarterly Journal of Economics 121(3): 973-998. Weldon, S. Laurel. 2002. “Beyond Bodies: Institutional Sources of Representation for Women in Democratic Policymaking.” Journal of Politics 64(4): 1153-1174. Young, Damon. 2016. “What’s the “Blackest” Last Name? Washington, Jefferson, Jackson, or Jenkins.” The Root: Very Smart Brothas, August 22. https://verysmartbrothas.theroot.com/ whats-the-blackest-last-name-washington-jefferson-1822522570 (Accessed May 29, 2018).

NOTES 1 The survey received Institution Review Board certification at the University of Pittsburgh (#PRO17030095).

Phillips, Christian. 2018. “Wanting, and Weighting: White Women and Descriptive Representation in the 2016 Presidential Election.” Journal of Race, Ethnicity and Politics 3(1): 29-51. Reingold, Beth, and Jessica Harrell. 2010. “The Impact of Descriptive Representation on Women’s Political Engagement: Does Party Matter?” Political Research Quarterly 63(2): 280-294. Rocha, Rene R., Caroline J. Tolbert, Daniel C. Bowen, and Christopher J. Clark. 2010. “Race and Turnout: Does Descriptive Representation in State Legislatures Increase Minority Voting?” Political Research Quarterly 63(4): 890-907. Rosenthal, Cindy Simon. 1995. “The Role of Gender in Descriptive Representation.” Political Research Quarterly 48(3) 599-611. Rossenwasser, Shirley Miller, and Norma G. Dean. 1989. “Gender Role and Political Office: Effects of Perceived Masculinity/ Femininity of Candidate and Political Office.” Psychology of Women Quarterly 13(1): 77-85. Schwindt‐Bayer, Leslie A., and William Mishler. 2005. “An Integrated Model of Women’s Representation.” Journal of Politics 67 (2): 407-428. Surawski, Melissa K., and Elizabeth P. Ossoff. 2006. “The Effects of Physical and Vocal Attractiveness on Impression Formation of Politicians.” Current Psychology 25(1): 15-27. Tackett, Michael. 2017. “Women Line Up to Run for Office, Harnessing Their Outrage at Trump.” The New York Times, December 4. https://www.nytimes.com/2017/12/04/us/politics/ women-candidates-office.html (Accessed May 28, 2018). Tyson, Alec, and Shiva Maniam. 2016. “Behind Trump’s Victory: Divisions by Race, Gender, Education.” (2016, November 9). Pew Research Center. http://www.pewresearch.org/facttank/2016/11/09/behind-trumps-victory-divisions-by-race-gendereducation/ (Accessed June 1, 2018). U.S. Census Bureau. 2016. “Annual Estimate of the Resident Population by Sex, Single Year of Age, Race Alone or in Combination, and Hispanic Origin for the United States: April 2010 to July 1, 2016. 2016 Population Estimates.” Census 2010, using American FactFinder. https://factfinder.census.gov/bkmk/ © Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

The Determinants of Death Penalty Policy Christina Martin, Susquehanna University The use, and the debate, of the death penalty has been more prominent in the United States than in most other parts of the world. While more than two thirds of the world have abolitionist policies regarding the death penalty in law or in practice, the United States remains one of only 58 countries to retain the punishment of death. Currently, 31 states opt to have the death penalty (four of those currently have gubernatorial moratoriums), while 19 states have either abolished or overturned the death penalty. This begs the question: what macro-level state factors are the most important in determining whether a state has death penalty policy? Using linear regression, this paper demonstrates that a state’s political affiliation, GDP per capita, and education level are the most important factors in determining the existence of this punishment. Interestingly, education only becomes significant once a state’s population reaches the level of an advanced degree (one higher than Bachelor’s). Furthermore, this study finds that neither murder rate nor southern state classification has a significant impact, although other studies have noted these as the most important factors.

INTRODUCTION

ince the death penalty was reestablished in 1976 in the Supreme Court decision of Gregg v. Georgia, each state has been able to adopt or deny death penalty policy as a form of criminal punishment. As of October 2017, 31 states have the death penalty, although governors of four of these states have currently declared a moratorium on the use of this punishment. Nineteen states have abolished, overturned, or never adopted death penalty legislation1 . Because adoption of death penalty legislation is a state-by-state decision, it is interesting to examine what factors go into the existence of death penalty policy in states. Is it simply a decision of the state legislature? Or are there more factors that are outside legislative control that may also have an impact on determining whether a state has the death penalty? By examining the question of what macro-level factors most impact whether a state has the death penalty, one can draw the conclusion that education level, GDP per capita, and a state’s political affiliation are the most important variables. Through testing several different political and socioeconomic variables against the independent variable of a state’s death penalty policy, the data also negate the importance other studies have found in murder rates and southern state classifications. Further studies may seek to look at death penalty cases over time to understand more of a timeline approach by running multiple regression tests at different periods as the death penalty is used to determine whether frequency influences the variables. With the various issues to explore in political science and public policy, one might question why it is important to take into consideration research on the death penalty, a form of punishment that has been implemented in the United States since 1608 (Waksman 2012). The simple answer is that laws regarding the death penalty are constantly changing and public opinion on the topic is always shifting. Beginning in 20

2014, Tennessee reinstated the electric chair, moratoriums have been declared on the death penalty, Utah reinstated the use of the firing squad, states have abolished the death penalty, citizens have voted to reinstate the death penalty, and State Supreme Courts have ruled it as unconstitutional (ProCon 2018). As of May 30, there have been seven executions in the year of 2018. In 2017, 23 people were executed and 39 were sentenced to death (Death Penalty Information Center 2017). Although research centers, such as the Pew Research Center, have conducted public opinion polls reporting that support for the death penalty is at its lowest in more than four decades, executions continue, death sentences are given, and voters reinstate the punishment. It is clear to see this topic is of consistent importance and worthy of constant examination. A complete figure of executions per year since 1976 put together by the Death Penalty Information Center (2017) can be found in Figure 1.

BACKGROUND Prior to the Declaration of Independence, capital punishment was enforced in all the American colonies. From that time until present day, the usage, method, and reasoning for the death penalty have gone through transformations. In 1608, Captain George Kendall was the first recorded person to be sentenced to death in the British North American colonies. As a punishment for spying on the Spanish government, Kendall was executed by firing squad in Jamestown, the Colony of Virginia (Waksman 2012). Over time, the reason for which a person may be put to death has been clarified and narrowed. Crimes, such as horse stealing and kidnapping, were once punishable by death. Alabama has been one to frequently sentence its citizens to death for crimes other than murder- in1822, Thomas Davis was put to death for counterfeiting (Niles 1822) and in 1941, Frank Bass was sentenced to death for burglary (“Negro is First

The Determinants of Death Penalty Policy

Figure 1: U.S. Executions since 1976

Source: Death Penalty Information Center (2017) to...” 1941). For the last fifty years, homicide has essentially been the only crime for which someone has been sentenced to death. The last person who was put to death for a crime other than homicide was James N. Coburn, when he was executed by electric chair in Alabama for robbery on September 4, 1964 (Savage 2008). The death penalty is still listed on the books as a possible punishment in some states for crimes other than homicide, however, it is applied very rarely. In the Supreme Court case of Kennedy vs. Louisiana (2008), it was ruled unconstitutional to apply the death penalty for any other crime than one that resulted in death. Other states limited the use of the death penalty very early on in their statehood. Pennsylvania, for example, was the first state to put limits on crimes punishable by death. In 1682, William Penn, the founder of Pennsylvania, limited crimes punishable by death to murder (Woodham 2008). While some states have set limits on the death penalty, other states have gone a step further. Upon entering the Union, several states eradicated the death penalty as a form of punishment. The first state to completely abolish the death penalty was Rhode Island. In 1852, Rhode Island outlawed the death penalty for all crimes, including treason. Wisconsin followed one year later (Barak 2007). Some states, like Alaska and Hawaii, abolished the death penalty before gaining statehood. In 1972, the Supreme Court intervened. For four years, from 1972 to 1976, a suspension was placed on the use

of the death penalty because of the Supreme Court’s decision in Furman v. Georgia. In 1976, Gregg v. Georgia put an end to the suspension, but mandated that states that employ the death penalty must implement it in such a way that it does not violate the Eighth Amendment’s ban on cruel and unusual punishment. In this case, the Supreme Court also held that the death penalty did not violate the Fourteenth Amendment, which addressed citizenship rights and equal protection of the laws. After this new ruling, 37 states reinstated the death penalty under new statutes. The cases of Furman v. Georgia and Gregg v. Georgia were not the only ones that discussed the role of the Eighth Amendment in death penalty sentences. The Supreme Court cases of Ford v. Wainwright, Roper v. Simmons, and Atkins v. VA all further clarified when it is unconstitutional to employ the death penalty under the protection of the Eighth Amendment. As mentioned previously, Kennedy v. Louisiana was the clarifying case that noted the strictest lines for which the death penalty may be applied – only in the case of a crime that results in death. In recent years, the debate surrounding the death penalty has become increasingly divided. From 2010 to 2017, there has been a large shift in terms of public opinion data to support and oppose the death penalty. In 2010, 64% of Americans supported the death penalty for a person convicted of murder and 29% opposed it, with 6% of Americans having no opinion. The most recent study that was released in October of 2017

Pi Sigma Alpha Undergraduate Journal of Politics

showed that support for the death penalty lowered to 55% and opposition increased to 41%, with only 3% of Americans not having an opinion (Gallup 2017). As public opinion shifted, Americans became more split on the issue, reaching the biggest divide the United States has seen so far in terms of public opinion of the death penalty. As the support for the death penalty slowly diminishes, more and more states have decided to eradicate their use of the death penalty. In 2007, New Jersey became the first state to repeal the death penalty during what is called “the modern era” of capital punishment. The modern era refers to the use of the death penalty after it was reinstated in 1976 (Death Penalty Information Center 2017). After the New Jersey State Commission reported that the death penalty “is inconsistent with evolving standards of decency,” New Jersey lawmakers voted 44 to 36 to repeal the death penalty and replace it with life in prison without parole (Richburg 2007). The opponents of the death penalty in New Jersey hoped that their actions would prompt other states to do the same, which is exactly what happened. After New Jersey repealed the death penalty, five states followed, specifically New Mexico, Illinois, Maryland, Connecticut, and Delaware. It is also worthwhile to note that in 2004, the New York Court of Appeals held a portion of the death penalty law of the state to be unconstitutional. In 2007, the court ruled that this applied to the last remaining person on New York’s death row. Since this ruling, every legislative attempt to restore the death penalty statute had been voted down until a Nebraska election, where the death penalty was voted to be reinstated in 2016 by the use of a referendum (Death Penalty Information Center 2017). In 2015, Pennsylvania Governor, Tom Wolf, announced a moratorium for the death penalty. On August 13, 2015, Connecticut banned the death penalty, making it one of the most recent states to repeal the death penalty. The Connecticut Supreme Court declared that the death penalty “no longer comports with contemporary standards of decency and no longer serves any legitimate penological purpose,” a statement very similar to the one made in New Jersey (Berman 2015). Other recent states that have repealed or limited the death penalty by way of the state Supreme Court have noted similar reasons for making their decision. Currently, the District of Columbia and 19 other states do not have a death penalty statute (Death Penalty Information Center 2017). One of the focuses of the debate regarding the death penalty has centered on the possibility of an innocent person being executed because of a mistake in eyewitness testimony or another common error in identifying and sentencing suspects of crimes. Since 1989, there have been 351 postconviction DNA exonerations in the U.S. (Death Penalty Information Center 2017). The average length of time that an exoneree spent on death row was 14 years. The leading causes of wrongful convictions include eyewitness misidentification, improper forensic science, false confessions, and informants (The Innocence Project 2017). Despite the possibility of an innocent person being executed, the death penalty continues 22

to be upheld as a possible punishment for crime in many states. As of November 8, 2017, between the years of 1976 and 2017, there were a total of 1,465 executions (Death Penalty Information Center 2017). As of July 1, 2017, there were 2,817 inmates on death row (Death Penalty Information Center 2017).

LITERATURE REVIEW Death penalty policy is a subset of morality policy, which is policy that seeks to regulate behavior to validate basic values that are not universally held by all members of an electorate (Mooney and Lee 1999). Besides the death penalty, issues such as abortion, pornography, and gambling fit within the confines of what is considered morality policy. Mooney and Lee (2000) have found that U.S. state policy on morality issues is determined mainly by characteristics focused not on economic and demographic features, but on state values.

State Supreme Courts Ideology and the Death Penalty

The most widely discussed variable in the study of death penalty policy is the influence of political ideology, which are sets of beliefs, ideals, or principles that offer guidance on how society should work. Langer and Brace (2005) offer a discussion on state supreme courts through the role of their ideological composition in the passive process of shaping public policy. Legislators and governors are the most interested in the ideological composition of state supreme courts, as they will be less likely to enact liberal legislation, for example, if a state court is more conservative (and vice-versa). Legislators will thus reference the ideological predispositions of the court when creating law as policymakers wish to pass legislation that will not be overturned by the court. Langer and Brace (2005) find state supreme courts to be the most crucial determinants of death penalty policy. The role of state supreme courts in the policy process is to interpret and assess the constitutionality of laws, as well as to provide advisory opinions in some cases. In an interview of state legislators, 65% stated they would rewrite or postpone adoption of policy to avoid a negative response from the state supreme court (Langer and Brace 2005). These findings demonstrate the key role state supreme courts play in the legislative process. Langer and Brace (2005) found policymakers are more influenced by preferences of courts than preferences of the public.

Public Ideology and the Death Penalty

In a similar study, Brace and Boyea (2008) examined the role of the public’s ideology in its shaping of state supreme court ideology regarding the death penalty. This study assessed the difference between elected state supreme court judges and appointed judges. The impact of death penalty opinion is contingent on judicial elections: in states with high levels of

The Determinants of Death Penalty Policy

support for capital punishment, those that elect their judges will have courts with a lower propensity to reverse capital sentences (Brace and Boyea 2008). Public opinion about the death penalty, in the case of states with elected judges, is significantly correlated with judicial ideology.

Legislator Ideology and Death Penalty

Just as public opinion affects elected supreme court justices in their decision-making on the death penalty, so too does it affect the decision-making of legislators. Mooney and Lee (2000) note the incentive of policymakers to listen carefully to their constituents as this can impact their chances at election or reelection, and additionally serve as a topic for a potential stump issue. When considering the death penalty in the eyes of a liberal versus a conservative, Mooney and Lee (2000), as well as Nice (1992), agree that liberals will tend to go against death penalty policies, while conservatives will push for such policy. Nice (1992) argues ideology is a great influence in state death penalty decisions, and therefore the ideological composition of a state will be a significant determinant in such policy. States that are shown to have a supportive orientation toward the death penalty will tend to have relatively conservative electorates (Nice 1992). Langer and Brace (2005) also found that the policymaker’s ideology exerts a statistically significant effect on the enactment of death penalty legislation, and that this effect is heightened when there is unified party control of the legislature. Both policymaker ideology and unified government control had a significance of .05 on death penalty legislation (Langer and Brace 2005). However, in the study done by Mooney and Lee (2000), mass public liberalism and elected official liberalism were not significant in either the pre- or postFurman v. Georgia period as a determinant of death penalty reform. Although party activist liberalism was not significant in the pre-Furman period, it was found to be significant in the post-Furman period at the .01 level when considering death penalty legislation (Mooney and Lee 2000). In public opinion studies, political ideology is often linked to attitudes toward the death penalty. A general shift occurs in public opinion in the pre- and post-Furman periods. The pre-Furman period was classified as a contentious time, as national support for capital punishment averaged 49.1% and opposition averaged 39.1%. After Furman v. Georgia (1972), the United States moved to a period of consensus, as 64.3% supported the death penalty and 29.4% opposed it from 1972 to 1982 (Mooney and Lee 2000, 226). Mooney and Lee (2000) found that public support for the death penalty was significant at a .01 level in the pre-Furman period but was not found to be significant in the post-Furman period, with public opinion being the only variable found to be significant in the preFurman period (2000, 233). This means that public opinion of the death penalty was found to be the most consistently significant figure in their study. In democracies that have thus far abolished the death penalty, legislators frequently have

gone against public opinion to make this decision, as Mooney and Lee (2000) found that public opinion was actually in favor of the death penalty at most times when the legislation was abandoned. Some assert that state elected officials are considered too accountable to the electorate to take such bold action against public opinion. A specific way to use public opinion to determine support for or against legislation is through a referendum system. For example, Morgan and Meier (1980) find that states with a referendum system will see a plethora of public controversies about moral issues make their way onto the ballot.

Crime Rate and the Death Penalty

The next variable that is often discussed at great length in relation to adoption of the death penalty is the crime rate, more specifically focusing on murder. Nice (1992) stresses that when killings are a common occurrence in a state, the public will perceive the death penalty as necessary for the preservation of social order. Although the deterrent ability of the death penalty is shown to be inconsistent, the death penalty at least provides public officials with a symbolic response to heinous crimes (Martin 2016). For this reason, high rates of murder and nonnegligent homicides, especially seen at an increasing rate, are likely to increase support for capital punishment (Nice 1992). Tyler and Weber (1982) agree with Nice’s (1992) assertions, stating that support for the death penalty is the result of a desire for a lower crime rate. A belief that harsher punishments will lessen crime will result in a support for these punishments. Vidmar and Ellsworth (1973) found the judgment that crime was increasing directly related to death penalty support. In Langer and Brace’s (2005) study, the murder rate significantly predicted support for the death penalty, as state policymakers were 20% more likely to enact the death penalty when the murder rate increased by a unit.

Socioeconomic Status and the Death Penalty

When examining predictors of support for the death penalty, the socioeconomic context is also taken into consideration by scholars. A disagreement exists regarding whether wealth will relate to a support for the death penalty. Nice (1992) argues that wealthier states will be more receptive to capital punishment, as public opinion shows wealthier people are more supportive of the death penalty, and killers of wealthy people are especially likely to receive such punishment. However, median-income level was not significant when he performed his study (see also Tyler and Weber 1982). Morgan and Meier (1980) disagree with the view that income is linked to death penalty support and state that because upper-income groups are more sympathetic to social and political change, they will be less likely to impose conservative moral principles on others. They found median family income, when combined with other socioeconomic factors, particularly median years of education, percentage of population living in poverty, and percentage employed in white collar occupations, is significant.

Pi Sigma Alpha Undergraduate Journal of Politics

Vidmar and Ellsworth (1972, 1253) had similar findings, and state that generally, “people who support the death penalty tend to be older, less educated, male, wealthier, white, and from urban areas.” Through the examination of data that surround the death penalty, there are several studies that examine how race influences support for the death penalty. According to the American Civil Liberites Union (2017) “the color of a defendant and victim’s skin plays a crucial and unacceptable role in deciding who receives the death penalty in America. People of color have accounted for a disproportionate 43% of total executions since 1976 and 55% of those currently awaiting execution.” Throughout the year, there have been several court cases that have specifically examined the role of racial bias in the criminal justice system, including Batson v. Kentucky, Miller-El v. Dretke, and McCleskey v. Kemp. These focus on removing or excluding members of the jury based on their race and the possible impact that this could have on a decision. Other organizations have published statements that also address the influence of nonwhite percentages of a state in the impact of state death penalty policy, including the National Coalition to Abolish the Death Penalty and Amnesty International USA. Tyler and Weber (1982) found that socioeconomic variables, such as religion and race, did not appear to significantly influence support for the death penalty. Morgan and Meier’s (1980) findings refute this, noting considerable support for the basic proposition that religion appears to be significant in terms of referendum voting, with an effect on moral votes in the same way as socioeconomic characteristics. Morgan and Meier (1980) note the propensity of religious groups to use the political process to further their own moral beliefs and Mooney and Lee (2000) see religious affiliation as an important factor in measuring public values. Kohut, Green, Keeter, and Toth (2000) found the views of black Christians reflect the mixed liberal/conservative agenda of their clergy, as they express less support for the death penalty. Overall, 18% of those interviewed stated that religion is the most influential factor influencing their opinion toward the death penalty and 21% said education is the most influential factor (Kohut et al. 2000). The inconsistent findings among these studies can be attributed to the fact that Vidmar and Ellsworth (1973) examine public opinion data at the individual level while other scholars, such as Morgan and Meier (1982), examine aggregatelevel data. While these data can still be compared and used to construe further differences in death penalty variables, it is important to note that these disparities may be attributed to the examination of both aggregate versus individual level data. The geographical location of states is used as a variable for death penalty policy, as well as urbanization level. Langer and Brace (2005) use southern states to control for the distinctive policies of a region, as southern states are known for implementing the death penalty. Since 1976, more than half of 24

all state executions have occurred in a southern state. In their study, southern states were found to have a significance of .05 in determining death penalty policy (Langer and Brace 2005). Although Mooney and Lee (2000) found no such significance in the pre-Furman period, they found southern states to significantly influence support for death penalty policy in the post-Furman period. Geographic disparities of executions are clear as an unseemly majority of all executions in the United States have been in a southern state.

THEORY Of the nine variables that are discussed in this paper and presented with hypotheses, eight predictors have significant research that has been conducted to determine their relationship with the death penalty. While some variables are highly contentious among scholars in the way that they interact with the death penalty (e.g., nonwhite percentage, education level, and gross domestic product per capita), others, like the southern state classification and the Republican percentage, have been agreed upon by most scholars. I have hypothesized that each variable will be correlated with the existence of the death penalty that is largely consistent with the prior body of research, apart from the variable of murder rates. In my previous research examining deterrence, I have found there to be a negative relationship between murder rates and the death penalty, and thus expected this to hold true in this study (Martin 2016). I have also included a variable representing the existence of a referendum system in each state. The latter lacks extensive research, but I assert that states with a referendum system will be more likely to have the death penalty.

Hypothesis section: Hypothesis one: The higher the murder rate of a state, the less likely it will have death penalty policy. Hypothesis two: The higher the nonwhite percentage of a state, the more likely it will have the death penalty. Hypothesis three: The higher the percentage of a state that has a high school diploma, the less likely it will have the death penalty. Hypothesis four: The higher the percentage of a state that has a Bachelor’s degree, the less likely it will have the death penalty. Hypothesis five: The higher the percentage of a state that has an advanced degree, the less likely it will have the death penalty. Hypothesis six: The higher the gross domestic product per capita of a state, the less likely it will have the death penalty.

The Determinants of Death Penalty Policy

Hypothesis seven: If a state has a referendum system, it will be more likely to have the death penalty. Hypothesis eight: If the state is classified as a southern state, it will be more likely to have the death penalty. Hypothesis nine: The higher the percentage of Republican voters in a state, the more likely it will have the death penalty.

METHODOLOGY Dependent Variable

This research seeks to address the following research question: What macro-level state factors are the most important in determining whether a state has death penalty policy? To empirically test this question, a binary logistic regression is estimated with a series of state-level socioeconomic and political independent variables against the dependent variable - death penalty policy. Death penalty policy is defined as whether an individual state has death penalty policy that can be implemented for criminal sentences. This variable does not consider the frequency with which the death penalty is employed. For example, Nebraska has not carried out an execution since 1997, but because there remains the ability to sentence a person to death, it is included as a state that has the death penalty. Additionally, the four states – Colorado, Oregon, Pennsylvania, and Washington - that currently have a gubernatorial moratorium on the death penalty are also included as death penalty states using this definition. Information regarding which states are classified as death penalty states is derived from the Death Penalty Information Center (2018). States with the death penalty were coded as a 2, while states without the death penalty were coded as a 1.

Independent Variables

To measure a state’s murder rate, it is important to define what type of murder will be included in this percentage. Examining homicides, the only definition that will be included is what is considered capital homicide. In most states, this is defined as first-degree murder. Some states, such as Ohio, identify this as aggravated murder2 . Capital murder in the U.S. usually involved one or more of the following factors: a victim is a police officer, firefighter, paramedic, or child; the murder is committed during commission of another violent felony, such as burglary, sexual assault or kidnapping, multiple murders are committed pursuant to one another, and murderfor-hire. The statistics that have been gathered in the Uniform Crime Reporting article published by the Federal Bureau of Investigation (2016) are used to measure the murder rate for each state. Including the murder rate as a variable in this study is important because the murder rate is most typically

used to justify deterrence. The deterrent factor of the death penalty is one of the largest reasons that supporters justify the punishment. As the murder rate and the deterrent factor are often viewed hand-in-hand in death penalty cases, it is important for this to be included as an independent variable. To assess the socioeconomic level of each state, three independent are examined: race, education, and GDP. The percentage of nonwhites in the state measure racial makeup. Nonwhite residents of a state include those who classify as black, Hispanic, Asian, American Indian/Alaska native, or native Hawaiian/other Pacific Islander. To create the percentage of a state that identifies as nonwhite, data were pulled from the Kaiser Family Foundation’s Population Distribution by Race/ Ethnicity (2016). Adding together the columns for all races other than white, including the column that included two or more races, each state was given a percentage of nonwhite residents. Because of the racial bias that has been asserted by many organizations and studies, it is important to include this variable to evaluate the possible influence it may have on death penalty legislation in a state. Racial bias extends beyond just white and black defendants, which is why the entirety of the nonwhite population of a state is included as a variable. Education level is broken down into three categories that include the percentage of a state with high school diplomas, the percentage of a state with Bachelor’s degrees, and the percentage of a state with an advanced degree. The varying degrees of education were broken down into categories to provide a clearer picture of whether the specific level of education in a state (from high school diplomas to advanced degrees) impacts its death penalty policy, or if it is simply education as a whole that has an influence. Information regarding state-by-state education levels was pulled from data published by the United States Census Bureau in the 2011-2015 American Community Survey 5-Year Estimates. Education variables were included because of the many scholars that asserted that with more education, the more “liberallyminded” a person will become, in turn making them more likely to oppose the death penalty (National Public Radio 2016). Transitioning to a macro-level view of education of the state, the more citizens of a state that are educated to higher levels will in turn represent a more “liberally-minded” view from the state as a whole. The final socioeconomic variable is GDP or the gross domestic product per capita of a state. This measure was chosen over gross domestic product in millions of current dollars in order to provide a clearer understanding of the monetary situation in a state. Data for this variable is derived from the Bureau of Economic Analysis of the United States Department of Commerce for the year of 2016. GDP is an influential and important variable as the wealthier a state is, the more resources they have at their disposal to spend on state-sponsored education. This could, in turn, lead to influence the prior variable, as well as the overall systems that are put in place by the state.

Pi Sigma Alpha Undergraduate Journal of Politics

The first of the three political variables is the existence of a referendum system in a state. A state with a referendum system is defined as a state that has in place a referendum system where issues can be placed on the ballot for citizens to vote. Using Ballotpediaâ&#x20AC;&#x2122;s (2017) list of states with initiative or referendum systems, the existence of a referendum system is coded as a yes or no answer. States with a referendum system were coded as a 2 and states without a referendum system were coded as a 1. As previously discussed, referendum votes are a direct way in which citizens can show their opinion of the death penalty in terms of measurable public opinion. Therefore, it is expected that states with a referendum system will be more likely to have the death penalty than states without the referendum system. The next political variable is whether a state is classified as being in the southern region. Although there are a variety of ways that states have been classified into regions within the United States, this study will use the information that is presented by the U.S. Census to identify a state as southern. In the U.S. Census, there are three breakdowns of the southern classification which include the South Atlantic, East South Central, and West South Central. This study does not examine each sub-region separately, but rather looks at the classification of southern states as a whole. States that were classified as southern were coded as a 2, while states that were not considered southern were coded as a 1. Southern states are typically stereotyped to be the ones to employ the death penalty most often. Although stereotyped, southern states have historically employed the death penalty the most frequently. The final political variable examines the political leaning of a state, specifically to the Republican party. Although there are countless ways to measure the political affiliation of a state, this study used previous U.S. presidential elections to identify the percentage of a state that is Republican. As the election of Donald Trump was a unique presidential election, to control for this, the results of the previous three presidential elections (2008, 2012, 2016) were used and averaged together in order to provide a more legitimate percentage for Republicans in a state. The information that was used to calculate an average for this variable came from the New York Times (2008 and 2016) and NBC (2012)3 . This variable was also included to capture the public opinion of a state, as partisanship is often closely tied to individual opinions of the death penalty. States that lean Republican are often more likely to be death penalty states while states that lean Democrat are more often to be non-death penalty states.

FINDINGS Examining the results of the regression tests in Table 1, several conclusions can be drawn. First, there are many independent variables that do not have any significant impact on the presence of death penalty policy in a state when run in a combined test, including the murder rate of a state, the existence of a referendum system, the nonwhite percentage of a 26

state, southern state classification, and the percentage of a state with high school diplomas (Note: The influence of high school diplomas is not reported in Table 1). When an isolated test is estimated (Table 2), the murder rate, however, does become significant. Second, turning to the significant variables, one wonders why some variables, particularly those that have been studied so extensively in other research, are not statistically significant in this analyses. The combined tests illustrate how the different variables relate to each other and the dependent variable of the death penalty, while the isolated tests allow a closer examination of the variables on their own in their relationship to the death penalty. The following section provide a more in-depth look at the differences between Tables 1 and 2 as well as the differences between the significant variables. In Table 1, the first model (Model 1) that was run in the combined section sought to examine the influence of the Republican affiliation of a state. Thus, this model did not include the variables of education levels or GDP per capita. The Republican affiliation variable was the only one that was significant, and was significant at the .01 level, along with the constant that was significant at the .001 level. It also had a positive correlation, which is consistent with hypothesis that the more Republican voters in a state, the more likely a state will be to have the death penalty (Hypothesis 9). Although none of the other variables were significant in this model, the murder rate was correlated negatively and the existence of a referendum system, the nonwhite percentage of a state, and southern state classification were positively correlated (Hypotheses 7, 2, and 8). The second model in Table 1 examines the variables of murder rate, the existence of a referendum system, the percentage of nonwhite residents, and the southern state classification more closely (Hypotheses 1, 7, 2, and 8). However, nothing in Model 2 was significant other than the constant, which was significant at the .01 level. Although not significant, all the variables in this model were positively correlated. This is a change from Model 1 in Table 1, where the murder rate was negatively correlated (Hypothesis 1). The third model in Table 1 adds the Republican variable again and eliminate the southern variable in order to discover whether Republican affiliation and southern state classification were too closely related in model one, creating the large significance in the Republican variable (Hypotheses 9 and 8). However, eliminating the southern variable only made the Republican variable more significant, this time at the .001 level, along with the constant (Hypotheses 8 and 9). As it was in Model 1, the Republican variable was positively correlated in this model, continuing to confirm the hypothesis that as more Republican voters are in the state, the more likely it will have the death penalty (Hypothesis 9). The variables for the existence of a referendum system and the nonwhite percentage continued to be positively correlated, and the variable for murder rate switched back to its negative correlation (Hypotheses 7, 2, and 1).

ÂŠ Pi Sigma Alpha 2018

The Determinants of Death Penalty Policy Table 1: Logistic Regression of Referendum Status, Socioeconomic Variables, Political Makeup on Death Penalty Policy Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

Model 7

Murder Rate

-.028 (.220)

.072 (.201)

-.024 (.206)

.065 (.190)

.033 (.195)

-.021 (.223)

.056 (.189)

Nonwhite

.047

.014 (.022)

.047

.004 (.020)

-.001 (.020)

.052 (.031)

.002 (.020)

(.028)

Bachelor’s Degree

-.121 (.080)

Advanced Degree

-.314 (.145)*

.114 (.205)

GDP

.000 (.000)*

Referendum

.668

.768 (.777)

.650

.922 (.737)

.975 (.712)

.663 (.833)

1.045 (.737)

(.833) Southern region

(.770)

.064

.531 (1.08)

1.32 (1.02)

1.24 (1.00)

-.163 (1.199)

1.09 (1.04)

(1.13) Republican

.201 (.065)**

.202 (.063)***

.233 (.091)**

Constant

-11.364 (3.480)***

-5.386 (2.091)**

-11.335 (3.439)***

.425 (2.355)

.864 (3.191)

-14.034 (6.101)*

1.898

.496

.388

.495

.313

.310

.501

.333

(2.966)

* = Significant at .05 level ** = Significant at .01 level *** = Significant at .001 level Note: For simplicity, the influence of the percentage of a state with high school diplomas, which is not statically significant, is not reported in Table 1. Table 2: Isolated Tests of Selected Variables Isolated 1

Isolated 2

Isolated 3

Isolated 4

Isolated 5

Isolated 6

Murder Rate

.281 (.143)*

High School

-.226 (.111)*

Bachelor’s Degree

-.164 (.069)*

Advanced Degree

-.356 (.138)**

Republican

.159 (.047)***

GDP

.000 (.000)**

Constant

-.831 (.711)

20.524 (9.863)*

5.298 (2.072)*

4.335 (1.531)**

-7.098 (2.206)***

5.902 (2.037)**

* = Significant at .05 level ** = Significant at .01 level *** = Significant at .001 level

Pi Sigma Alpha Undergraduate Journal of Politics

The fourth model in Table 1 focuses on the variable of an advanced degree while eliminating the Republican variable that had already proved to be significant (Hypotheses 5 and 9). As in the previous three models, the first four variables were not significant. The variable of an advanced degree was significant at the .05 level and correlated negatively, which correctly aligned with its hypothesis- as a larger percentage of a state has advanced degrees, it will be less likely to have the death penalty (Hypothesis 5). The variables for referendum systems, nonwhite percentage, and southern classification continued to be positive, while the variable for the murder rate also switched back to a positive correlation (Hypotheses 7, 2, 8, and 1). The fifth model in Table 1 attempted to examine another measure of education, which was the percentage of those in a state with a Bachelor’s degree (Hypothesis 4). To control for multicollinearity, this was not run with the variable of the percentage of those in a state with an advanced degree (Hypothesis 5). This was the first model where no variables were significant. There were some changes in correlation among the variables in Model 5. The murder rate, referendum system, and southern state classification carried through to this model as positively correlated, while the nonwhite percentage of a state became negatively correlated for the first time (Hypotheses 1, 7, and 2). Although it was not significant, the variable for Bachelor’s degrees in a state was also negative, which would have held to the hypothesis (Hypothesis 4). The influence of the percentage of a state with high school diplomas was not statistically significant (Hypothesis 3) and is not reported in Table 1. The sixth model in Table 1 tests the two variables that have proved to be significant thus far (Republican affiliation and education level) into one model, in order to see if they would both still hold their significance when run together (Hypotheses 9 and 5). The result was a continuation in the significance in Republican affiliation and the elimination of the significance in the percentage of those with an advanced degree in a state (Hypotheses 9 and 5). As it was in other models, the Republican variable continued to be positively correlated (Hypothesis 9). There was quite a heavy shift in the correlation of the other variables. The variables for murder rate and southern state classification were negatively correlated in this model (Hypotheses 1 and 8). This was the third time murder rate was negatively correlated and the first time southern state classification was. The existence of a referendum system, the nonwhite percentage of a state, and the percentage of those with an advanced degree were all positively correlated (Hypotheses 7, 2, and 5). This was a change in the variable for advanced degrees, as it was negatively correlated in the model where it proved to be significant. The final model (Model 7) added the variable of GDP per capita for the first time (Hypothesis 6). It was not run with any of the education level variables, as that might present a problem with multicollinearity. Additionally, it was not run with the Republican variable in order to control for the fact 28

that the Republican variable seemed to be overwhelmingly significant in each of the previous models. In this, GDP per capita was significant at the .05 level with no correlation to the constant. Additionally, all four of the other variables were positively correlated, which presented another switch for the variable of the murder rate, as well as a second switch for both the nonwhite and southern variables (Hypotheses 1, 2, and 8). The only variable to be positively correlated across all seven models was the existence of a referendum system (Hypothesis 7). In order to more deeply determine the relationship between the individual independent variables and the dependent variable, each variable was run in a single test with the constant (See Table 2). The variables of the existence of a referendum system, the nonwhite percentage of a state, and the southern state classification were still not significant when run individually with the constant (Hypotheses 7, 2, and 8)4 . Beginning with the first variable that is significant with the constant, the murder rate was significant at the .05 level and positively correlated (Hypothesis 1). Next, the percentage of those with a high school diploma were also examined in this model, and had a significance of .05 and negatively correlated (Hypothesis 3). Similarly, the percentage of those with a Bachelor’s degree was also significant at the .05 level and negatively correlated (Hypothesis 4). The variable of the percentage of those with an advanced degree continued in the pattern of negative correlation, but had an increased significance at the .01 level (Hypothesis 5). The most significant variable in this test, as well as in the combined test, was the percentage of Republican voters in a state (Hypothesis 9). Here, it was positively correlated and significant at the .001 level. Finally, the GDP per capita of a state had no correlation and was significant at the .01 level (Hypothesis 6). The constant was also significant at varying ranges for all variables other than the murder rate.

DISCUSSION AND CONCLUSION Throughout the various models and tests that were run, it is clear that education level, political affiliation, and GDP per capita are the factors that most influence whether a state has death penalty policy. In terms of education, there is an increasing significance seen as a state increases in its level of education from a high school diploma, to a Bachelor’s degree, and to an advanced degree. As each of these variables are negatively correlated when they appear to be significant, it is possible to say with some degree of confidence that as a state becomes more educated, it is less likely to have the death penalty. This asks the question: what does education have to do with a state’s likelihood to not have the death penalty? Could this possibly be connected to the increased tolerance and open-mindedness that may come from meeting a variety of individuals and learning a variety of information as one progresses through high school into an advanced curriculum?

The Determinants of Death Penalty Policy

Or perhaps, increased critical thinking skills that develop throughout the education timeline? Additionally, should those who oppose the death penalty also seek to increase education levels in states that have lower education percentages in order to hope for a future change in death penalty policy? It is not surprising that political affiliation was the most significant variable in this study. This aligns with the overwhelming support from literature that agrees that political affiliation is one of the most important variables in considering the death penalty. Additionally, Republicans have been seen to hold accepting views of the death penalty in public opinion data, which further pushes that the positive correlation of Republican voters and the death penalty is not a surprise. Finally, it is also not entirely a surprise that the variable for gross domestic product per capita follows education level in both significance and correlation. If access to education increases as one has more money, it is expected that these two variables would prove to be similar in their results in different tests and models. Future research should perhaps examine the death penalty legislation in a different way. For example, it is possible that this test could be skewed by the fact that those states in gubernatorial moratoriums were included with states that have the death penalty, along with states that have not employed the death penalty in several years and states that have only carried out the death penalty once or twice in their historical use. States that have not employed the death penalty in several years could technically be classified as having a form of de facto moratorium on the death penalty put in place, rather than de jure moratoriums, such as Pennsylvania. It is important to remember states that have both de jure moratoriums and de facto moratoriums were included as death penalty states in this model. In future studies, as stated before, it would be interesting to run the same test with these states included as non-death penalty states to see if a difference occurs. Another subsequent study might run several tests considering each of the previous thoughts, separating states in different ways in order to examine whether frequency has an influence on the variables. Additionally, another study that examines the death penalty policy at different times in its legislative life may also provide interesting results that differ from this study. For example, going back fifteen years when the death penalty was legal in several other states and running a test with data from that time and comparing it to now to see change over time may provide a clearer reason and understanding as to why some variables are significant and why some are not. This could be an extremely interesting addition to the study as much of the legislation that impacts the death penalty in many states has remained unchanged in the recent years. A study that examines the death penalty over time or during the time surrounding the court case that reinstated the ability to use the death penalty could allow for new data to surface. Overall, this study provides interesting results as it does not find two of the most cited variables of scholars to be

significant in combined tests- the murder rate and the southern state classification. Additionally, as this study finds education level, political affiliation, and gross domestic product per capita to be significant, this could provide for future analysis of the death penalty in its relation to each of these variables, as well as the study of the relationship between each of these variables themselves. Another variable that could be added to the set in future studies could be a form of public opinion data. In this study, public opinion is mostly shown through variable of party affiliation, as partisanship has been shown to be very attached to opinions of the death penalty. In future studies, an additional variable could be included if there were a way to find public opinion data for each individual state, allowing this study to shift from more of a macro-level examination to a micro-level examination. To summarize, these findings are significant because they contribute a new layer of conversation and data to the ever-growing collection of information and studies that surrounds the death penalty. Especially in terms of the insignificance of variables like murder rate and southern states that are commonly thought to be significant in death penalty debates, it could call for a shift in variables that are most typically studied and stereotyped to be important influences of the death penalty. As referendum items become more common on ballots in elections, this issue retains its significance as the topic of the death penalty could one day be placed on the ballot of a citizen of any state. n

ABOUT THE AUTHOR Christina Martin graduated from Susquehanna University in May of 2018 with a triple major in Political Science, Public Policy, and Spanish. At Susquehanna, Christina served as the President of the Student Government Association, the Finance Vice President of her sorority, Alpha Delta Pi, and the President of the Theta Xi chapter of Pi Sigma Alpha. While her research throughout university centered on the death penalty, she also became interested in political communication and did her second senior thesis on the influence of Facebook on political candidates. She has presented research at the Pennsylvania Political Science Association Conference, the National Conference on Undergraduate Research, Undergraduate Research Day at the Pennsylvania Capitol, and the Pi Sigma Alpha Conference. As of September 2018, she is living in Spain for a year to teach English to 3-6 year olds through the CIEE program. After, she plans to combine her experience in education and public policy through a job in the international education sector.

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

Appendix A: States with/without death penalty legislation States with the Death Penalty Alabama Arizona Arkansas California Florida Georgia Idaho Indiana Kansas Kentucky Louisiana Mississippi Missouri Montana Nebraska Nevada New Hampshire North Carolina Ohio

States without the Death Penalty

States in Moratorium

Alaska

Colorado

Connecticut

Oregon

Delaware

Pennsylvania

Hawaii

Washington

Illinois Iowa Maine Maryland Massachusetts Michigan Minnesota New Jersey New Mexico New York North Dakota Rhode Island Vermont West Virginia Wisconsin

Oklahoma South Carolina South Dakota Tennessee Texas Utah Virginia Wyoming

Source: Death Penalty Information Center 2017

ÂŠ Pi Sigma Alpha 2018

The Determinants of Death Penalty Policy

Appendix B: State homicide classifications for death penalty punishment State

Homicide classifications for death penalty punishment

Alabama

Intentional murder with 18 aggravating factors

Arizona

First degree murder, including pre-meditated murder and felony murder, accompanied by at least 1 of 14 aggravating factors

Arkansas

Capital murder with a finding of at least 1 of 10 aggravating circumstances

California

First-degree murder with special circumstances (sabotage, train wrecking causing death, treason, perjury causing execution of an innocent person, fatal assault by a prisoner serving a life sentence)

Colorado

First-degree murder with at least 1 of 17 aggravating factors, first-degree kidnapping resulting in death

Florida

First-degree murder, felony murder

Georgia

Murder with aggravating circumstances, kidnapping with bodily injury or ransom when the victim dies

Idaho

First-degree murder with aggravating factors, perjury resulting in the execution of an innocent person

Indiana

Murder with 16 aggravating circumstances

Kansas

Capital murder with 8 aggravating circumstances

Kentucky

Capital murder with the presence of at least one statutory aggravating circumstance

Louisiana

First-degree murder

Mississippi

Capital murder

Missouri

First-degree murder

Montana

Capital murder with 1 of 9 aggravating circumstances, felony murder

Nebraska

First-degree murder with a finding of at least 1 statutorily-defined aggravating circumstance

Nevada

First-degree murder with at least 1 of 15 aggravating circumstances

New Hampshire

Murder committed during rape, kidnapping, drug crimes, or burglary, killing of a police officer, judge or prosecutor, murder for hire, murder by an inmate while serving a sentence of life without parole

North Carolina

First-degree murder with the finding of at least 1 of 11 stator aggravating circumstances

Ohio

Aggravated murder with at least 1 of 10 aggravating circumstances

Oklahoma

First-degree murder in conjunction with a finding of at least 1 of 8 statutorily-defined aggravating circumstances

Oregon

Aggravated murder

Pennsylvania

First-degree murder with 18 aggravating circumstances

South Carolina

Murder with 1 of 12 aggravating circumstances

South Dakota

First-degree murder with 1 of 10 aggravating circumstances

Tennessee

First-degree murder with 1 of 16 aggravating circumstances

Texas

Criminal homicide with 1 of 9 aggravating circumstances

Utah

Aggravated murder

Virginia

First-degree murder with 1 of 15 aggravating circumstances

Washington

Aggravated first-degree murder

Wyoming

First-degree murder, murder during the commission of sexual assault

Source: Death Penalty Information Center 2017

ÂŠ Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

Appendix C: Percentage of state that voted for the Republican presidential candidate in 2008, 2012, and 2016 along with their allotted Republican average Â

Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming

2008

2012

2016

Average

60 60 54 59 37 45 38 37 48 52 27 62 37 49 45 57 57 59 41 37 36 41 44 56 49 50 57 43 45 42 42 37 50 53 47 66 41 44 35 54 53 57 56 62 31 46 41 56 42 65

61 55 54 61 38 47 40 40 49 53 28 65 41 54 46 60 61 58 41 37 38 45 45 56 54 55 61 46 46 41 43 36 51 59 48 67 43 47 36 55 58 59 57 73 31 48 42 62 46 69

62 51 48 61 32 43 41 42 49 50 30 59 38 57 51 56 63 58 45 34 33 47 45 58 56 56 59 46 47 41 40 37 50 63 51 65 39 48 39 55 62 61 52 45 30 44 37 68 47 68

61 55.3 52 60.3 35.7 45 39.7 39.7 48.7 51.7 28.3 62 38.7 53.3 47.3 57.7 60.3 58.3 42.3 36 35.7 44.3 44.7 56.7 53 53.7 59 45 46 41.3 41.7 36.7 50.3 58.3 48.7 66 41 46.3 36.7 54.7 57.7 59 55 60 30.7 46 40 62 45 67.3

Note: Sources: NBC 2017; New York Times 2018.

ÂŠ Pi Sigma Alpha 2018

The Determinants of Death Penalty Policy

Appendix D: Variables not significant against the constant in isolated tests in Table 2

Constant

Referendum

Isolated 1

Isolated 2

Isolated 3

-.477

.262

-1.231

(.923)

(.568)

(.939)

.644 (.590)

Nonwhite

Southern

.008 (.017)

1.349

(.727)

REFERENCES American Civil Liberties Union. 2017. “Race and the Death Penalty.” https://www.aclu.org/other/race-and-death-penalty (Accessed April 3, 2017). Ballotpedia. “States with Initiative or Referendum.” https:// ballotpedia.org/States_with_initiative_or_referendum (December 3, 2017). Barak, Gregg. 2007. Battleground: Criminal Justice. Westport: Greenwood. Brace, Paul, and Brent Boyea. 2008. “State Public Opinion, the Death Penalty, and the Practice of Electing Judges.” American Journal of Political Science 52(2):360-372. Berman, Mark. 2015. “Connecticut Supreme Court Says the Death Penalty is Unconstitutional and Bans Executions for Inmates on Death Row.” Washington Post. August 13. https:// www.washingtonpost.com/news/post-nation/wp/2015/08/13/ connecticut-supreme-court-says-the-death-penalty-isunconstitutional-banning-it-for-remaining-inmates-on-deathrow/?utm_term=.c07e2d572380 (Accessed December 3, 2017).

death-penalty-support-lowest- years.aspx (Accessed September 20, 2017). Kaiser Family Foundation. 2016. “Population Distribution by Race/ Ethnicity.” https://www.kff.org/other/state-indicator/distributionby-raceethnicity/ (Accessed May 30, 2018). Kennedy v. Louisiana. 2008. 554 US 407. Kohut, Andrew, John Green, Scott Keeter, and Robert Toth. 2000. The Diminishing Divide: Religion’s Changing Role in American Politics. Brooking Institution Press. Langer, Laura, and Paul Brace. 2005. “The Preemptive Power of State Supreme Courts: Adoption of Abortion and Death Penalty Legislation.” Policy Studies Journal 33(3):317-340. Mooney, Christopher Z., and Mei-Hsien Lee. 2000. “The Influence of Values on Consensus and Contentious Morality Policy: U.S. Death Penalty Reform, 1956-82.” The Journal of Politics 62(1):223-239.

Death Penalty Information Center. 2017. https://deathpenaltyinfo. org/ (Accessed December 3, 2017).

Mooney, Christopher Z., and Mei-Hsien Lee. 1999. “Morality Policy Reinvention: State Death Penalties.” The Annals of the American Academy of Political and Social Science 566:80-92.

Federal Bureau of Investigation: Uniform Crime Reporting. 2016. Crime in the United States https://ucr.fbi.gov (Accessed May 30, 2018).

Morgan, David R., and Kenneth J. Meier. 1980. “Politics and Morality: The Effect of Religion on Referenda Voting.” Social Science Quarterly 61(1):144-148.

Furman v. Georgia. 1972. 408 US 238.

NBC. 2017. “Presidential Election Results.” http://elections.nbcnews. com/ns/politics/2012/all/president/#.WiWjH7GZNE6 (Accessed December 4, 2017).

Gallup. 2017. “U.S. Death Penalty Support Lowest Since 1972.” http://news.gallup.com/poll/1606/death-penalty.aspx (Accessed April 2, 2018). Gregg v. Georgia. 1976. 428 US 153.

New York Times. 2008. “Election Results 2008.” https://www. nytimes.com/elections/2008/results/president/map.html (Accessed December 4, 2017).

The Innocence Project. 2015. “DNA Exonerations Nationwide.” https://www.innocenceproject.org/dna-exonerations-in-the-unitedstates/ (Accessed December 3, 2017).

New York Times. 2017. “Presidential Election Results: Donald J. Trump Wins.” https://www.nytimes.com/elections/results/ president (Accessed December 4, 2017).

Jones, Jeffrey M. 2013. “U.S. Death Penalty Support Lowest in More Than 40 Years.” Gallup. http://www.gallup.com/poll/165626/

Nice, David C. 1992. “The States and the Death Penalty.” The Western Political Quarterly 45(4): 1037-1048.

Pi Sigma Alpha Undergraduate Journal of Politics Niles, William Ogden. 1822.”Chronicle.” Niles’ Register. Baltimore.160. http://www.nilesregister.com/NRessay.htm (Accessed May 30, 2018). National Public Radio. 2016. “Why Are Highly Educated Americans Getting More Liberal?” https://www.npr. org/2016/04/30/475794063/why-are-highly-educated-americansgetting-more-liberal (Accessed May 30, 2018). ProCon. 2018. “Historical Timeline of the Death Penalty.” https:// deathpenalty.procon.org/view.timeline.php?timelineID=000025 (Accessed May 30, 2018). Richburg, Keith B. 2007. “N.J. Approves Abolition of Death Penalty; Corzine to Sign.” Washington Post. December 14. http://www. washingtonpost.com/wp-dyn/content/article/2007/12/13/ AR2007121301302.html (Accessed December 3, 2017).

NOTES 1 Please see Appendix A for list of states that do and do not have the death penalty.

2 Please see Appendix B for a complete list of what is considered homicide that may be charged with the death penalty.

3 Please see Appendix C for a complete list of the percentage of a state that voted Republican in 2008, 2012, and 2016, along with the average that was given to that state. 4 Please see Appendix D for tables with variables that were not significant in Table 2.

Savage, David G. 2008. “Supreme Court to Hear Case on Death Penalty Restriction.” The Baltimore Sun. January 5. http://articles. baltimoresun.com/2008-01-05/news/0801050295_1_deathpenalty-supreme-capital-punishment (Accessed December 3, 2017). The Tuscaloosa News. 1941. “Negro Is First to Die for Night Burglary.” 2. https://news.google.com/newspapers?nid=1817& dat=19410808&id=SfY-AAAAIBAJ&sjid=Yk0MAAAAIBAJ& pg=6526,4499844 (Accessed May 30, 2018). Tyler, Tom R., and Renee Weber. 1982. “Support for the Death Penalty: Instrumental Response to Crime, or Symbolic Attitude?” Law & Society Review 17(1): 21-46. United States Census Bureau. 2015. “Census Regions and Divisions of the United States.” https://www2.census.gov/geo/pdfs/maps-data/ maps/reference/us_regdiv.pdf (Accessed May 30, 2018). United States Census Bureau. 2016. “2011-2015 American Community Survey 5-Year Estimates.” https://www.census. gov/programs-surveys/acs/technical-documentation/table-andgeography-changes/2015/5-year.html (Accessed May 30, 2018). United States Department of Commerce: Bureau of Economic Analysis. 2017. “GDP Per Capita for 2016.” https://www.bea.gov/iTable/iTable.cfm?r eqid=70&step=10&isuri=1&7003=200&7035=1&7004=sic&7005=1&7006=xx&7036=1&7001=1200&7002=1&7090=70&7007=-1&7093=levels#re qid=70&step=10&isuri=1&7003=1000&7004=naics&7035=1&7005=1&7006=xx&7001=11000&7036=1&7002=1&7090=70&7007=-1&7093=levels (Accessed May 30, 2018). Vidmar, Neil, and Phoebe Ellsworth. 1973. “Public Opinion and the Death Penalty.” Stanford Law Review 1245-1270. Waksman, David. 2012. “Is There a Death Penalty in America?” Law Enforcement Today. https://www.lawenforcementtoday.com/isthere-a-death-penalty-in-america (Accessed December 3, 2017). Woodham, Chai. 2008. “Eastern State Penitentiary: A Prison with a Past.” Smithsonian. https://www.smithsonianmag.com/history/ eastern-state-penitentiary-a-prison-with-a-past-14274660/ (Accessed December 3, 2017). Zimring, Franklin Ester, and Gordon Hawkins. 1989. Capital Punishment and the American Agenda. Cambridge University Press.

The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees

The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees Kerry Dugandzic, Connecticut College This research analyzes the political neutrality of Palestinian refugee employees of the United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNRWA). Four propositions precede the analysis: 1) UNRWA’s neutrality is compromised by the majority-Palestinian identity of its staffers and by the agenda of its main donor, the United States; 2) The staffers’ conflict of identity is exacerbated by the decline in their intra-Agency power; 3) Neutrality violations committed by Palestinian refugee employees increase as their power declines; and 4) Palestinian refugee employees are stripped of more power as the neutrality violations increase. Case studies of alleged UNRWA employee neutrality violations in Lebanon and Jordan lead to the hypothesis that UNRWA’s current balancing of the competing interests of its major donor and its employees is unstable. Findings suggest that UNRWA’s future viability may rely on a shift in leadership back to the Palestinians. INTRODUCTION Approximately 99 percent of the employees of the United Nations Relief and Works Agency for Palestine Refugees (UNRWA) in the Near East are Palestinian refugees themselves. These employees are caught in the midst of two identities: Palestinian refugee and international civil servant. The most controversial aspect of this dichotomy of identities is that, as UN employees, Palestinian refugees who work for UNRWA are held to UN neutrality principles. As a central actor in the highly polarized Israel-Palestine conflict, UNRWA’s supposed “neutrality” must be further analyzed. In reality, it has come to resemble, especially due to its employee composition, a pseudogovernment of the Palestinian people. Although it serves them as a government, it does not represent their interests like one, because its main donor’s interests do not necessarily align with Palestinian interests. Rather than being neutral, UNRWA is strained to balance these competing priorities. The priorities are juxtaposed as the Palestinian refugees seeking an end to the status quo (their displacement) and the U.S. government seeking to appease its ally, Israel, whose government does not want an influx of Palestinians returning to the country. In this research, the ramifications of the fact that 99 percent of UNRWA’s employees are Palestinian refugees,1 both for the organization’s neutrality as a humanitarian actor and for the political identity of its staff members, are examined. Four interrelated propositions are presented: 1) UNRWA’s neutrality is compromised both by the overwhelmingly Palestinian identity of the majority of its staff members and by the agenda of its main donor, the United States; 2) The staffers’ conflict of identity is exacerbated by the decline in the intra-Agency power of Palestinian staff members; 3) Neutrality violations committed by UNRWA’s Palestinian refugee employees increase as their power declines; and 4)

Palestinian refugee employees are stripped of more power as the neutrality violations increase. These propositions lead to the hypothesis that the UNRWA’s current balancing of the competing interests of its major donor, the United States, and its employees, majorityPalestinian, is unstable. Though the 1948 and 1967 periods of mass displacement seem distant, the troubling realities of refugee-hood for Palestinians in UNRWA field areas loom ever larger in the face of the UNRWA’s severe funding deficits and the increase in regional instability as a byproduct of the Syrian civil war and its waves of refugees. Support for the proposed hypothesis is demonstrated through two case studies, UNRWA in Lebanon and UNRWA in Jordan. Lebanon and Jordan are logical choices for this study, given that the two countries host the largest numbers of Palestinian refugees in the Middle East. Gaza and the West Bank could also be used in case studies. However, the dynamics between UNRWA, the Palestinian leadership, and the Israeli government are much more muddled in those cases. A different analysis would be necessary. UNRWA’s perceived neutrality (or lack thereof ) affects its day-to-day operations across the Middle East, as well as its incoming donations. The current state of affairs regarding UNRWA neutrality and its Palestinian refugee employees is not stable for future operations or for maintaining donor funding.

LITERATURE REVIEW Deconstructing Neutrality

“Neutrality” is and has long been a problematic term. The original definition of neutrality applied to nation-states only, in that “a neutral state will not take part either directly or indirectly in any forthcoming war between third-party states” (Goetschel 2011, 1696). Declaring neutrality was a way to help

Pi Sigma Alpha Undergraduate Journal of Politics

a state defend its interests in “an international environment marked by interstate conflicts” (Goetschel 2011, 1696). Credibility to do so, however, relied upon the acceptance of a state’s declared ‘neutral’ status by third parties - namely those who were at war (Goetschel 2011). From the very inception, the concept of neutrality is subject to debate due to this aspect of its definition. If a state is only neutral when validated in its neutrality by third parties, then “neutrality”’ is subjective to those third parties. In an international system of sovereign states, no belligerent state is held to any legal obligation to acknowledge and respect the declared neutrality of a “‘neutral“’ state (Komarnicki 1952). Similarly, humanitarian organizations that seek to operate as neutral actors rely upon the validation of their neutrality by third parties. Humanitarian actors must negotiate with local leaders and act within the territory administered by those leaders, which means that humanitarian actors are not only held to the agendas of local actors but also that they can legitimate belligerent non-state actors in the process (Belloni 2007), inadvertently taking sides. Their neutrality is also compromised by the practical matter of funding. Humanitarian actors accept donors’ money and therefore accept donors’ priorities (Belloni 2007). UNRWA’s neutrality statement on its website (2017b) illustrates these two issues: “Neutrality is critically important to United Nations humanitarian agencies – as well as ICRC and NGOs – to gain and maintain the confidence of all needed to operate independently, safely and effectively, especially in politically-charged or conflict situations. Neutrality is also a core obligation and value of UN staff and an important condition for our donors’ continued trust and financial and political support.” UNRWA and other humanitarian actors predicate their neutrality upon recognition by belligerent actors in a conflict situation and upon recognition by donors. It makes sense: if both donors and belligerent actors recognize the neutrality of an organization, the organization can carry out its daily operations unhindered and achieve the humanitarian mission it seeks to implement. There is no barring of access to refugee camps or attacks on humanitarian convoys, for instance. In the case of UNRWA, this means the day-to-day functioning of its human development and humanitarian services. However, there are further ideological complications to the idea of a humanitarian mission. Belloni (2007) analyzes humanitarianism as an ideology in two senses: 1) in that it comprises a “set of ideas and values” about “the importance of protecting individual and group rights” with the aim of “guiding political behavior accordingly,” and 2) in that its values “originate from and legitimate unequal relationships of domination and subordination,” particularly between “the West and the lessdeveloped world” (Belloni 2007, 454). Humanitarian actors 36

oppose policies in conflict with the international humanitarian law and thus are not neutral in that respect. They are, by their mission, bound to uphold the tenets of humanitarianism, here defined as “the political, economic and military interference in the domestic affairs of a state justified by a nascent transnational morality” (Belloni 2007, 451). Humanitarian agencies are also not neutral if their mission “hides a Western agenda of containment” by localizing “the underlying reasons for distress” (Belloni 2007, 454). Belloni (2007, 463-464) writes: “Humanitarianism’s main function is not so much that of improving the human condition by changing the structural circumstances which permit human rights violations but that of temporarily sedating political crises, preventing their escalation into wars with cross-national and cross-border consequences, and limiting their impact on Western countries.” UNRWA may be worthy of this criticism - after all, it has done exactly as Belloni (2007) described. For over half a century, it has served to temporarily mollify the Palestinian refugees it serves, although their plight of displacement remains ever present. While the organization was not created to solve the Palestinian crisis, it is inextricably tied to it by its employee composition and the aims of its main donor. Neutrality at the state, international organization, and humanitarian actor levels is consequently rife with both conceptual and practical difficulties. At the individual level, problems also abound. Researchers Mele, Anderfuhren-Biget, and Varone (2016, 451) conducted a study on the “intrinsic tensions” faced by individual employees of international organizations who are bound to “neutrality.” International organizations (IOs) are defined by Mele et al. (2016, 490) to be “membership organizations where member states not only express their eminently political representation through delegates or diplomats operating at the plenary legislative level, but also provide, more or less indirectly, nationals who are hired by the IO to keep the machine running.” Mele et al. (2016, 490) specifically focus on two (anonymous) United Nations humanitarian agencies in their analysis of the “systematic concerns” about “the ability of IOs to remain neutral playing fields with a homogenous administrative body the International Civil Service (ICS).” Employees of the United Nations formally agree “not to solicit or receive directives from any state” and to “act in the sole interest of the IO” (Mele et al. 2016, 492). They agree, in essence, to “neutrality”’ in the international arena and their presumed neutrality provides “the basis for the credibility and legitimacy of the system vis-a-vis the public” (Mele et al. 2016, 492) - just as in the prior cases of neutrality, the neutrality of IOs by virtue of the neutrality of their employees legitimates the IOs in the eyes of the public (third parties). This concept of ICS neutrality is complicated by the aforementioned

The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees

political reality of the United Nations and of humanitarian organizations. In particular, mandates of IOs have increasingly become more politicized due to the increase in earmarked government funding (Mele et al. 2016). Mele et al. (2016, 500) found that “G staff” (general staff) at the local level “frequently face tensions between national and international interests,” compromising their neutrality more so than international staff. This conclusion, in context with the problematization of the concept of neutrality, validates the premise upon which this research is based: that the United Nations Relief and Works Agency for Palestine Refugees in the Near East is inherently not a neutral organization. In reality, it represents a quasi-governmental institution, containing the trappings of a Palestinian state. It is responsible for providing services traditionally assigned to national governments in the fields of education, health, and social services (Al Husseini 2000), and it has even expanded to micro-credit and microfinance services (Al Husseini and Bocco 2010). The fact that 99 percent of its employees are Palestinian refugees themselves, in context with the current lack of a unified Palestinian non-IO, only validates this observation. Palestinian refugees are active political agents, not passive victims (Richter-Devroe 2013). Those who work for UNRWA informally represent the concerns and needs of the refugee constituents at large and have even formed staff committees through which to advocate for their interests both as employees and as Palestinians (Schiff 1995). Therefore, although UNRWA is not an actual state government, Palestinian concerns are often its concerns, and the socioeconomic development and legal protection of the Palestinian refugee population are integral to its mission. Meanwhile, UNRWA’s leadership is primarily Western, as is its donor base, necessitating that it caters to a pro-Western agenda - much to the dismay of its refugee employees and constituents.2 In terms of its “neutrality” in a more traditional sense, UNRWA cannot be separated from Israel or Palestine, which invalidates its claim to neutrality. That is, since its leadership often caters to a Western, pro-Israel agenda (Gabiam 2016) and its employees are majority-Palestinians, it cannot be separated from the two sides to be called a third-party “neutral.” In this case, Israel and Palestine are not third parties to UNRWA’s “neutrality” - they are aspects of its organizational mission and membership. Just one example of this is found on the UNRWA website’s main page (2017c): “UNRWA highlights the international community’s obligation to provide a just and durable solution for Palestine refugees.” UNRWA directly involves itself in politics by including a solution to the conflict as an aspect of its organizational mission. Thus, UNRWA has “unstable boundaries” and is susceptible to pressures from all sides rather than being independent of them (Gabiam 2016, 61). Moreover, since neither the Israeli government nor Palestinian refugees view UNRWA as completely neutral, its neutrality loses credibility.

UNRWA and Palestinian Nation-Building

Belloni (2007, 464) accuses humanitarian agencies of being “better in alleviating human suffering, rather than addressing the underlying problems which led to conflict, and thus providing a cover for strategic inaction.” UNRWA, he writes, represents “this placebo humanitarian function” in its “most grotesque form” (Belloni 2007, 465). Though UNRWA has done much to help the Palestinian population, Palestinian displacement remains as unresolved as it was when UNRWA was created in 1949. The Western governments who dictate the funding and consequently the mission (Gabiam 2016) of UNRWA “prefer to prevent short-term negative side effects, such as massive refugee flows,” instead of “putting forward long-term and potentially costly political initiatives” (Belloni 2007, 465). UNRWA, however, has historically been associated with Palestinian identity as a symbol of both their displacement and the international community’s responsibility to provide a just solution to the Israel-Palestine conflict. For example, UNRWA’s ration cards that symbolize Palestinian refugee status “came to be regarded by the refugees as a legal justification for their right to return and obtain compensation. The ration card itself has constituted an official- and often unique-piece of documentary evidence attesting to a physical link with Palestine, thereby becoming a symbol of Palestinian identity” (Al Husseini 2000, 52). This specific example indicates a general perception of UNRWA by many Palestinian refugees as “a unique reflection of the international community’s recognition of their ‘right to return.’” (Al Husseini and Bocco 2010, 268). Of course, defining what Palestinian identity means or to whom it refers is a complex task. Palestinians living in Lebanon and Jordan have different experiences, as do those living in the other UNRWA fields or the greater diaspora. For this research, when referring to “Palestinians” or “Palestinian identity,” this is defined as Palestinians living in the areas where UNRWA operates: Lebanon, Jordan, Gaza, the West Bank, and Syria. Before the formation of the Palestine Liberation Organization (PLO), Palestinians looked to UNRWA as “an informal institutional framework within which the rehabilitation of Palestinian society could take place” (Al Husseini 2000, 53). UNRWA also served as a “quasi-political” representative for the Palestinians on the international stage, in the absence of an internationally-recognized Palestinian leadership (Al Husseini 2000). Now, after the decline of the PLO and the rise of the weaker Palestinian Authority (PA), UNRWA has a similar capacity. The Oslo agreements of the 1990s “engendered a [Palestinian] national crisis and a schism between the Palestinian authority and the vast majority of refugees” that “catapulted UNRWA again to center stage, whereby it regained its significance, not only as a welfare government, but as a space through which refugees could indirectly negotiate their political and legal rights” (Farah 2009, 411). Indeed, UNRWA “has established itself as a quasi-state institution, taking on responsibilities traditionally

Pi Sigma Alpha Undergraduate Journal of Politics

assigned to national governments in the field of basic education and vocational training, preventive and curative health care, regular and emergency relief and social welfare services and, especially since the early 1990s, micro-credit and microfinance” (Al Husseini and Bocco 2010, 262). It has resumed, albeit on a smaller scale, its pre-PLO political and developmental positioning (Al Husseini 2000). The presence of its “evergrowing number of local employees” further implicates its status as a quasi-governmental institution (Al Husseini and Bocco 2010, 267).

Hiring Palestinians

Regarding the process of hiring, the UNRWA Careers website (2017d) claims that “Priority consideration for area staff posts is given to applications from registered Palestine refugees.” In 2017, approximately 30,800 area staff (presumably mostly Palestinian refugees3) and just 178 international staff worked for UNRWA according to its records (UNRWA 2017a). UNRWA is open, even proud, about its Palestinian staff base. This is because the hiring of Palestinians falls in line with its philosophy of “participatory development.” As Rempel (2009, 414) writes, “While UNRWA traditionally viewed participation as a means of programme implementation, stakeholder participation acknowledges participation as a human right with the primary objective of empowering refugees to realize their rights and improve social equity.” In the initial period following their displacement, hiring Palestinians was UNRWA’s attempt to involve refugees in their socioeconomic reintegration (Rempel 2009, 415). The goal was twofold: ensure project implementation by employing refugees and facilitate refugees to become self-reliant and eventually integrate themselves economically into their host countries or other countries abroad. This was problematic from the start - refugees largely agreed that the solution to their problem was the right of return, not economic integration elsewhere (Rempel 2009, 415). Although UNRWA has since expanded its services and even allowed some refugees roles in the actual planning, monitoring, and implementation of its programs (Rempel 2009, 432), the power of the Palestinian voices within UNRWA is still weak compared to the influence of its leadership and donor base.

Implications of the Literature Review

It is clear that UNRWA’s claim to neutrality is contestable for a variety of reasons. The concept of the neutrality of international organizations is theoretically flawed, especially for one so politicized as UNRWA. Further, the study conducted by Mele et al. (2016) implies that all local staff members of United Nations agencies face conflicts of neutrality. Given the extended duration and politicization of the IsraeliPalestinian conflict, it can be assumed that Palestinian employees of UNRWA face a particularly difficult duality of identities as both international civil servants and Palestinian refugees. In turn, this duality strains UNRWA as it attempts to 38

balance its Palestinian-ness and its donor loyalties. The tug of war between UNRWA’s identity as a quasi-Palestinian-state and its identity as a Western-funded organization is increasingly more tense as the reality of displacement endures.

CASE STUDIES The current practical limits of UNRWA’s neutrality are analyzed in terms of employee violations thereof and connect those neutrality violations to a marked decline in the leadership positions of Palestinians within UNRWA. Finally, it is demonstrated how the decline in leadership and the increase in neutrality violations continue to be a self-fulfilling prophecy that undermines UNRWA’s efficacy as a whole.

Background Arab League Guidelines for Palestinian Integration in the Near East.

Shortly after the creation of the state of Israel, the Arab League formally established the legal and political guidelines for Palestinian refugees’ integration in host states. Today, these guidelines are still regarded as the standards to which Arab states ought to adhere (Al Husseini and Bocco 2010). The major principle these guidelines established as an international norm was that Palestinian refugees should be maintained as stateless persons in order to retain their Palestinian nationality and thus preserve their “right of return” (Al Husseini and Bocco 2010). This resulted in the decadeslong implementation of a “positive discriminatory regime” in most of the host states (Al Husseini and Bocco 2010, 264). For UNRWA, the states’ adherence to these guidelines has meant that it is limited in its capacity to provide services that would enhance the quality of living for Palestinian refugees; “The notion of the ‘right of return’ has… conditioned the development of the refugee camps’ physical and housing infrastructure” (Al Husseini and Bocco 2010, 261) Thus, under the Arab League Guidelines, UNRWA is incredibly politicized. In addition, because of the “positive discriminatory regime” implemented in most host states towards Palestinians, UNRWA is the main vehicle through which Palestinian refugees selfadvocate and attain various services and rights. The states’ determination to maintain the statelessness of Palestinians under the guidelines also ensures the growth of UNRWA’s refugee population, since it provides services to descendants of Palestinian refugees (Gabiam 2016).

Lebanon.

According to UNRWA statistics (2017f ), approximately 450,000 Palestinian refugees are registered in Lebanon, the majority of whom resides in the twelve recognized camps. They represent about ten percent of the population but are denied from a variety of rights, including the right to work in twenty different professions. This denial of the right to work means

The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees

that for many Palestinians in Lebanon, UNRWA is the main source of employment. Among the five UNRWA field areas, Lebanon has the highest percentage of Palestinian refugees living in abject poverty (UNRWA 2017f ). Furthermore, the situation in Lebanon has been exacerbated by the high influx of Syrian refugees, some of whom are of Palestinian origin and were registered with UNRWA in Syria. Treated with much more discrimination than ordinary Syrians, Palestinian refugees from Syria are deprived of even more rights, live in the most poverty, and are troubled by their ever-precarious legal status in the country (UNRWA 2017f ). It is due to these factors that UNRWA also has come to advocate for the protection of Palestinian refugees alongside its service provisions.4 UNRWA’s Protection Unit “promotes the rights of Palestine refugees under international law, through the monitoring and reporting of violations and by engaging in private and public advocacy” (UNRWA 2017g).] It is also interesting to note that, if UNRWA now claims the responsibility to promote the rights of Palestinian refugees under international law, it opens itself up to the responsibility of advocating for Palestinian refugees’ right to return to Palestine as established by the UN General Assembly Resolution 194 (Al Husseini 2000). Resolution 194 is an aspect of international law. UNRWA’s protection policy is yet another example of how the organization has become increasingly politicized.

Jordan.

The two million Palestinian refugees in Jordan (UNRWA 2017h) enjoy the most rights and privileges out of UNRWA’s fields. Most of them have full citizenship and are more integrated into the local population. However, there are still camps: ten recognized Palestinian refugee camps throughout the country accommodate approximately 370,000 Palestinian refugees - 18 percent of the country total (UNRWA 2017h). Similar to Lebanon, Jordan has also become the recipient of a number of Palestinian refugees from Syria, in addition to the influx of Syrian citizens. Nearly 10,000 Palestinians from Syria have sought refuge in Jordan and, unlike their local counterparts, they suffer from uncertain legal status and abject poverty (UNRWA 2017h). Palestinians in Jordan have more agency, but those from Syria and those displaced after 1967 are very vulnerable. There are two political cultures among Palestinian refugees in Jordan, that of those who do not hold citizenship and that of those who do (Richter-Devroe 2013). Palestinian refugees who are either from Syria or were displaced from Gaza after 1967 are “highly marginalized, lack protection, and have little possibilities for political engagement,” giving them far less agency than naturalized Palestinians (Richter-Devroe 2013, 97). While naturalized Palestinians may lobby for more rights and support from the Jordanian government (Richter-Devroe 2013), those without citizenship are in a similar position to Palestinian refugees in Lebanon as they rely on UNRWA for

legal protection and advocacy alongside basic distribution of goods and services.

Palestinian Political Identity in Lebanon and Jordan.

The demographic weight of Palestinian refugees in Jordan and Lebanon meant that their active political involvement “only reinforced the fear amongst Arab leaders that the refugees’ presence would remain a potential threat to the country’s stability” (Al Husseini and Bocco 2010, 264). In combination with a history of opposition political movements and Palestinian nationalism mobilization, both countries’ histories of ethnic conflict contributed to this fear5. However, Jordan, from the outset, departed from the overall Arab state trend of positive discrimination. In 1949, the Kingdom granted citizenship to all Palestinians residing within it (Al Husseini and Bocco 2010). However, it was in the interest of the Jordanian government to do so: “To legitimize the claim and control over the West Bank, the transformation of Palestinian refugees into disciplined subjects was paramount” (Achilli 2014, 237). This meant that Jordan effectively “created” a new type of refugee: “the refugee-citizen,” who could enjoy the benefits of Jordanian citizenship while maintaining their “temporary” refugee status (Al Husseini and Bocco 2010, 263). Palestinians in Jordan thus did not have to choose between renouncing their Palestinian refugee status and integrating into Jordanian society. That being said, informal discrimination occurs in the fields of employment in the Jordanian public sector - especially in the military and intelligence services - and of representation in national institutions like Parliament (Al Husseini and Bocco 2010, 265). Additionally, “any notion of a separate ‘Palestinian identity’ in Jordan’s internal politics was banned by royal decree” (Al Husseini and Bocco 2010, 265). Palestinians could symbolically maintain their refugee status and still live and work as Jordanians, but were not allowed to mobilize as a nation. The roots of this policy reach back to the 1967 ArabIsraeli war, which pushed the number of Palestinians in Jordan to comprise over half of the total population (Achilli 2014). After the 1967 war, the Palestinian Liberation Organization (PLO) was also created, which contributed to the development of a much stronger Palestinian national consciousness and a subsequent mass mobilization of Palestinian refugees (Achilli 2014). Then, the 1970-1971 Black September civil war “gradually precipitated a discourse of unity within the context of a guest/host relationship” (Achilli 2014, 239). The government’s agenda began privileging non-Palestinian aspects of national identity (Achilli 2014). These events resulted in Palestinians in Jordan struggling to navigate the seeming contradiction of being both Palestinian refugees and Jordanian citizens (Achilli 2014). After nearly 70 years, many of them still live in “temporary” spaces that are simultaneously spatially and economically integrated into Jordan’s cities (Achilli 2014, 246). Against this backdrop, overt forms of political resistance and assertion of Palestinian

Pi Sigma Alpha Undergraduate Journal of Politics

national identity have fallen by the wayside (Achilli 2014). Meanwhile, authorities in Lebanon have embraced a policy of fierce discrimination against Palestinians with the goal of their permanent emigration (Al Husseini and Bocco 2010). Due to the delicate Lebanese political system, in which sectarianism reigns, the country’s policy has made clear that any major improvements in Palestinian living and working conditions could lead to their upsetting the sectarian balance through assimilation - since most Palestinians are Sunni Muslims, they would tip the country’s Muslim population into the majority (Al Husseini and Bocco 2010). Moreover, as in Jordan, the history of ethnic conflict between Lebanese citizens and Palestinians contributed to the marginalization of the Palestinian refugee community. Again, the rise of the PLO and the mass mobilization of Palestinians after the 1967 war resulted in conflict: the Lebanese civil war lasted more than 15 years (Halabi 2004). Further, the state of Lebanon continued to suffer from the ramifications of the Palestinian-Israeli conflict, most obviously during the Israeli invasions of 1982 and 2006. The “Palestinian cause” became “the ‘Palestinian problem’ and for the first time, there was popular consensus against granting the refugees political support and civil rights” (Halabi 2004, 42). Today, Palestinians in Lebanon have “few allies or supporters” (Khalidi 2010, xxvii) and, like their Jordanian counterparts, are politically disillusioned (Allan 2013).

UNRWA Employees and Neutrality

UNRWA is extremely dependent on its local staff members. Yet, these local employees are not considered to be part of the UN’s regular labor force (Schiff 1995). Thus, matters of neutrality violations must be dealt with by UNRWA on a case-by-case basis, since each local government has its laws and interpretations of what such violations could mean (Buehrig 1971). As Buehrig (1971, 97) writes, “The vexed question at issue is who decides the official or private character of a particular act.” Ultimately, governments have the power to constrain UNRWA’s freedom and the freedom of its employees (Buehrig 1971). In Jordan, Lebanon, and Syria, the host governments review local job applicants for UNRWA employment. Moreover, major donors such as the United States have attempted, and succeeded in many cases, to influence the hiring process. According to the U.S. Senate General Accounting Office (U.S. Congress. Senate. 2003), the United States placed a “Refugee Coordinator” in Amman to monitor UNRWA staff specifically in 2003, and that position, now based in Jerusalem (U.S. Department of State 2017), expanded to monitor UNRWA across all of its fields. Lawmakers in Washington have made clear that donations to UNRWA hinge on their ability to monitor and influence the organization. It is not necessarily anti-U.S. interest that most hires are Palestinian - if allowing them to work placates their resistance to perpetual displacement, all the better. Additionally, recall the (Western40

originated) ideology of “participatory development” discussed in the literature review.

From mighty to marginalized: The decline of Palestinian power in UNRWA.

As was previously mentioned, local Palestinian refugee staff comprise the grand majority of UNRWA employees. When UNRWA was created, its “temporary nature made working for it morally defensible, because its employees could argue that they were helping their people while still defending their right to return” (Schiff 1995, 144). Over time, however, UNRWA lost its temporary air, expanding into socio-economic development and shifting away from its traditional forms of immediate assistance (Gabiam 2016). This mission shifting was accompanied by a shift in leadership of the organization. In the mid-1980s, UNRWA centralized power under the international employees in its field offices, thereby eroding the power and influence that local employees previously held (Schiff 1995). International employees, though presumably more “neutral” than their local counterparts, are not necessarily better leaders: local employees “may be under more pressure or more biased than the internationals, but the structure of the situation forces them to analyze the interests, positions, and rationalities of all actors more fully and with more nuance than must the internationals” (Schiff 1995, 148). The ability of local UNRWA employees to balance UN duties and community loyalties, however, has eroded largely in great part due to the organization’s limiting of their power. Schiff believed in 1995 that UNRWA had the potential to transform into a Palestinian organization, “part of the bureaucracy of an independent or confederated Palestinian state, run by Palestinians” (279). Yet, in the post-Oslo period, “the already formidable obstacles to a Palestinian state in any meaningful sense of that word - a state that is independent, sovereign, possessed of a continuous territory, and economically viable - have in fact been growing rapidly” (Khalidi 2010, xxixxii.). In depriving local UNRWA staff members of their negotiating and diplomatic power, UNRWA’s donor base weakened UNRWA itself. Though UNRWA still provides many important services, the increasing amount of staff neutrality violations actively contribute to its undermining. UNRWA’s Palestinian employees no longer feel the power within and commitment to UNRWA that they once did. As the following section will illustrate, this decline in power seems to be associated with a decline in donor funding to UNRWA as well as an increase in presumed violations of neutrality by Palestinian employees of UNRWA.

Neutrality violations in Lebanon and Jordan.

Neutrality violations in UNRWA take two forms: internal and external. Internal violations are committed by UNRWA employees, usually local staff members. Meanwhile, external violations occur when an independent actor violates

The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees

the neutrality of UNRWA premises. This paper will focus on internal neutrality violations, which occur intellectually on social media and physically in the form of protest. This research focuses on internal violations because those are violations that are made by staff members, as opposed to external violations that are made by an unaffiliated actor. UNRWA does not often publicly respond to allegations of internal neutrality violations, which often come from Israeli watchdog news sites, and, more officially, from the Genevabased NGO, UN Watch. Though UN Watch, which was created in 1993, is not officially an Israeli organization, it has strong ties to Israel and focuses especially on combating antiIsrael and anti-Semitic sentiments within UN organizations such as UNRWA. A 2017 UN Watch Report sought to do just that by searching for UNRWA-related keywords on Facebook6 and discovering (what UN Watch believes to be)7 UNRWA staff neutrality violations. According to UN Watch (2017a, ii), “Were UNRWA itself to examine its employees, on and off Facebook, it can reasonably be estimated that thousands of UNRWA employees would be implicated.” Some of the alleged violations included: “UNRWA teachers, staffers and institutions celebrating the terrorist kidnapping of Israeli teenagers, cheering rockets being fired at Israeli civilian centers, endorsing various forms of violence including random knife and vehicular attacks against Israeli civilians, erasing Israel from the map, praising Hitler and posting his photo, and posting overtly anti-semitic videos, caricatures, and statements” (UN Watch 2017a, v). The organization also notes, “These cases of incitement are additional to the 30 revealed by UN Watch in 2015. More examples are found by UN Watch every day” (UN Watch 2017a, i). The majority of such violations came from UNRWA teachers and other local staff members (UN Watch 2017a). UN Watch (2017a) cites 21 incidents tied to UNRWA employees in Lebanon and 18 tied to ones in Jordan. These figures are particularly relevant since the other incidents listed are only 11 each for Gaza and Syria (UN Watch 2017a). These statistics are small, but the effects of the accusations on UNRWA are influential. For instance, the 2017 UN Watch report was aimed at convincing the Canadian government to re-examine its funding to UNRWA, and after the report was released a debate ensued in the Canadian parliament over just that (UN Watch 2017b). Though UN Watch may not be able to track every violation that occurs, its reports are important to examine. Due to limits on UNRWA’s transparency, there may be no true way to know the number of internal violations that have occurred over time. Nevertheless, these examples of social media and those physical protests analyzed below illustrate the importance of beginning this research and asking these questions about UNRWA. Based on the data presented by UN Watch, employees of UNRWA in Lebanon and Jordan seem to be committing the most alleged violations of their neutrality as international civil servants. Moreover, those committing violations are, as

expected, mostly local staff members who work directly in the camps as opposed to the field offices. Just over 71 percent of those accused in Lebanon were cited to be UNRWA teachers or at least to work in UNRWA educational facilities (UN Watch 2017a). In Jordan, just over 61 percent of those accused are UNRWA teachers or involved in UNRWA education (UN Watch 2017a). Approximately 21,571 UNRWA staff members are education staff (UNRWA 2017i). Given that the total employee population of UNRWA is around 30,000 people, more than half of the population – 72 percent - is significant. Since over half of the Palestinian refugee population is comprised of children and people under the age of 25 (Chatty 2010), this means that UNRWA teachers are particularly influential members of the refugee community. This may partially explain why the teachers are more inclined to invoke declarations of their overt dislike or hatred of Israel - because their loyalty is primarily to the Palestinian community as opposed to the United Nations. Moreover, the fact that they are not considered members of the UN’s labor force, combined with the fact that their influence over the years has declined, can only contribute to these feelings. At the same time, neutrality violations, which appear to be increasing over time,8 compromise donor’s faith in UNRWA to maintain the neutrality associated with legitimate international organizations - especially in context with other destabilizing factors, such as the Syrian civil war and extremist violence in the camps (Gunness 2017). As was aforementioned, donors such as the United States have inserted themselves into the UNRWA hiring process. Moreover, UNRWA has been criticized both by donors and by the broader international community for its lax staff screenings and its lack of an antifraud system, “which is imperative for employees to whistle blow and report suspicious activity.”9 According to the UNRWA website (2017c), “the Agency has been forced to implement austerity measures over the last few years due to lack of funds: financial contributions have not increased sufficiently to keep pace with inflation and a rising refugee population. This has resulted in a reduction in services as is evident in the fact that average annual spending per refugee has fallen from about $200 in 1975 to around $110 today.” In the context of an apparent lack of faith in UNRWA to monitor the neutrality of its staff members, there is certainly a correlation between a decline in funding and neutrality violations. It is also important to note that Westerners, either Europeans or Americans, hold the main leadership positions in UNRWA Headquarters and its field offices: for instance, the offices of the Director of UNRWA, the Deputy Director, and the majority of the Department Heads in both Lebanon and Jordan are Western in origin (UNRWA 2017j). The fear that UNRWA cannot control the neutrality of its staff members may contribute to a lack of Palestinians in leadership positions. As discussed in the literature review, of the 30,000 total UNRWA staff members, only about 100-200 are international

Pi Sigma Alpha Undergraduate Journal of Politics

staff funded by the UN Secretariat. The remaining 8,329 staff members not working in education are mostly local employees in other fields such as relief and social services. To give context for Lebanon and Jordan, UNRWA (2011) showed that the field refugee staff in Jordan was comprised of 5,452 refugee staffers out of a total 6,782 staffers; in Lebanon, refugee staffers comprised 2,785 out of 2,931 employees. Yet, again, the leadership positions are not held by Palestinians. Furthermore, outside of Facebook and other social media platforms, local staff have shown their loyalties during protests at UNRWA field offices. Though refugees have historically used protest as a tactic with which to enforce change (Schiff 1995, 89), the types of protests seen today are unprecedented, as is the Palestinian staff’s participation in them. Largely due to UNRWA funding cuts, refugee staffers have protested multiple times throughout recent years, several times per year. For this reason, protests broke out multiple times in the last few years. For example, in September 2016, a staff strike that originated in Gaza spread to Lebanon and Jordan, as well as to Syria and the West Bank. The UNRWA administration was accused of “deliberately” putting off the hiring of employees “in order to win some time and save money at the expense of refugees.”10 UNRWA was further charged with “not seeking to palliate the lack of staff in schools and health centers, despite there being hundreds of unemployed graduates qualified to hold positions in these fields.”11 Meanwhile, protests in Lebanon in January 2016 closed the UNRWA field office in Beirut for days due to funding cuts to UNRWA’s healthcare services.12 The protests sparked after a 23-year-old refugee, Omar Khudeir, self-immolated outside of a health clinic - the cuts made him unable to afford hospitalization services for a hereditary blood disorder.13 Palestinian refugees not employed by UNRWA raised the question as to why, if it lacks enough staff members and money, it continues to prioritize the “lavish” lifestyle of international staff: “‘Why does the UNRWA hire foreign workers and pay for their accommodation, travel, etc. when we, qualified Palestinians, can work for the organization and then they will be providing a salary, saving a family, and reducing wages?’” demanded one refugee living in Lebanon’s Bourj El Barajneh camp.14 In 2009, UNRWA employees in Jordan protested in the form of a sit-in due to the same issue of funding cuts causing a decline in services to Palestinian refugees.15 The same tactic of and the reason for protest occurred again in 2015 for multiple months.16 Earlier, the current political disillusionment for Palestinians in both Lebanon and Jordan was discussed. In the context of these protests at UNRWA field offices and buildings, it seems as though Palestinian refugees, both those who are UNRWA employees and those who are not, are only politically inactive towards their host governments. Indeed, the refugees view the UNRWA as their main government, since it was supposedly created for them and provides goods and services. 42

UNRWA is the pseudo-government of the stateless Palestinians, and its Palestinian refugee employees are the representatives of the people. Yet, the trend of the protests illustrates that representatives who are deprived of any intra-Agency power turn against their government, rendering it ineffective. The increasing regularity of these protests, in conjunction with a continued decline in donor funding and consistent accusations of social media neutrality violations, clearly undermines UNRWA’s efficacy as a whole.

CONCLUSION Although UNRWA has been criticized both by refugees and nation-states for its lack of neutrality (Gabiam 2016), it has still received somewhat steady support for a variety of reasons. Its services are relatively efficient, its existence symbolizes the preservation of Palestinian refugee rights (in the view of refugees), and its presence in the Near East serves as a stabilizing factor (in the view of Western donors) (Al Husseini and Bocco 2010). UNRWA can continue serving in these positive respects, but the thousands of Palestinians in its employment cannot continue to remain powerless. As the case study section illustrated, the status quo of Palestinian subordination grows less and less manageable. Palestinian refugees are not “passive victims, and they variously appropriate, renegotiate, or subvert humanitarian classifications and practices, and challenge the intentions and interests of more powerful actors” (Farah 2009, 391). The cyclical nature of the Palestinian employee disempowerment and neutrality violations within UNRWA undermines its efficacy both in the eyes of its donors and in the eyes of the refugees it serves. Therefore, all four propositions are realized by the research completed in this study. UNRWA’s neutrality is compromised by the Palestinian identity of the majority of its staff. This duality of identity is exacerbated by the decline in the intra-Agency power of those staff members. Moreover, neutrality violations committed by UNRWA’s Palestinian staff members have increased in recent years, correlating with the marked decline in power of such staffers. The increase in violations has contributed to donor doubts in UNRWA and thus has contributed to the further stripping of the Palestinians’ power within UNRWA. These statements lend support to the hypothesis that UNRWA’s current balancing of the competing interests of its major donor, the United States, and its employees, majority-Palestinian, is unstable. In addition, the research suggests that UNRWA’s future viability may rely on a shift in leadership back to the Palestinians. Staff and refugee protests of UNRWA are increasing alongside donor funding cuts - it seems as though UNRWA is faltering in its organizational efficacy. The limits of this study exist in terms of the amount of data that can be gathered from a distance further research might be more conclusive if data are gathered on the ground in Jordan, Lebanon, and UNRWA’s other field areas. Furthermore, research might be conducted on the other

The Politics of Neutrality: UNRWA’s Palestinian Refugee Employees

field areas in which UNRWA operates, as well as on the effects of external neutrality violations of UNRWA premises. It would also be a worthy research task to analyze the implications of the Syrian civil war and other politically destabilizing factors in the region on UNRWA funding cuts.

ABOUT THE AUTHOR Kerry Dugandzic is a graduate of Connecticut College, where she majored in Government and Global Islamic Studies and minored in Arabic. At Connecticut College, Kerry was also a scholar of the Toor Cummings Center for International Studies and the Liberal Arts (CISLA). Her senior project for CISLA was converted into this manuscript for the Pi Sigma Alpha Undergraduate Journal of Politics. She spent four months in Jordan and four months in Lebanon studying Arabic, undertaking an internship at UNRWA Lebanon, and conducting research to inform the manuscript. Kerry is currently finishing up an internship in the Middle East Program at the Center for Strategic and International Studies in Washington, D.C., and she expects to attend graduate school in the coming years to hone her Arabic skills and further develop her knowledge of Middle East politics. Her goal is to someday work in the U.S. Department of State in the Bureau of Near Eastern Affairs.

Goetschel, Laurent. 2011. “Neutrality.” In International Encyclopedia of Political Science, eds. Bertrand Badie, Dirk Berg-Schlosser, and Leonardo Morlino. Thousand Oaks: Sage, 1696-1699. Gunness, Chris. 2017. “UNRWA Strongly Condemns Violations of the Inviolability and Neutrality of United Nations Installations in Ein el Hilweh.” March 4. https://www.unrwa.org/newsroom/ official-statements/unrwa-strongly-condemns-violationsinviolability-and-neutrality-united (Accessed May 7, 2018). Gunness, Chris. 2015. “Allegations of Neutrality Violations Unfounded.” August 28. https://www.unrwa.org/newsroom/ official-statements/allegations-neutrality-violations-unfounded (Accessed May 7, 2018). Halabi, Zeina. 2004. “Exclusion and Identity in Lebanon’s Palestinian Refugee Camps: A Story of Sustained Conflict.” Environment and Urbanization 16 (2): 39-48. Khalidi, Rashid. 2010. Palestinian Identity. New York: Columbia University Press. Komarnicki, T. 1952. “The Problem of Neutrality under the United Nations Charter.” Transactions of the Grotius Society 38: 77-91. Mele, Valentina, Simon Anderfuhren-Biget, and Frédéric Varone. 2016. “Conflicts of Interest in International Organizations: Evidence from Two United Nations Humanitarian Agencies.” Public Administration 94 (2): 490-508. Rempel, Terry. 2009. “UNRWA and the Palestine Refugees: A Genealogy of ‘Participatory’ Development.” Refugee Survey Quarterly 28 (2–3): 412–437. Richter-Devroe, Sophie. 2013. “‘Like Something Sacred’: Palestinian Refugees’ Narratives on the Right of Return.” Refugee Survey Quarterly 32(2): 92–115. Schiff, Benjamin N. 1995. Refugees Unto the Third Generation: UN Aid to Palestinians. 1st ed. Syracuse: Syracuse University Press.

REFERENCES Achilli, Luigi. 2014. “Disengagement from Politics: Nationalism, Political Identity, and the Everyday in a Palestinian Refugee Camp in Jordan.” Critique of Anthropology 34 (2): 234-257. Allan, Diana. 2013. Refugees of the Revolution: Experiences of Palestinian Exile. Stanford, CA: Stanford University Press. Al Husseini, Jalal. 2000. “UNRWA and the Palestinian NationBuilding Process.” Journal of Palestine Studies 29 (2): 51-64. Al Husseini, Jalal, and Riccardo Bocco. 2010. “The Status of the Palestinian Refugees in the Near East: The Right of Return and UNRWA in Perspective.” Refugee Survey Quarterly 28 (2-3): 260285. Belloni, Roberto. 2007. “The Trouble with Humanitarianism.” Review of International Studies 33 (3): 451-474. Buehrig, Edward H. 1971. The UN and the Palestinian Refugees: A Study in Nonterritorial Administration. Bloomington: Indiana University Press.

Taubenfeld, Howard. 1953. “International Actions and Neutrality.” The American Journal of International Law 47 (3): 377-396. UNRWA. 2017a. “UNRWA in Figures.” https://www.unrwa.org/sites/ default/files/content/resources/unrwa_in_figures_2017_english.pdf (Accessed May 7, 2018). UNRWA. 2017b. “UNRWA and Neutrality.” https://www.unrwa.org/ userfiles/2011033075942. pdf (Accessed May 7, 2018). UNRWA. 2017c. “Frequently Asked Questions.” https://www.unrwa. org/who-we-are/frequently-asked-questions (Accessed May 7, 2018). UNRWA. 2017d. “Careers.” https://www.unrwa.org/careers/you-apply (Accessed May 7, 2018). UNRWA. 2017e. “Working at UNRWA.” https://www.unrwa.org/ careers/working-unrwa (Accessed May 7, 2018). UNRWA. 2017f. “Where We Work: Lebanon.” https://www.unrwa. org/where-we-work/lebanon (Accessed May 7, 2018).

Chatty, Dawn. 2010. “Palestinian Refugee Youth: Agency and Aspiration.” Refugee Survey Quarterly 28 (2-3): 318-338.

UNRWA. 2017g. “What We Do: Protection.” https://www.unrwa.org/ what-we-do/protection (Accessed May 7, 2018).

Farah, Randa. 2009. “UNRWA: Through the Eyes of its Refugee Employees in Jordan.” Refugee Survey Quarterly 28 (2-3): 389–411.

UNRWA. 2017h. “Where We Work: Jordan.” https://www.unrwa.org/ where-we-work/jordan (Accessed May 7, 2018).

Gabiam, Nell. 2016. The Politics of Suffering: Syria’s Palestinian Refugee Camps. Bloomington: Indiana University Press.

UNRWA. 2017i. “What We Do: Education.” https://www.unrwa.org/ what-we-do/education (Accessed May 7, 2018).

Pi Sigma Alpha Undergraduate Journal of Politics UNRWA 2017j. “Who We Are: Our Leadership.” https://www.unrwa. org/who-we-are/our-leadership (Accessed May 7, 2018). UNRWA. 2011. “UNRWA Statistics 2010.” https://www.unrwa.org/ userfiles/2011120434013.pdf (Accessed May 7, 2018). UN Watch. 2017a. “Enhanced Due Diligence? An Examination of Canada’s Pledge to Stop UNRWA Teachers from Inciting Jihadist Terrorism and Antisemitism.” April. https://www.unwatch.org/ wp-content/uploads/2009/12/Canada-and-UNRWA-EnhancedDue-Diligence.pdf (Accessed May 7, 2018). UN Watch. 2017b. “Clash in Canadian Parliament over UNRWA Teachers’ Incitement Exposed by UN Watch.” April 21. https:// www.youtube.com/watch?v=iZU6hD6ivJw&t=0s&list=PLrf9i5 CED35oFiJFAlVdBkn5nPBS-UOtv&index=2 (Accessed May 7, 2018). U.S. Congress. Senate. General Accounting Office. 2003. Department of State (State) and United Nations Relief and Works Agency (UNRWA) Actions to Implement Section 301(c) of the Foreign Assistance Act of 1961. GAO-04-276R. U.S. Department of State. 2016. Framework for Cooperation Between the United Nations Relief and Works Agency for Palestine Refugees in the Near East and the United States of America for 2017, 4 November.

NOTES 1 Of the 30,977 UNRWA employees registered in 2017, only 178 were international staff (UNRWA 2017a). 2 Meaning those whom UNRWA serves. 3 The UNRWA employment website states that “UNRWA has over 30,000 employees, most of them Palestine refugees and a small number of international staff” (UNRWA 2017e). 4 UNRWA adopted its protection policy in 2012. 5 Black September in Jordan (1970-1971) and the Lebanese Civil War (1975-1990). 6 It is important to note that more neutrality violations may be discovered today due to the technology available to most people. Before social media platforms, such as Facebook, staff’s private views may never have been public and therefore contestable. 7 UN Watch has no way to confirm whether or not all UNRWA Facebook pages or UNRWA employee pages are valid - in the past, UNRWA has admitted to false pages being created (Gunness 2015). 8 Based on UN Watch (2017a), which states that 30 incitements were found in 2015 - compared to the 61 described for 2016. 9 Chaker Khazaal, “Palestinian Refugees Protest Against UNRWA,” Huffington Post, 21 January, 2017, https:// www.huffingtonpost.com/chaker-khazaal/palestinian-refugees-prot_b_9038304.html (Accessed May 7, 2018). 10 “UNRWA Workers Strike in Four Countries over Low Wages,” Albawaba, September 19, 2016, https://www.albawaba. com/news/unrwa-workers-strike-four-countries-over-low-wages-884434 (Accessed May 7, 2018). 11 Albawaba, 2016. 12 Khazaal, 2017. 13 Khazaal, 2017. 14 Khazaal, 2017. 15 “Jordan: UNRWA Employees, Activists Stage Sit-in to Protest Services Cutback,” Deutsche Presse Agentur, August 19, 2009, https://reliefweb.int/report/occupied-palestinian-territoryjordan-unrwa-employees-activists-stage-sit-protest-services (Accessed May 7, 2018). 16 Laila Azzeh “UNRWA Workers Protest ‘Premeditated Plot’ Against Refugees,” The Jordan Times, August 5, 2015, http:// www.jordantimes.com/news/local/unrwa-workers-protest-premeditated-plot%E2%80%99-against-refugees (Accessed May 7, 2018).

Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level

Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level Zoe Bouras, Illinois Wesleyan University Globally, the citizens of developed democracies are exhibiting high levels of political distrust and anti-elite sentiment, a situation which poses a serious threat to the health of democracy and established political systems. While there are system-level trends that contribute to anti-elitism, there remain variations in anti-elite sentiment at the individual level. Two broad schools of thought suggest explanations for this; one contending that the way an individual perceives and responds to their economic situation leads to anti-elite sentiment, and the other, that anti-elite sentiment has risen as a reactionary undercurrent to the “silent revolution” of the last half of the 20th Century. Using Round 7 of the European Social Survey (ESS) from 2014, this research investigates a third approach which brings together insights of the cultural and economic explanations: the “left-behind” hypothesis. This study finds that in the European context, those who perceive themselves to be “left-behind” by economic and cultural trends express stronger anti-elite sentiments than even the poorest “losers of globalization,” or the most traditionally conservative. INTRODUCTION “‘Howard Dean,’ the husband says, ‘should take his tax-hiking, government-expanding, latte-drinking, sushi-eating, Volvo-driving, New-York-Timesreading’ – at which point the wife interrupts to finish the sentence - ‘body piercing, Hollywood-loving, left-wing freak show back to Vermont, where it belongs’” (Tierney 2004). The last 50 years in international politics can be characterized by a steep decline in political trust and the rise of anti-elite and antiestablishment attitudes. These lower levels of political trust have many serious repercussions, and scholars suggest that wide-spread political distrust can lead to the eventual dissolution of existing political systems and the downfall of modern liberal democracy (Dalton 2004; Gidron and Hall 2017; van der Meer and Hakhverdian 2017). Within the overarching theme of “political distrust” there are various sub-categories to be identified, such as distrust of supranational organizations, government institutions, elites, and the media. This study deals specifically with the decline of public trust in elites and the rise of anti-elite sentiment. Some academics suggest that anti-elite sentiment is one of the least important dimensions of trust because it has the least severe repercussions, as elites are subject to change, and because anti-elite sentiment is not entirely surprising in a representative system (Dalton 2004; McLaren 2012; Seyd 2008). This explanation, however, does not give due consideration to the consequences of widespread anti-elite sentiment. Scholars should be concerned when voters are distrustful of politicians and other political, economic, and cultural elites, because that distrust and cynicism is a defining characteristic of populism,

and populism is viewed as a threat to liberal democracy (Mudde 2007; Müller 2016). Indeed, the ideological core of populism involves a deep-seated distrust of elites and intellectuals, and the rhetoric often used by populist parties and actors pits the pure, regular people against the corrupt and wicked elite. This study seeks to address the question of why specific individuals develop anti-elite attitudes, and while distrust is embedded in a larger structural context, this cross-national European study will attempt to identify specific markers that influence anti-elitism at the individual level.

Anti-Elite Sentiment in Context

Scholarly literature suggests a variety of factors to explain why citizens are losing faith in their elites at the international, societal, and individual levels. Some scholarship emphasizes the role of just one level of influence, while other academics, such as Dalton (2004), suggest that the real explanation is a combination of influences. At the systemic level, explanations include: actual cases of corruption; the changing nature of the media; and the activities of politicians and elites themselves. The latter of which can reinforce and exacerbate perceptions that elites are untrustworthy, incompetent, or corrupt. At the individual level, the demographic variables associated with cultural change, in combination with a feeling of being “left behind” or abandoned by mainstream society, play a definitive role (Gidron and Hall 2017; Hochschild 2016; McLaren 2012). However, before examining these two levels of explanation, it is important to define elites and anti-elite attitudes.

Who are the Elites?

The term elite once commonly referred to those with money or power, however, there has been a shift in the way

Pi Sigma Alpha Undergraduate Journal of Politics

that “elite” is used in common parlance. Now, elite not only includes the politically powerful or rich, but also carries a cultural connotation, and lifestyle has created a new stratum of elitism. This concept of a cultural elite, sometimes referred to as the “metropolitan elite,” has permeated many post-industrial societies around the globe, and is perpetuated by populist parties (Tierney 2004). The pitting of the regular, proper, and “pure” general citizenry against the “corrupt” and distant elite is a hallmark of populist platforms (Betz and Johnson 2004; Mudde 2007; Müller 2016). Not only are elites seen as different and disinterested, but they are publicized as being the antithesis of good government and the representation of the general will of the people. “Central to the contemporary radical right’s politics of resentment is the charge that in liberal capitalist democracies power has been usurped by a self-serving political and cultural elite that pursues its own narrow agenda without concern for the legitimate concerns and interests of ordinary citizens” (Betz and Johnson 2004, 313). There are many contemporary examples of this rhetoric internationally. The 2004 U.S. presidential election offered a clear example of this definitional shift, when a right-wing political group known as the “Club for Growth” launched an advertising campaign that identified democratic presidential candidate, Howard Dean, as someone who, “should take his tax-hiking, government-expanding, latte-drinking, sushi-eating, Volvo-driving, New-York-Times-reading… body-piercing, Hollywood-loving, left-wing freak show back to Vermont” (Tierney 2004). This quote bears a striking resemblance to a statement by a politician from the New Zealand First political party which categorized elites as being, “the spa bath, Chardonnay sipping, social [i.e. left-wing] elitists who have more interest in the fine arts than they do in working class Kiwis” (Betz and Johnson 2004, 314). In the European context, much of the 2016 Brexit campaign focused on anti-elitism. Additionally, Mudde (2007) identifies Romania as an example of anti-elite sentiment, as the Greater Romania Party (PRM) once published, “a list of top intellectuals who should be shot for the greater good of the country” (Mudde 2007, 67). Given this, the discussion of anti-elite sentiment becomes very broad, but the political consequences remain severe.

System Level

Contemporary anti-elitism is nestled in a system-level, international context. Trends associated with changing media standards, globalization, the rise of internet technology, and the rise of right wing populism, contribute to a wide-ranging suspicion of elites that is more generalizable than simply an increase in corruption (Dalton 2004). In Democratic Challenges, Democratic Choices (2004), Dalton addresses this by stating that in America, the conversation focuses on, “…Vietnam, Watergate, Abscam,” and “Iran-Contra,” but that most places in the world have no experience of those specific incidents (Dalton 2004, 31). While the Watergate scandal may have seriously changed thinking about politicians and 46

the presidency in the U.S., it should not have affected levels of trust in elites in New Zealand or Switzerland, yet people in New Zealand and Switzerland do exhibit declining levels of political trust (Dalton 2004). It is clear that “…coincidental unique political crises across a large and diverse set of nations is an improbable explanation of general patterns...” and as such, we need to look to the individual-level to ascertain what truly drives anti-elite attitudes (Dalton 2004, 4). Corruption is an important factor when discussing political distrust and anti-elite sentiment, as it is reasonable to be distrustful of a corrupt system (Anderson and Tverdova 2003). However, there is evidence to suggest that real cases of corruption do not carry as much weight as the perception of corruption, and there are many systematic influences on the prevalence of these perceptions (van der Meer and Hahkverdian 2017; Ziller and Schübel 2015). People and organizations that have public platforms are able to influence how other people decide which issues are worth caring about. This agendasetting ability has become more problematic since the largescale introduction of the Internet, as now essentially anyone can have a visible platform. This high level of accessibility poses problems on two fronts: firstly, it is easier than ever to spread misinformation, and secondly, the real facts and positions of experts and elites become diluted by the rest of the discourse. Similarly, politicians and experts themselves may contribute to the public view that elites should not be trusted by contradicting each other or identifying each other as untrustworthy, as “crooks,” or as incompetent; political campaigns are increasingly run on a platform of “cleaning up” government (CNN 2011; Hoschchild 2016; Mudde 2007; Vaccines 2015). These contradictions and challenges to elites and the establishment work to reinforce ideas of anti-elitism. The public is constantly confronted by a discourse that suggests that there is indeed something inherent to politicians and elites of which they should be skeptical. These systemic influences are plausible and can explain broad trends in anti-elite sentiment in the developed world, but they cannot account for differences in the level of anti-elite sentiment across individuals in post-industrial democracies. It becomes important to examine individual level influences to ascertain why some people are more likely than others to exhibit anti-elite sentiment.

Individual Level

An analysis at the individual level involves examining the ways in which an individual’s circumstances influence their perspectives and responses to systemic trends. One way that scholars view individual attitudes toward politics and elites is through the lens of economic loss. Scholars use various phrases such as, “losers of globalization,” and “economic deprivation” to describe this framework. These theories contend that the economic problems that have arisen as a result of an increasingly globalized world, such as outsourced jobs and income inequality, are the main source of frustration for

Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level

citizens (Gidron and Hall 2017; Hessami 2011). The economic perspective “emphasizes the consequences for electoral behavior arising from profound changes transforming the workforce and society in post-industrial economies;” it is a very clear-cut way to pit the “us” against the “them,” or, as Inglehart and Norris (2016) phrase it, the “haves” against the “have-nots.” Those most likely to feel these impacts of economic insecurity are the low-income earners; the unskilled, the “long-term unemployed;” those who are unable or unwilling to take advantage of the “travel and consumption” opportunities of increased globalization and modern openings in the job market; and the poorer sectors of society with vested interests in redistributive policy and economic stability (Gidron and Hall 2017; Hessami 2011). Those at the bottom of the income ladder are more likely to favor redistributive policy and objectives, and as such, they are likely to be left-wing populists. The problems that these people face can be seen as the fault of the economically and culturally powerful; “…blaming ‘Them’ for stripping prosperity, job opportunities, and public services from ‘Us’ ”(Inglehart and Norris 2016, 2). In comparison, the “cultural backlash” hypothesis provides an explanation that focuses on the undercurrent of reaction in post-industrial societies to the liberalizing cultural shifts that occurred in the 1960s and 1970s (Inglehart and Norris 2016, 5). This negative reaction to progressive cultural change builds on the idea that in the 1970s there was a silent revolution which resulted from high levels of economic security, education, peace-time, and a new emphasis on “new” politics and postmaterialist values (Inglehart and Norris 2016). This rhetoric is more appealing to right-wing populists as it focuses on the maintenance of the status quo and incorporates nativism and tradition. The rise of the radical right is seen as a cultural backlash against the growing hegemony of postmodern values and New Left elites in top political, economic, and cultural positions. As stand-alone explanations, both the economic grievance and the cultural backlash schools offer important insights to the discussion surrounding antielite sentiment. Empirical studies tend to find that cultural values and attitudes are more telling of political leanings than economic status and demographic indicators, however, the social and economic trends that lead to the backlash are not new and have been in play for decades. Given this, it is not clear why some people have reacted to those systemic trends with strong anti-elite attitudes while others have not. So, what makes some individuals more trusting of elites than others? To answer that question, it is necessary to bring economic grievances together with cultural reactions. As post-materialist values have become entrenched in society, and levels of education, equal rights, and secularization have risen, those with more traditionalist values and those who previously experienced status security have become sensitive to societal changes they see as threatening (Betz and Johnson 2004; Gidron and Hall 2017; Hochschild 2016). In Strangers in Their Own Land, Hochschild (2016) uncovers a feeling

that the government does not work for all individuals, and finds that some people do feel left behind by both their elites and government in the American context. There is evidence to suggest that this idea of a cultural cleavage is deeply seated, that it has led to a school of thought that identifies the “left behind” as a distinct group, and that it is present outside of American society (Dalton 1988; Gidron and Hall 2017). “Important segments of the white working class are said to be alienated from societies whose main currents seem to have left them behind. They feel as if they are ‘strangers in their own land’ whose value goes unrecognized by the affluent elites…” (Gidron and Hall 2017, 4). It is this distinction between the “regular folk” and the “affluent elites” who are responsible for the perversion of society that leads to anti-elite sentiment among the left behind. It is the elite who are ushering in the very changes that are unwelcome. Gidron and Hall (2017) also examine the “left-behind” through the operationalization of social connectedness and standing; they model the ways that employment status and skill levels have direct effects on social integration, and self-identified social status (Gidron and Hall 2017). Being “left behind” is the result of both cultural and economic factors, and the people who are likely to feel the most left behind are not necessarily the poorest or the culturally traditional, but rather they are those who feel they have more at stake in a shifting world (Gidron and Hall 2017). This research argues that in the European context, antielite sentiment emerged from a feeling of being left-behind and from low levels of social integration. One would anticipate that, in general, the population sectors that feel the most left behind would be majority racial males, with low skills and employment prospects, and who hold traditional values (Dalton 2004; Gidron and Hall 2017; Hochschild 2016; Mudde 2007; Rothstein and Stolle 2008; van der Meer and Hahkverdian 2017). Additionally, groups that have evolved to be the “butt of the joke” are likely to feel left behind and betrayed by their politicians and pundits (Hochschild 2016). These expectations stem from a variety of societal changes during the last 50 years or so. Theoretically, just as the rise of gender equality puts men at odds with women and feminism, the rise of racial equality puts the majority racial group at odds with the culture of multiculturalism and integration. An example of this can be seen in the recent increase in immigration to the European Union (EU) and recent controversies over refugee flows from the Middle East, which have led to the prevalence of strong anti-immigrant attitudes across the continent. McLaren (2012, 204) links these antiimmigrant attitudes to declining political trust and anti-elite sentiment, “negative perceptions of the impact of immigration may reduce both willingness to support the existence of a group of individuals who engage in policy-making and willingness to support the institutions through which these groups of elites govern.” Those who feel most threatened by immigration are most likely to feel as though the elites have “sold out” and betrayed them (McLaren 2012). For example, equal rights for

Pi Sigma Alpha Undergraduate Journal of Politics

the LGBTQA+ community elicits negative reactions from the very religious and traditional conservatives. The less educated may feel left behind due to the increase of tertiary education in developed post-industrial society and the resulting loss of status for secondary and vocational education (Gidron and Hall 2017). Additionally, as the hub of economic activity has moved from rural areas to cities and towns, the rural and provincial, and those employed in the agricultural sector, will feel left behind. Those who now face unemployment or underemployment due to brain drain or outsourcing will view themselves in opposition to the “big-wig” elites who made those choices. There are, of course, nuances to consider with this model of resentment and betrayal, specifically when considering gender and age. Women are not necessarily any less populist than men, and, as author and scholar Stephanie Coontz explains in a 2016 interview, “…women who have the fewest opportunities to compete successfully in the labor market are the ones who are much more likely to support the policies and values that reward a traditional division of labor in the household…,” (Denvir 2016). These women also see themselves as losing in the face of new politics, and consequently, we should anticipate gender to be an indefinite indicator (Denvir 2016). Similarly, the impact of age differs according to scholastic literature. Some academics suggest that there is a substantial age-gap in anti-elite sentiment and support for the populist radical right, with the elderly feeling estranged as their generation is displaced, and other studies find that youth are particularly populist (Arzheimer and Carter 2006). Final important considerations for the European context are the Eurocrisis and the consequent increase of wide-spread Euroscepticism (Werts, Scheepers, and Lubbers 2012, 187). The existence of the European Union and Eurozone can serve as examples of how the government and elites are sacrificing the wellbeing of the nation and its citizenry (Müller 2016, 24). “Mainstream politicians and political parties of both the centerleft and center-right have both been the drivers of the European Union, making the EU an elite-led, politically centrist project” (Polk and Rovny 2017, 4). The ways in which the European Union limit national sovereignty can be seen as direct attacks on the nation-state, and the power of the EU is often criticized (Polk and Rovny 2017). In the wake of the Eurocrisis, those who are economically and socially well off are less likely to be Euroskeptic than those that have suffered economically, particularly when considering the various austerity measures that were imposed in nations such as Greece (Müller 2016; Polk and Rovny 2017, 3; Werts, Scheepers, and Lubbers 2012). Additionally, some EU decisions facilitate increased job competition and the idea that those outside the state are taking jobs, “the opening of borders and the EU-enlargement might have increased competition on lower levels of the labor market in particular,” which leaves low-skilled citizens feeling economically insecure (Werts, Scheepers, and Lubbers 2012). These impingements on sovereignty, culture, and the citizen’s 48

way of life can all be seen as failures of the elites to protect and serve their constituents and culminate in strong anti-elite and anti-establishment attitudes across Europe. These feelings are clearly evidenced by both the populist right and left; for example, by Greece’s Syriza and by the UK’s Brexit campaign (Polk and Rovny 2017). Ultimately, the “left behind” are expected to be those who perceive themselves to have a social standing and status to protect that has been threatened in the contemporary moment. The left-behind approach offers a set of hypotheses that will be tested; rather than following Gidron and Hall (2017) directly, this study investigates social integration and self-identified social status separately. H1: The more socially integrated a person feels, the less likely they will be to hold anti-elite sentiments. H2: The more secure a person’s self-identified social status (SSS), the less likely they will be to hold anti-elite sentiments. In testing these hypotheses, one can organize independent variables by the hypothesis in which they most closely relate, for example, issues such as level of social interaction is relevant to social integration but not to SSS, as such it is used to test hypothesis H1. Independent variables of social integration (used to test H1) are: rural or urban living, new political ideas, social trust, political trust, employment, and social interaction. The socially integrated person will be an employed, urbanite with new political ideas, high levels of social and political trust, low nativist feelings, and will meet with friends and family regularly. Variables of self-identified social status (used to test H2) are: years of education, gender, resentment, and age. A person who is vulnerable because of their SSS will have lower levels of education, and will harbor resentful feelings. They will probably be an elderly male, however, as was previously discussed, there are variations in results regarding both age and gender.

RESEARCH DESIGN AND METHODOLOGY This study draws upon data from Round 7 of the European Social Survey (ESS), which was conducted in 2014. The use of responses from European voters is logical because it builds upon a theoretical framework which suggests that developed democracies that have undergone post-material values change will be the most likely to exhibit anti-elite sentiments, where the elites are defined as political and cultural elites, as much as, or more than, purely economic elites. This is not to dismiss the rise of political distrust in other areas of the world, but the literature suggests that manifestations of populism, political distrust, and cynicism outside of post-industrial liberal democracies are driven by different phenomena (Dalton 2004; Gidron and Hall 2017; Hochschild 2016; Inglehart and Norris 2016; Mudde

Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level

2007; Seyd 2008; van der Meer Figure 1: Heat Map of Anti-Elite Attitudes across Europe and Hahkverdian 2017; Ziller and Schübel 2015). The European case selection allows for control over the basic systemic influences that can be identified from the literature, such as post-materialist value changes, and the shift to an information economy. By controlling for these variables, one can better examine the predictors of individual-level attitudes. The current study analyzes survey responses from approximately 37,500 core voters, between the ages of 18 and 100, from twenty European countries. Countries included in the survey and the number of respondents from each country can be found in Appendix A. An ordinary least squares regression model was conducted to model individual level anti-elite sentiments, where elites are defined as politicians, and level of trust and support is measured on a 4-point Note: These data are taken from Round 7 of the European Social Survey 2014 scale. Independent variables included are age (1 = younger, 4 = older), gender (1 = male; 0 = female), employment status (1= their elites. As a whole, the heat map does illustrate the overall employed in paid work, 0 = other), years in education (1 = 12+ trend for rising anti-elite sentiment in advanced democracies, and the variation across states suggests that different historical years, 0 = 0-12 years) resentment (1 = high resentment 0 = low and cultural influences anti-elite sentiment at the national level. resentment), rural or urban living (0 = rural, 1 = urban), new The first stage of analysis for the individual level employs political opinion (1 = new politics, 5 = traditional politics), bivariate correlations in order to assess the relationships among social interaction (0 = low social interaction, 7= high social variables and the degree and direction of the association interaction), a perception of income (1= positive perception between each independent variable and the dependent of income, 4 = negative perception of income) and various variable (anti-elite sentiment). To assess H1 (“The more composite measures including social trust, political trust, and socially integrated a person feels, the less likely they will be to nativist sentiment. The full operationalization of these variables hold anti-elite sentiments”) I use the independent variables is listed in Appendix B. previously labeled “indicators of social integration.” Similarly, to test H2, (“The more secure a person’s self-identified social status, the less likely they will be to hold anti-elite sentiments,”) ANALYSIS AND RESULTS SSS indicators were used. The bivariate correlation results are Figure 1 shows international dispersal of anti-elite attitudes presented in Table 1. All variables are significant, and the most across Europe. The heat map is based on the percentage of respondents who scored a 1 on the scale of anti-elite sentiment, robust relationship is between anti-elite sentiment and political trust (r = .735; p = .01), followed by social trust (r = .401; p and thus have the least favorable attitudes towards elites. Generally, the results indicated on the map are satisfying and fit = .01), and nativist sentiment (r = .287; p = .01). Most of the measures of social integration: social interaction, political trust, with popular epistemology about attitudes in different groups social trust, urban living, and employment, do correlate in the of nations. For example, Scandinavian states exhibit the lowest anticipated direction. New politics, however, does not. This levels of anti-elite sentiment, with Norway reporting less than could result from the deference to authority that is common in 10% of people at the lowest operationalized level of anti-elite traditional politics. H2 is partially supported, and both age and sentiment. Similarly, former Soviet states have some of the gender do correlate as anticipated. The gender result suggests highest levels of distrust; for instance, in Poland, over 50% of respondents reported the lowest operationalized level of trust in that men are slightly more trusting of elites than women, and © Pi Sigma Alpha 2018

Pi Sigma Alpha Undergraduate Journal of Politics

that the elderly exhibit more anti-elite sentiment than the young. One interesting indicator is years of education, as it is significantly correlated with every other variable in ways that make intuitive sense. Results from the multivariate regression can be seen in Table 2. Table 2 reports that many variables become statistically insignificant, with social trust, political trust, social interaction, nationalism, age, and income perception remaining significant. A significant regression equation was found (F(12,28854)=3014.806, p < .000), with an adjusted R² value of .556. This means that the model can account for roughly 55% of the variance in the dependent variable (anti-elite sentiments). Analysis of a step-wise regression (not reported in this paper) indicates that much of the variance in anti-elite sentiment can be explained by a single indicator: political trust. Political trust, social trust, and attitudes toward income together account for 55.4% of the variance in the model. In the regression model in Table 2, both education (β = .016, p < .036) and nativism (β = .043, p < .00) become much weaker predictors of anti-elite sentiment than the bivariate correlations indicate in Table 1. Despite this, education is substantively significant as it is so correlated with other measures. Similarly, nativism seems to become less robust in the regression model than reported in the bivariate correlation table. In Table 1, nativism is significantly correlated with every other variable at

the .01 level, and could be washed out in the control model. For both H1 and H2, it is possible to reject the null hypothesis: there are significant correlations between SSS and anti-elite sentiment, and between social integration and antielite sentiment. For H1, these relationships are evidenced by significant relationships between anti-elite sentiment and rural/ urban living (r = .047; p = .01), new political ideas (r = .193; p = .01), social trust (r = .401; p = .01), political trust (r = .735; p = .01), employment (r = .075; p = .01), nativism (r = .131; p = .01), and social interaction (r = .150; p = .01). In Table 1, the same significant correlations can be seen between anti-elite sentiment and the IVs for H2, years of education (r = .038; p = .01), gender (r = .141; p = .01), resentment (r = .287; p = .01), age (r = -.068; p = .01), and income feeling (r = -.275; p = .01). The empirical findings are consistent with the idea that attitudinal measures are more telling than demographics. Furthermore, data analysis lends support to the direction and causal relationship between the independent variables and dependent variable.

CONCLUSION The last half-century has been characterized by rising anti-elite sentiment and general political distrust among the public in developed democracies. These broad trends can be explained at

Table 1: Correlation Matrix of Independent Variables on Anti-Elite Sentiment Variables

1.Anti-Elite Sentiment

2.Rural/Urban

.047**

3.New Politics

.193**

.049**

4.Social Trust

.401**

.027**

.203**

5.Political Trust

.735**

.056**

.200**

.413**

6.Employment

.075**

.030**

.123**

.054**

.067**

7.Nativism

.131**

.014**

.186**

.152**

.135**

.101**

8. Interaction

.150**

.117**

.203**

.139**

.133**

.004

.125**

9.Years of Education

.038**

.002

-.026**

-.003**

.057**

.261**

.207**

.057**

10.Gender

.141**

.073**

.176**

.132**

.139**

.076**

.017**

.023**

-.021**

11.Resentment

.287**

.091**

.300**

.280**

.305**

.050**

.270**

.088**

.138**

-.025**

12.Age

-.068**

-.026**

-.170**

.003

-.059**

-.247**

.046**

-.201**

-.163**

-.037**

-.063**

13.Income Feeling

-.275**

-.008

-.194**

-.276**

-.264**

-.171**

-.195**

-.170**

-.187**

-.066**

-.141**

.046**

Note: *p < .05. **p < .01

10.

11.

12.

Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level

the system-level by the rise of globalization, the rise of Internet technology, mass education, and changing media and political practices. Figure 1 indicates that there is an international trend toward anti-elite sentiment, and that there are differences in attitudes at the level of the individual: some states, despite being subject to the same international trends, are more antielitist than others, and some individuals are anti-elitist where others are not. This study sought to understand why those individual Europeans have adopted anti-elite attitudes to a greater extent than others. Scholars have proposed explanations involving feelings of economic insecurity and a sense of impending cultural threat, and the “left-behind” hypothesis incorporates both of these insecurities. This study indicates that in the European context, an individual’s levels of social and political trust are the most telling indicators of their attitude toward elites, which is in keeping with the trend for attitudes to be more important explanatory indicators than demographic measures. This study also finds that the left behind theory is useful for explaining anti-elite sentiment, as both economic and social variables emerge as significant. There are limitations to this study that future research could address. One such limitation was that Round 7 of ESS did not include a question about race. Existing literature suggests that race-driven resentment is an important indicator of anti-elite attitudes and distrust; as multi-culturalism and equal-rights for minorities become more important and gain more attention, majority-race groups may feel angry and out of step with society. Another limitation relates to the type of elite the study was able to address. This study was only

able to evaluate attitudes towards political elites because of the questions in the data set. Without questions addressing attitudes towards academics and cultural elites, this study remains incomplete, as it cannot consider the full range of elites. Future research should seek to include a specific marker for race, as well as including questions that tap into attitudes toward academics, experts, and cultural elites such as television hosts and news anchors. Also, because of the differences in trust levels between states which are indicated in Figure 1, it may be interesting to use dummy variables and control for state. This study contributes to a question of growing importance in the international community; growing anti-elite sentiment is a phenomenon with severe consequences for existing social and political systems, and getting to the heart of these attitudes could help to quell them. n

ABOUT THE AUTHOR Zoe Bouras is a recent graduate of Illinois Wesleyan University, where she double majored in political science and international studies with a concentration in development studies. As an undergraduate, she was a member of a variety of honor societies including Pi Sigma Alpha, Phi Beta Kappa, Phi Kappa Phi, Phi Beta Delta, and Mortar Board. She also was active on campus in Model United Nations, the IWU Rock Climbing Club, Kappa Delta Sorority, and IWU College Democrats. Moving forward, Zoe plans to pursue a Master’s degree in Latin American Studies at the University of Oxford in the fall of 2018.

Table 2: Predictors of Anti-Elite Sentiment Variables

Standard Error

Beta

Significance

Rural/Urban Living

.055

.008

-.001

.797

Social Trust

.067

.004

.079

Political Trust

.487

.003

.677

New Politics

-.003

.002

-.006

.145

Employment

-.005

.008

-.003

.474

Social Interaction

-.005

.002

.009

.037*

Years of Education

.016

.008

.009

.036*

Gender (Male)

.001

.007

.000

.929

Resentment

.008

.004

.340

Nativism

.043

.005

.041

.000**

Age

-.018

.005

-.016

.000**

Income Feeling

-.053

.005

-.046

.000**

Adjusted R 2

.556

37,471

.000** .000**

Note: *p < .05. **p < .01 These data are taken from Round 7 in the European Social Survey 2014

Pi Sigma Alpha Undergraduate Journal of Politics

Appendix A: List of included countries and their number of cases in the analysis Country

Cases

Country

Cases

Austria

1,791

Netherlands

1,910

Belgium

1,768

Norway

1,436

Czech Republic

1,522

Poland

1,605

Denmark

1,497

Portugal

1,256

Estonia

2,047

Slovenia

1,224

Finland

2,078

Spain

1,920

France

1,913

Sweden

1,785

Germany

3,038

Switzerland

1,522

Hungary

1,682

United Kingdom

2,253

Ireland

2,376 TOTAL

37,471

Lithuania

2,239

Note: Case selection from Round 7 of the European Social Survey 2014

Appendix B: Operationalization and Coding of Variables Dependent Variable: Anti-elite Sentiment

New Politics

Measured on an additive scale

B27. “Gay men and lesbians should be free to live their own life as they wish?” (scale of 1-5)

B5. “How much do you personally trust each of the institutions I read out…politicians?” B1e. “How much would you say that politicians care what people like you think?

01-05 =01

06-10=02

11-15=03

16-20=04

(1-high anti-elite sentiment, 4-low anti-elite sentiment)

(1-new political beliefs, 5- traditional political beliefs)

Social Trust: Measured on an additive scale of 30 that was collapsed into 6. A3. “Generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people?”

Independent Variables

A4. “Do you think that most people would try to take advantage of you if they got the chance, or would they try to be fair?”

Social Integrations Variables

A5. “Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves?”

Rural/Urban Living

01-05=01

06-10=02

11-15=03

F14. “Which phrase on this card best describes the area where you live?”

Big city-suburb of a city=01

16-20=04

A Small Town or City-A Farm or Home in the Countryside =0

21-25=05

26-30=06

(1-urban, 0-rural)

(1-low trust, 6-high trust)

Dom Pérignon vs. Miller Lite: Investigating Anti-Elite Sentiments at the Individual Level Political Trust

07-Every day

Measured on an additive scale of 30 that was collapsed into 6.

(1-lowest social interaction, 7-highest social interaction)

B2. “How much you personally trust each of the institutions I read out…[country]’s parliament?”

Self-identified Social Status Variables

B3. “How much you personally trust each of the institutions I read out…the legal system?” B6. “How much you personally trust each of the institutions I read out…political parties?”

Education F16. “About how many years of education have you completed, whether full-time or part-time?

0-12=0

13-30=1

(1-highly educated)

01-05=01

06-10=02

11-15=03

16-20=04

21-25=05

26-30=06

Gender is taken from self-reported gender, using question F2. The question was recoded to address Male or Female gender.

(1-low political trust, 6- high political trust)

0-Female

Employment

01-Male

F17c. “Which of these descriptions best describes your situation (in the last seven days)?”

Resentment

Gender

01-In paid work

0=Other Nativist Sentiment Measured on an additive scale of 20 that was collapsed into 4 B33. “Would you say that [country]’s cultural life is generally undermined or enriched by people coming to live here from other countries” B34. “Is [country] made a worse or a better place to live by people coming to live here from other countries?

D17b. “Compared to people like yourself who were born in [country], how do you think the government treats those who have recently come to live here from other countries?”

The same-much worse=0

Much- A little better= 01

(1-high resentment, 0-low resentment) Age Age is calculated from self reported year of birth, using question F3.

01-25=01

01-05=01

26-50=02

06-10=02

51-75=03

11-15=03

76-100=04

16-20=04

(1-younger, 4-older)

(1-high nativist sentiment, 4-low nativist sentiment)

Social Interaction

Income Perception

C2. “How often do you meet socially with friends, relatives, or work colleagues?

F42. “Which of the descriptions on this card comes closest to how you feel about your household’s income nowadays?”

01-never

01-Living comfortably on present income

02-Less than once a month

02-Coping on present income

03-Once a month

03-Finding it difficult on present income

04-Several times a month

04-Finding it very difficult on present income

05-Once a week

06-Several times a week

(1-positive income perception, 4-highly negative _ income perception)

Pi Sigma Alpha Undergraduate Journal of Politics

REFERENCES Anderson, Christopher J., and Yuliya V. Tverdova. 2003. “Corruption, Political Allegiances, and Attitudes Toward Government in Contemporary Democracies.” American Journal of Political Science 47(1): 91-109.

Polk, Jonathan, and Jan Rovny. 2017. “Anti-Elite/Establishment Rhetoric and Party Positioning on European Integration.” Chinese Political Science Review 2(3):356-371. Seyd, Ben. 2015. “How do Citizens Evaluate Public Officials? The Role of Performance and Expectations on Political Trust.” Political Studies 63:73-90.

Arzheimer, Kai, and Elisabeth Carter. 2006. “Political Opportunity Structures and Right-Wing Extremist Party Success”. European Journal of Political Research 45(3):419-443.

Tierney, J., 2004, January 11. The 2004 Campaign Political Points. https://www.nytimes.com/2004/01/11/us/the-2004-campaignpolitical-points.html (Accessed August 2, 2018).

Betz, Hans-Georg Betz, and Carol Johnson. 2004. “Against the Current-Stemming the Tide: The Nostalgic Ideology of the Contemporary Radical Populist Right.” Journal of Political Ideologies 9(3):311-327.

Van der Meer, Tom, and Armen Hakhverdian. 2017. “Political Trust as the Evaluation of Process and Performance: A Cross-National Study of 42 European Countries.” Political Studies. 65(1):81-102.

Center for Disease Control and Prevention. “Vaccines Do Not Cause Autism.” https://www.cdc.gov/vaccinesafety/concerns/autism.html (Accessed July 10, 2018). CNN Wire Staff. 2011. Retracted Autism Study an ‘Elaborate Fraud,’ British Journal Finds. CNN.com. January 5. http://www.cnn.com/2011/HEALTH/01/05/ autism.vaccines/index.html (Accessed July 10, 2018) Dalton, Russell J. 2004. Democratic Challenges, Democratic Choices. Oxford: Oxford University Press.

Werts, Han, Peer Scheepers, and Marcel Lubbers. 2013. “EuroScepticism and Radical Right-Wing Voting in Europe, 2002-2008: Social Cleavages, Socio-Political Attitudes and Contextual Characteristics Determining Voting for the Radical Right.” European Union Politics 14(2):183-205. Ziller, Conrad, and Thomas Schübel. 2015. “The Pure People” versus “the Corrupt Elite”? Political Corruption, Political Trust and the Success of Radical Right Parties in Europe. Journal of Elections, Public Opinion and Parties 25(3):368-386.

Denvir, Denvir. 2016. “Why Women are still Voting for Trump, Despite his Misogyny.” The Vox. October 25. https://www.vox. com/conversations/2016/10/25/13384528/donald-trump-womenstephanie-coontz (Accessed November 25, 2017). Gidron, Noam, and Peter A. Hall. 2017. “Populism as a Problem of Social integration.” Working Paper. https://scholar.harvard.edu/ hall/publications/populism-problem-social-integration (Accessed August 2, 2018). Hessami, Zohal. 2011. “Globalization’s Winners and Losers-Evidence from Life Satisfaction Data, 1975-2001.” Economics Letters 11(3):250-253. Hilger, Nathaneial G. 2016. “Why Don’t People Trust Experts?” The Journal of Law and Economics 59(2):293-311. Hochschild, Arlie Russell. 2016. Strangers in Their Own Land: Anger and Mourning on the American Right. La Vergne: The New Press. Inglehart, Ronald F. and Pippa Norris. 2016. “Trump, Brexit, and the Rise of Populism: Economic Have-nots and Cultural Backlash.” Faculty Research Working Paper Series. https://research.hks.harvard. edu/publications/getFile.aspx?Id=1401 (Accessed July 10, 2018). McCombs, Maxwell E., and Donald L. Shaw. 1984. “The AgendaSetting Function of the Press.” In Media Power in Politics ed. Doris A. Graber. Washington D.C.: CQ Press, 63-72. Mclaren, Lauren M. 2012. “The Cultural Divide in Europe: Migration, Multiculturalism, and

Political Trust.” World Politics 64(2):199-241.

McQuail, D. 1984. “The Influence and Effects of Mass Media.” In Media Power in Politics ed. Doris A. Graber. Washington DC: CQ Press, 36-54. Mudde, Cas. 2007. Populist Radical Right Parties in Europe. Cambridge: Cambridge University Press. Müller, Jan-Werner. 2016. What is Populism? Philadelphia, Pennsylvania: University of Pennsylvania Press. 54