TECHNOLOGY ADOPTION AND IMPACT ON CROP PRODUCTIVITY, BENEFICIARY’S INCOMES AND EFFICIENCIES IN WEST-AFRICA: CASES OF CASSAVA FARMERS IN GHANA, BAKERS AND MILLET FARMERS IN SENEGAL.
A PhD Field-Work Report Submitted by
Kodjo Kondo University of New England, Armidale 2351 NSW-Australia 9/15/2014
ACKNOWLEDGEMENT Before anything, I would like to express my sincere gratitude to the Australian Government, the University of New England (UNE), the West and Central African Council for Agricultural Research and Development (CORAF/WECARD) and to the implementing partners of the West Africa Agricultural Productivity Program (WAAPP) in Ghana and Senegal for the resources and the excellent facilities provided in the course of the field-work.
Thank you.
Page | 1
ACKNOWLEDGEMENT
1
ACCRONYMS
3
I. INTRODUCTION
4
1.1 BACKGROUND INFORMATION 1.2 RESEARCH OBJECTIVES 1.3 FIELD REPORT OUTLINE
4 5 5
II. RESEARCH TOPICS, QUESTIONS AND HYPOTHESES
6
III. METHODOLOGY 3.1 INTERVIEWS 3.2 FOCUS GROUP DISCUSSIONS 3.3 QUANTITATIVE SURVEYS 3.3.1 MILLET FARM-HOUSEHOLDS SURVEY IN SENEGAL 3.3.2 BAKERS’ SURVEY IN DAKAR REGION IN SENEGAL 3.3.3 CASSAVA FARM-HOUSEHOLDS SURVEY IN GHANA.
8 8 10 10 10 13 14
VIII. ROADMAP FOR REMAINING ACTIVITIES
21
VI. WORKSHOPS ATTENDED & OTHER ACTIVITIES UNDERTAKEN
23
V. DIFFICULTIES ENCOUNTERD AND RECOMMENDATIONS
24
APPENDICES
25
ANNEX 1. DOCUMENTS PRODUCED ANNEX 2. LIST OF MILLET HOUSEHOLD SURVEY COMMUNITIES IN SENEGAL ANNEX 3. LIST OF CASSAVA HOUSEHOLD SURVEY COMMUNITIES IN GHANA
25 26 28
Page | 2
ACCRONYMS ACRONYM
DEFINITION
ANCAR
Agence Nationale de Conseil Agricole et Rural
ASPRODEB
Association pour la Promotion du Développement à la Base
CORAF/WECARD
West and Central African Council for Agricultural Research and Development
CSIR/CRI
Crops Research Institute of the Council for Scientific and Industrial Research.
DAES
Directory of Agricultural Extension Services/MOFA
DCS
Directory of Crop Services/MOFA
FBO
Farmer Based Organisation
FNBS
Federation Nationale des Boulangeries du Senegal / National Association of Bakers in Senegal
FNRAA
Fond National pour la Recherche Agricole et Agroalimentaire du Senegal
FONGS
Fédération des ONGs du Sénégal
GAP
Good Agronomic Practice
ISRA/BAME
Bureau des Analyses Macro-Economiques de l’Institut Sénégalais de la Recherche Agricole
ITA
Institut de Technologie Alimentaire
MDES
Minimum Detectable Effect Size
M&E
Monitoring and Evaluation
MOFA
Ministry of Food and Agriculture
PCU
Program Coordination Unit
RTIMP
Root and Tubers Improvement and Marketing Program
UGPM
Union des Groupements des Producteurs de Meckhé
UNCAS
Union Nationale des Coopératives Agricoles du Sénégal
UNE
University of New England
WAAPP
West Africa Agricultural Productivity Program
WASP
West Africa Seed Program
Page | 3
I. INTRODUCTION 1.1 Background Information In March 2014, the West and Central African Council for Agricultural Research and Development, hereafter referred to as CORAF/WECARD, and Mr Kodjo Kondo - PhD candidate at the University of the New England in Armidale – Australia, referred to as the “PhD Candidate”, entered into a six-month agreement to document the causal effects of the technologies’ adoption under the World Bank funded West Africa Agricultural Productivity Program (WAAPP) and the USAID and DFAT funded Seed systems projects’ generated technologies. The joint venture aimed at allowing both parties to: o Share understanding on the various impact evaluation designs; o Decide on the technologies to assess, the target crops and countries; o Allow a critical review of the research questions and hypotheses by the Scientists at CORAF; o Review and pilot-test the methods and tools proposed; o Document lessons learned, tools and methods used; and o
Disseminate research findings through conference papers, journal articles and a thesis.
On arrival in Dakar in March 2014, the PhD candidate exchanged with the Director of programs, the WAAPP Coordinator and the Chief of Party of the USAID-funded West Africa Seed Program (WASP). The interactions helped precise the internship objectives and to collect all projects documents for initial review. The desk review resulted not only in the identification of the technologies being disseminated in six initially proposed countries (Ghana, Senegal, Mali, Nigeria, Burkina Faso and Cote d’Ivoire) but also in the formulation of research topics, questions and hypotheses. These outcomes were documented in an inception report submitted to CORAF/WECARD on April 9, 2014. Elements of this report were presented during a seminar that gathered 14 CORAF/WECARD Scientists at the Executive Secretariat on April 11, 2014. Recommendations from participants and further discussions with the WAAPP Coordinator helped reduce the scope of the study from six countries to two, with focus on millet in Senegal and cassava in Ghana. In Senegal, a number of existing and newly released improved millet varieties including Souna 3 were promoted by WAAPP-Senegal implementing agents along with good agronomic practices (GAP) in five regions of the “bassin arachidier” agro-ecologic zone of Senegal. Dissemination mechanisms used include demonstrations, radio programs and sensitization in market places. Besides technology development and dissemination, institutional arrangements aiming at valuing local cereals in bakeries and patisseries were also experimented. These arrangements consisted in linking up farmers from two unions with four millers in Dakar for the supply of quality millet and the production of fine millet flour appropriate for baking. Institutional arrangements also exist between the millers and bakers who buy the millet flour produced and Page | 4
mixt it with wheat flour, at 15% recommended dose, for composite bread baking. Substantial supports in training and equipment were provided to all actors in the millet value chain in Senegal by the project. Likewise, WAAPP-Ghana released four new improved cassava varieties which are being disseminated with existing ones along with GAPs across the country through demonstrations, field-days, video programs, community radios and E-agriculture.
1.2 Research Objectives The retained research topics aim at identifying the dissemination mechanism with high impact on the adoption of the improved varieties promoted as well as assessing the impact of adoption on millet and cassava productivities, farmers’ incomes, technical and allocative efficiencies. In addition, the study will assess the determinants of the adoption of the composite flours by the bakers in the Dakar region in Senegal. Finally it will throw light on the impact of the adoption of the composite flours on bakers’ business volumes, technical and allocative efficiencies.
1.3 Field Report Outline The current field-work report presents the research questions to be addressed by the topics retained; the hypotheses to be tested; the methods and tools used with their limitations; the difficulties encountered and some preliminary recommendations. It ends with a proposed road-map for the conclusion of the remaining activities. Results and discussions addressing the specific research questions will be presented in the form of conference papers, journal articles and a final doctoral thesis. A draft research report with preliminary results will be sent to CORAF after data checking, cleaning and processing are completed. Electronic copies of all documents produced during the field-work including trip reports and survey instruments are attached in annex 1.
Page | 5
II. RESEARCH TOPICS, QUESTIONS AND HYPOTHESES Table 1 below summarises the final research topics and questions retained and the hypotheses to test. Table1. Selected Research Topics with Research Questions and Hypotheses COUNTRY
RESEARCH TOPIC
RESEARCH QUESTIONS
HYPOTHESES
SENEGAL
1. Adoption of improved millet varieties and agronomic practices and impact on millet productivity and farmers’ income and efficiencies in the department of Tivaouane - Senegal.
Q1. What are the factors affecting farmers’ decision to adopt the improved millet varieties disseminated?
The dissemination mechanisms have a significant and positive impact on the adoption of the improved millet varieties
Q2. What are the factors affecting farmers’ decision to adopt one or more elements of the good agronomic practices (GAP) disseminated to enhance millet productivity
The dissemination mechanisms have a significant and positive impact on the adoption of more elements of the GAPs disseminated.
Q3. How does the impact of demonstration compare with the impact of the other demand creation and dissemination mechanisms used?
The impact of the demonstrations is significantly higher than the impact of the other dissemination mechanisms used.
Q4. What is the impact of the adoption of both the improved variety and GAPs on millet productivity, farmers’ income, technical and allocative efficiencies?
The adoption of the improved varieties and the GAPs has a significant and positive impact on millet productivity and farmers’ income.
Q5. What is the impact of the institutional arrangement with millers and bakers as well as the impact of millet cleaning and storage equipment supplied on beneficiaries’ income?
The institutional arrangements with millers and bakers have a positive and significant impact on the income of the millet producers.
(Millet)
The adoption of the improved varieties and the GAPs significantly decreases eventual inefficiencies among farmers.
Q5. How does the impact on income translate in terms of livelihood indicators (children education of, health…)? 2. Adoption of composite flour and impact on bakers’ business
Q1 What are the determinants of adoption of the composite flour by the bakers in the Dakar region?
The training programs, membership to the national association of bakers (FBNS) and the participation in business fares have a positive impact on the adoption of the composite flour by the bakers. Page | 6
COUNTRY
RESEARCH TOPIC
RESEARCH QUESTIONS
volumes, technical and allocative efficiencies in Dakar region - Senegal
GHANA (Cassava)
3. Adoption of improved cassava varieties and good agronomic practices and impact on cassava productivity and farmers’ income and efficiencies in the Ashanti and Brong Ahafo regions - Ghana.
HYPOTHESES The institutional arrangements between the farmers and millers on one side and the millers and bakers on the other side, affects positively the adoption of the adoption of the composite flour by the bakers.
Q2. What is the impact of the adoption of the composite flour on the bakers’ business volume, technical and allocative efficiencies?
The adoption of the composite flour technology has a significant and positive impact on the bakers’ business volume, technical and allocative efficiencies.
Q1. What are the factors affecting famers’ decision to adopt the improved cassava varieties and GAP promoted?
The dissemination mechanisms (demos and field-days, planting materials production and distribution, e-agriculture, radio, video…), the involvement of farmers in innovation platforms/value chains and the production level of the varieties have a significant and positive impact on the adoption of the new varieties and GAP promoted.
Q2. How does the impact of demonstration compare with the impact of the other dissemination mechanisms (eagric, radio, video, planting material multiplication, farmer-to-farmer, etc.)?
The demonstrations have more significant and positive impact on adoption of the improved varieties and GAP than the other dissemination mechanisms.
Q3. What is the impact of the adoption of both the improved varieties and the GAP on cassava productivity?
3.1 The adoption of both the improved varieties and GAP has a significant and positive impact on cassava productivity.
Q4. What is the impact of the adoption of both the improved varieties and the GAP on farmers’ income, technical and allocative efficiencies?
4.1 The adoption of both the improved varieties and GAP has a significant and positive impact on farmers’ income.
Page | 7
III. METHODOLOGY A sequential and interactive mix-method approach to research was adopted. The approach combines interviews with key implementing partners and focus group discussions with farmers. Outcomes from the qualitative surveys helped in the design of the instruments (sampling and questionnaires) used in the quantitative surveys. The quantitative surveys are very important in addressing the initial research questions and hypotheses formulated.
3.1 Interviews Fourteen (14) semi-structured interviews were held in Senegal and Ghana with WAAPP Program Coordination Units (PCU) personnel, key implementing partners and beneficiary bakers and farmer Based organisations (FBO). The interviews with the WAAPP-PCU helped confirm the list of technologies disseminated; gain national buy-in in the research activities; share understanding on the research methods, tools and responsibilities; identify the treated and the comparison/control districts and communities; gather relevant literature on the project activities and databases (baseline and follow-up surveys) as well as mobilising human, logistic and financial resources for the data collection and supervision. The interviews with the implementing agents helped confirm the list of the dissemination mechanisms used, the elements of the good agronomic practices promoted, the relevant farming practices related to the selected crops mostly on the inputs and equipment used in bakeries. All the interviews were audio-recorded with prior consent from participants as per the ethical requirements. The interviews were very useful in the design of the questionnaires for the quantitative surveys. Table 2 presents the list of the institutions where staff members were interviewed, the places and dates. These interviews were followed by four focus group discussions with millet farmers in Senegal and cassava farmers in Ghana.
Page | 8
Table 2: Institutions visited and results achieved Country
Institution
Place, date
Results achieved
Senegal
PCU/ WAAPPSenegal
Dakar, 25/04/14
National buying-in in the research project achieved; Contacts facilitated with implementing agents. Sampling frames established; Projects related reports collected.
ASPRODEB
Dakar, 30/04/14
Contacts facilitated with farmers unions (FONG and UNCAS) involved in the local cereals valorisation project and with FNBS.
FNBS
Dakar, 19/05/14
Bakery industry related activities learnt; Type of equipment, inputs, outputs and cost and prices applied known; Challenges and opportunities in the success of the project gathered.
FONGS
Thies, 20/05/14
Sampling frames established; List of varieties promoted and dissemination mechanisms used confirmed; Contact facilitated with farmers for focus group discussions.
UNCAS
Thies, 20/05/14
Same as above.
ISRA/BAME
Dakar, 10/06/14 to Sept 14.
Research objectives and tools shared; Survey budget developed; Enumerators, supervisors and data entry agents identified; Logistical support received in testing of questionnaires and follow-up of data collection.
ITA
Dakar, 17/06/14
In-depth knowledge gained on the local cereal valorisation project
ANCAR
Dakar, 17/06/14
List of varieties and dissemination mechanisms used confirmed
UGPM
MeckhĂŠ, 22/06/14
In-depth knowledge gained on the local cereal valorisation project and constraints to the adoption of the equipment supplied; Constraints and opportunities in the improved seed production and distribution understood.
Ghana
PCU /WAAPPGhana
Accra, 29/07/14 to 04/08/14
National buying-in in the research project achieved; Contacts facilitated with implementing agents. Sampling frames established; Projects related reports collected. Survey budget prepared and approved; Enumerators, supervisors and data entry agents identified; Survey instruments validated Logistical supports received in all research activities.
DAES/MOFA
Accra, 29/07/14
List of varieties promoted and dissemination mechanisms used confirmed; Districts with demonstrations in 2010 received
DCS/MoFA
Accra, 01/08/14
List of varieties promoted and dissemination mechanisms used confirmed; Documents on cassava planting materials production and certification collected and
CSIR-CRI
Kumasi, 06/08/14
Support in identification of enumerators, venue and logistics for the training of enumerators and supervisors.
DDAES
Foase
List of varieties and dissemination mechanisms used confirmed. Support in arranging for farmers and venue for the focus group discussions in Foase and Dida.
Page | 9
3.2 Focus group discussions Four focus group discussions involving 44 participants were organised in Thies Region in Senegal and in the Ashanti Region in Ghana. The group discussions resulted in a better understanding of the local farming systems; the listing of varieties grown in the areas; the type of equipment used plus other relevant information needed for the improvement of the questionnaire for the quantitative surveys. Table 3 summaries the date and the number of participants in the different focus groups. Table 3: Number of participants in the focus groups Country
Communities
Senegal1
Mboulouctene Seko Tassette 2 Ghana Foase Dida Total number of participants
Type of community Control Control Control Treatment
Date 21/05/14 22/05/14 07/08/14 07/08/14
Number participants 11 7 13 13 44
of Number of women 4 3 5 6 18(41%)
3.3 Quantitative Surveys Unlike the interviews and focus groups, the quantitative surveys aimed at collecting from representative samples of farm-households and bakers, adequate and sufficient physical, socio-economic, market, institutional, technological, inputs and production data to correctly address the research questions and test the hypotheses. Activities implemented in this area included the establishment of the retrospective evaluation designs and sampling frames; the choice of the appropriate sampling techniques; the finalisation of the questionnaires; the selection and training of enumerators and supervisors; the pretesting of the survey instruments; the surveys and their follow-up; the data checking and correction; the design and testing of the templates for data entry; the recruitment of data entry agents; and the supervision of data entry followed by data cleaning. 3.3.1
Millet Farm-Households Survey in Senegal o Evaluation and Sampling Designs, Power Analysis
The evaluation design in the millet farm households’ survey in Senegal consisted of the identification of the treatment and the comparison/control communities. To accurately address the research questions, the ideal In Senegal, the treatment community is defined as a community with demonstrations on improved millet varieties and GAPs as well as the presence of a farmer-based-organisation involved in the local cereals valorisation project. The two communities visited for the focus groups have FBOs but have not held the demonstrations on the improved millet varieties and GAPs. 1
In Ghana, the treatment community is a community that have had a WAAPP demonstration on improved cassava varieties prior to 2013. 2
Page | 10
treatment communities should be the ones with demonstrations on the new varieties and good agronomic practice as well as with grassroots-farmer based organisation (FBO) member of a union involved in the local cereals valorisation project. A control community in this case would be a community without demonstrations and FBO involved in the project. The Monitoring and Evaluation (M&E) Officer of the Senegalese WAAPP-PCU provided us with the list of the communities where the demonstrations plots on the new varieties and GAPs were implemented. A database containing all the communities in Senegal was also provided. In addition to this information, the lists of the grassroots FBOs and communities involved in the first phase of the valorisation of local cereals project in Senegal were also collected from FONGs and UNCAS in Thies. These pieces of information helped establish the study sampling frame composed of the list of communities tagged as community with demonstrations, community with FBO, community with demonstrations + FBO and control community. Based on the sampling frame established, it was obvious, as depicted in table 4, that it is only in Thies region in general and Tivaouane department in particular that there is the ideal treatment group composed of six communities. The decision was then made to have the study conducted in that department. Table 4. Number of communities per type in Senegal N 1 2 3 4 5 Total
Region Diourbel Fatick Kaffrine Kaolack Thies
Demo 27 26 21 14 27 115
Number of communities per type FBO Demo + FBO 0 0 0 0 11 0 2 0 86 6 99 6
Control
In the absence of the list of households in the communities, a cluster sampling approach with optimal allocation (Spybrook et al, 20113) was adopted taking into account the budget allocated for the study and the intra-community correlation coefficient estimated with data from a recent adoption study and the budget per community and per household in a similar survey. The application of the formula provided by Spybrook et al (see document on the sampling design attached in line 6 of annex1) gives a sample size of six communities with 167 households to be surveyed per community. This gives a total sample size of 1,002 farm-households. However, due to the impossibilities of having 167 households in all communities, it was agreed to interview 42 households in 24 communities (six communities with demo+FBO, six communities with demo, six communities with FBO and six control communities). All six communities with demo+FBO were automatically selected and the other 18 communities were randomly selected from each group using 3
http://sitemaker.umich.edu/group-based/optimal_design_software
Page | 11
Stata. Nine replacement communities were also randomly selected to serve as survey community in case the total number of 42 households is not reached in a particular community. Appendix 2 contains the list of selected communities. Power analyses conducted using the optimal design software reveals a minimum effect size (MDES) of 19% on millet productivity to be detectable by the design with an 80% statistical power. o Finalisation of the Questionnaire The initial questionnaire designed was improved based on the insights gained from the interviews and the focus group discussions as well as from existing questionnaires used by WAAPP-Senegal and ISRA/BAME in adoption studies. o Identification and Training of Enumerators and Supervisors Eight Enumerators (composed mainly of masters’ level students) and one Supervisor were selected with the assistance of ISRA/BAME. The Enumerators were taken through one day class-room training. The training allowed the PhD candidate to share the study objectives and research questions with the participants followed
by
thorough
review
and
a
amendment
of
the
questionnaire. The ethical issues and related documents (information sheet for
participants
and
consent
forms) were also
Picture 1. Team of enumerators for the farmers’ survey
presented and discussed. Moreover, the enumerators were instructed on the way of getting the list of households in the selected communities and how to use the random number table in sampling of 42 households in those communities that may have more than 42 households. o
Data Collection and Supervision
A 30-day agreement was reached with the enumerators to complete the data collection. A total number of 1,012 households were reached with a response rate of 100.4%. The surveys were supervised by both the PhD candidate and the ISRA/BAME proposed Supervisor. Hand-on supports were given to the Enumerators
Page | 12
and vetting meetings held with them to correct mistakes and enhance their understanding of the questionnaire. 3.3.2
Bakers’ survey in Dakar region in Senegal o Evaluation and Sampling Designs
For the bakers’ survey, the treated group is made of 50 bakers involved in the first phase of the local cereals valorisation project. These bakers have been supplied with equipment and trained in the composite
flour
formulation
and
baking
techniques. The sampling frame for the treatment group was provided by the Chairman of the bakers’ association of Senegal (FNBS). However, out of the 50 bakers, eight were out of Dakar (Mbour, Tabakounda, Kaolack) and two shut down leaving us out with 40 treated bakers. Among the 40 treated bakers only 28 actually participated in the survey. Refusal and time constraints are the reasons provided by the
Picture 2. Baking equipment supplied
enumerators for the 12 treated bakers not surveyed. In fact, the survey time coincided with the critical baking period during the Ramadan. The control group is composed of 122 bakers not involved in the project and identified through a snow-ball approach. This approach consisted in asking the treated bakers the contacts and locations of the nearby bakeries they know. o Finalisation of the Questionnaire The questionnaire for the quantitative surveys with bakers was designed based on the information gathered through the semi-structured interview held with the Chairman of the Bakers’ association. This questionnaire was pretested and improved during the training of the enumerators.
Page | 13
o Identification and Training of Enumerators and Supervisors for the bakers’ survey Three Enumerators and one Supervisor were identified with the assistance of ISRA/BAME. They have been trained by the PhD candidate at ISRA/BAME in Dakar. The training consisted in a one day-classroom based program on July 16, 2014 during which the study objectives were shared, the ethical considerations discussed and the questionnaire thoroughly reviewed and corrected. The second day of the training consisted in the pretesting of the questionnaire and it amendment before printing and copying. The pretesting resulted in the removal from the questionnaire of all questions related to patisseries in other to reduce the survey duration. o Data Collection and Supervision The bakers’ survey in Dakar region lasted for a month from July 23rd to August 21st 2014. The supervision consisted in providing guidance to the Enumerators and assisting them better master questionnaire so to complete them appropriately. Critical points in the questionnaire for the bakers’ survey are related to the matching of the measurement units to the quantities of labour, inputs and bread recorded. o Data Entry and Checking Six data entry agents were planned for the entry of both farmer and bakers’ survey data. However, due to time constraints, two more agents were added. The
data
are
being
entered from September 2nd to 11th into an SPSS based database designed and tested by the PhD candidate. The data entry
Picture 3. Data entry in Senegal is yet to be completed and will be followed by checking for entry mistakes and missing data. 3.3.3
Cassava farm-households survey in Ghana. o Evaluation and Sampling Designs, Power Analysis
In Ghana, the sampling frame used consisted of the list of districts with WAAPP demonstrations in nine Regions in Ghana (Ashanti, Brong Ahafo, Central, Eastern, Greater Accra, Northern, Upper West, Volta and Western Region). This list was provided by WAAPP-Ghana M&E Officer. The demonstrations on improved cassava varieties were implemented from 2010 across the regions with the existing improved varieties Page | 14
before the four WAAPP varieties were released in 2012. Due to the shortness of the study time and the limited resources available, the Research Team composed of the M&E Officer, the PhD Candidate and the Consultant involved in all WAAPP-Ghana baseline and impact studies selected the Ashanti and Brong Ahafo regions for the study. These two regions are dominant cassava producing zones of Ghana. The treatment group in these two regions are the districts with demonstrations from 2010 to 2012. The Team assumed that only the demonstrations plots established in 2010, 2011 and 2012 will affect the knowledge and the adoption of the improved varieties in 2013. The ideal treatment district would have been those with demonstrations in two or three successive years to enhance knowledge on the characteristics of the improved varieties and their adoption by the farmers.
Picture 4. Improved Cassava Demonstration Plot in Tano South, Brong Ahafo Region (picture taken by Danquah Augustine, WAAPP-Ghana M&E Officer)
Unfortunately we could not have such districts in the sampling frame. The Team then purposely selected six districts (three in each region). The selected districts have had demonstrations either in 2010, 2011 or 2012. Having identified the treated districts, the Research Team generated the list of all communities in these districts from Ghana districts’ website (http://www.ghanadistricts.com/home/). With the assistance of the MoFA district Directors, the communities where the demonstrations plots were established have been identified. As achieved in Senegal, a cluster sampling with optimal allocation combined with power analysis was performed to determine the number of communities to include in the survey and the number of households to survey per community.
Page | 15
Using the budget allocated for the study and the intra-district correlation coefficient estimated with data supplied by the M&E Officer and the budget per community and per household in the 2014 baseline survey, the application of the formula provided by Spybrook et al gives a sample size of six communities with 41 households to be surveyed per community. This gives a total sample size of 239 farm-households. Because the large size (57%) of the minimum effect detectable (MDES) by this design with a 80% power, the PhD Candidate recommended the implementation of the survey in 14 communities with 41 households per community (574 in total) with a reduced budget per community and per household and a MDES of 30%. Annex 3 presents the list of the sampled communities. It is to be noted that power analysis is mostly appropriate in perspective evaluation frameworks as it helps design the evaluation ex-ante and determine the appropriate sample size and the associated MDES and power. o
Finalisation of the questionnaire
The questionnaire used in Senegal was translated into English and amended according to the realities in Ghana and the insights gained from the interviews and the focus group discussions. A seminar was organised in Accra on August 13, 2014 in WAAPP-Ghana E-agricultural Conference room to validate the research objectives and amend the questionnaire designed. The three-hour seminar brought together the technical staff at WAAPP-Ghana and the implementing agents based in Accra. The revised questionnaire was reviewed, pretested and finalised during the training of enumerators and supervisors in Kumasi. o Identification and training of Enumerators and Supervisors Eight Enumerators (four MOFA Extension Agents and four Agricultural National Service Students) and two Supervisors (MOFA district level M&E Officers) were selected with the assistance of WAAPP-M&E Officer and the Team leader of the Socio-Economic Department of the Crop Research Institute (CSIR/CRI) in Kumasi. The overall supervision, gathering and entry of the completed questionnaires were contracted with the WAAPP-Ghana service provider Consultant. The enumerators and Supervisors were trained at CRI-Kumasi on the 14 of August 2014. The classroombased training program was followed by the pretesting of all survey instruments in Besease in Ejusu Juaben Municipal on August 15, 2014. During the training and pretesting of instruments, a special attention was giving to the technique of recruitment of the respondents in the communities. Large communities are to be divided into zones followed by a random selection of a zone. A complete enumeration of households in the selected zone was recommended by using the household listing. The form allowed the categorization of the households in agricultural and nonagricultural categories. It also helped identify the farmers who participated in the demonstrations the treated communities. Based on the list produced, the nonagricultural households are to be removed from the list. In the treated communities, the enumerators were called to randomly sample 23 farmers who participated in Page | 16
demonstrations and 20 households who don’t. In the control communities, they will just randomly sample 43 households from the agricultural households using a random table number. It is right noting that the number of households to be surveyed by community as per the power analysis performed is 41. The number 43 was unwillingly and mistakenly introduced during the training program leading to an increase in the overall sample from 574 to 602 households without effect on the survey budget. o
Data collection and supervision
The surveys in Ghana were completed in two weeks from August 18 to September 3, 2014. Data quality assessment was assured by PhD Candidate and WAAPP-Ghana Consultant as well as by the two districtlevel M&E Officers of MOFA. Supports were given to the enumerators from the beginning of the survey to correct mistakes, resolve difficulties and enhance understandings.
Picture 6. Checking of the completed questionnaire during follow-up visit in Gyansa
One particular problem raised by the Enumerators during the follow-up visits is the one related to the area planted with cassava ad the production measurement units. Most of the farmers don’t have a precise idea about the size of their plots and none of the production units options in the questionnaire apply to them. This imprecision in measurement units might bias the results if not appropriated addressed. The Research Team agreed to conduct accurate cassava yield surveys taking into account the different varieties and the diversity in the farming systems (mix and inter-cropping patterns). The PhD Candidate was tasked to suggest a form for collecting these data by upgrading the tools used by WAAPP-Ghana in previous yield surveys. The budget for this additional survey will be supported through WAAPP-Ghana M&E activities.
Page | 17
Another problem encountered during the follow-up mission is respondents asking Enumerators and ourselves for seating allowances as commonly practised by other projects and researchers. o Data Entry and Checking The data entry is currently on-going with five entry agents identified and supervised by the Consultant while the PhD Candidate is busy supervision the same activity in Senegal and completing the narrative and financial reports. Table 5 summarises the key features of three quantitative surveys.
Page | 18
Table 5. Key features of three quantitative surveys. Target regions Treatment
Sampling frame
Millet farm-households Survey - Senegal Tivaouane department in Thies region o o o o o o
Communities with demonstrations Communities with FBO involved in local cereal valorisation project; Communities with both demos and FBOs. List of districts (communauté rurale) and communities in the “bassin arachidier” (Thies, Diourbel, Fatick, Kaffrine, Louga) ; List of communities with demonstrations; List of communities with FBOs member of FONGS and UNCAS.
Sampling design techniques
Cluster sampling with optimal allocation design
Evaluation design
24 communities in two districts 6 communities with demo + FBO ; 6 communities with demo ; 6 communities with FBO ; 6 control communities.
Estimated sample size Response rate Power analysis
Bakers’ survey - Senegal Dakar region (Dakar, Pikine, Ruffisque, Guedjewaye). Bakeries involved in the first phase of the local cereals valorisation project.
Cassava farm-households survey -Ghana Ashanti and Brong Ahafo regions.
List of bakeries involved the first phase of the local cereals valorisation project
o
Communities with WAAPP and/or Root and Tubers Improvement and Marketing Project (RTIMP) demonstrations prior to 2013.
o o
List of districts in the Ashanti and Brong Ahafo regions List of all communities in the selected districts of the study List of communities with WAAPP demonstrations in the selected districts.
Census of the first phase bakeries and comparison bakeries identified through a snow-ball approach. 40 treated bakeries 110 control bakeries.
Cluster sampling with optimal allocation design.
1,008 100.4% Performed
28 treated & 122 Control Treated (70%) Control (111%) Not applicable
574 106% Performed
Validation of survey instruments designed
Instruments shared with partners (validation seminar not held)
Instruments shared with partners (validation seminar not held)
Survey instruments presented and validated by WAAPP-Ghana PCU and implementing partners in Accra on August 13, 2014.
Identification of Enumerators and supervisors
8 Enumerators University students identified with the assistance of ISRA/BAME.
3 Enumerators University students identified with the assistance of ISRA/BAME. One supervisor from ISRA/BAME.
8 Enumerators 4 extension agents identified by WAAPPGhana M&E Officer 4 National Service Agric-Students identified
14 Communities in six districts 8 treated communities 6 control communities
Page | 19
Training of Enumerators Pretesting of survey instruments Duration of data collection State of data collection Data entry Data checking
Millet farm-households Survey - Senegal One Supervisor from ISRA/BAME.
Bakers’ survey - Senegal
Cassava farm-households survey -Ghana by the socio-economics department of CSIRCRI in Kumasi. Two District levels Extension Officers’ Supervisors One Service provider Consultant at WAAPP-Ghana.
ISRA/BAME-Dakar on July 15, 2014.
ISRA-BAME-Dakar July 16, 2014 Dakar – July 17, 2014 One month. July 23 to August 21, 2014 Completed Completed On-going
CSRIR-CRI – Kumasi, August 14, 2013 Besease, August 14 2013 Two weeks. Aug 18 to Sept 3, 2014. Completed Completed Not started
One month. July 15 to August 14, 2014. Completed Completed On-going
Page | 20
VIII. ROADMAP FOR REMAINING ACTIVITIES The remaining activities were related to the completion of the data entry in Ghana, data checking for entry mistakes and missing values followed by the scanning of all questionnaires prior to their destruction. Preliminary analyses will be performed and results shared with stakeholders in Senegal and in Ghana before the PhD candidate returns to Australia. The transcription of the audio-recordings and in-depth analyses of the quantitative data addressing the research questions and hypotheses will be performed in Australia. The study results will be communicated and published in conference papers, journal articles and in a doctoral thesis. Before that, a draft research report addressing the research questions will be produced, shared and discussed with CORAF/WECARD by December 31st 2014. Table 6 highlights the deadlines for the remaining activities. Table 6. Remaining tasks with deadlines #
Activity
Place
Deadline
1
Reports writing and acquittal for advances received (CORAF and DFAT)
Senegal
12/09/2014
2
Completion of data entry, checking and cleaning in Senegal
Dakar-Senegal
17/09/2014
3
Scanning of questionnaires and destruction of hard copies by one of the data entry agent.
Senegal/Ghana
14/10/2014
4
Performance of primary analyses on the cleaned variables
Senegal
15/09/2014
5
Presentation of preliminary results to CORAF and WAAPP-Senegal Stakeholders
Dakar - Senegal
16/09/2014
6
Trip to Ghana
Accra-Ghana
18/09/2014
7
Acquit for finances with WAAPP-Ghana
Accra-Ghana
19/09/2014
8
Completion of checking and cleaning in Ghana.
Accra-Ghana
22/09/2014
9
Presentation of preliminary results to WAAPP-Ghana Stakeholders
Accra-Ghana
23/09/2014
10
Trip to Australia
Australia
23-26/09/2014
11
Rest
Australia
29-30/09/2014
12
Transcription of interviews and focus group discussions’ audio-recordings
Australia
03/10/2014
13
Translation of Senegal databases and questionnaires into English
Australia
03/10/2014
14
Quantitative data analyses addressing the research questionnaire and testing for hypotheses.
Australia
10/10/2014
15
Submission of an abstract for the 59th AAERES Conference in New Zealand
Australia
14/10/2014
16
Submission of a draft research report to CORAF with all research questions addressed, hypotheses tested and
Australia
31/12/2014 Page | 21
#
Activity
Place
Deadline
Australia
Anytime until 26/06/2016
findings discussed. 17
Publication of results in conference papers, journal articles and thesis.
Page | 22
VI. WORKSHOPS ATTENDED & OTHER ACTIVITIES UNDERTAKEN During the time spent at CORAF, I attended staff meetings, workshops and training programs that are, in most cases, very relevant to the current study. Upon arrived in Dakar, I participated from March 17 to 22, 2014 in a regional workshop organised for all WAAPP M&E Officers and Service Providers on the harmonisation of methodologies used in baseline surveys and follow-up impact evaluations. The workshop was facilitated by Consultants from the School of Statistics in Cote d’Ivoire. Parallel training sessions were held at the same period in the use of the newly developed online WAAPP M&E database. From April 22 to 24, 2014, I was again privileged to attend a training program on impact assessment in Dakar organised by WAAPP and the Development Impact Evaluation (DIME) of the World Bank. The program gathered more than 50 participants from 10 countries. It helped me update knowledge in the various evaluation designs; get to know about the optimal design software used the in power analysis and discuss the research objectives and methods with Economists and Experts from the Bank. During the period, I also attended WAAPP review mission for Senegal and CORAF and the wrap-up meeting in Lome-Togo in May 2014. Among other activities implemented are: o The participation and reporting for the partnership meeting between IFPRI and CORAF (April 9-10, 2015); o The participation in the panel for the selection of an M&E Specialist and o The review of series of institutional documents including the 2nd Operational Plan.
Page | 23
V. DIFFICULTIES ENCOUNTERD AND RECOMMENDATIONS Challenges faced during the internship are related to administrative procedures, the work load and the evaluation designs in farmers’ survey in both Senegal and Ghana. In fact, the delay with administrative procedures in addressing the introduction letter to WAAPP partners jeopardised the smooth unfolding of the initial program established for Senegal. Partners became unwilling to collaborate and to release project information needed for the evaluation and sampling designs. The bias in the selection of communities in Senegal for the focus groups is one of the consequences as well as the current lack of time for a proper data cleaning and analyses in both Senegal and Ghana before departing. The presentation of full preliminary results addressing the research questions would have allowed the validation of the research findings by the project’s stakeholders and the discussion of any unexpected results. The study could be more documented if conducted on one topic instead of the current three. Work-load has been constant throughout the study resulting in some interviews scheduled not being currently possible to implement. Millers in Senegal needed to be interviewed to gather their perceptions about the local cereals valorisation project. Interviews with Extension Officers working on the demonstrations are also important to assess the effectiveness, the quality and the diversity in the demonstrations and field-day activities conducted in the two countries. Moreover, the diversity of topics from a country to another may not allow cross-country analyses. Future studies should be limited to one central topic across countries to allow indepth analyses and comparisons. As mentioned in the methodology, the technology dissemination and the valorisation of local cereal projects in Senegal were implemented separately without a planned synergy on the field making the retrospective evaluation design complicated. For more impact, the two projects needed to be implemented in the same communities, districts and regions. FNRAA could make sure these projects align before funding and implementation. In Ghana, the strategy adopted in spreading the demonstrations across districts without repetition may not yield enough evidence for farmers to know about and adopt the improved varieties. It might be better focussing efforts in few communities and districts for two or three successive years and establish linkages with marketers and processors before moving to other districts. Furthermore, potentials exist in the field for a strong collaboration with the International Fund for Agricultural Development (IFAD) funded Root and Tubers Improvement and Marketing Program (RTIMP) to avoid duplications and leverage resources for more impacts. Finally, the design and conduct of the 2013 cassava yield survey is very important to avoid measurement unit biases in the current cassava farm households’ survey in Ghana.
Page | 24
APPENDICES Annex 1. Documents produced #
Document title French
1
Double click to download English
Inception report Inception Report.pdf
2
WAAPP-PCU and partners’ interview guides
5.KK_Ethics_Agricult 4_Agricultural ural Research and Development Research agents' andinterview Development guide_Fr_310114.pdf agents' interview
3
Bakers interview guide Guide d'entretien avec FNBS.pdf
4
Checklist of questions for focus groups 6.KK_Ethics_Focus 6.KK_Ethics_Focus Groups discussion guide_Fr_310114.pdf Groups discussion guide_Eng_310114.pdf
5
Terms of reference for the studies in Senegal Termes de Reference etude Senegal_kodjo_230614.pdf
6
Millet farmers’ survey sampling design Plan d'echantillonnage et de collecte des donnees_Kodjo_vs3_080714.pdf
7
Questionnaire for millet farmers in Senegal Questionnaire_Produ cteurs_Mil_KK_160714_final.pdf
8
Questionnaire for the bakers in Senegal Questionnaire Boulangeries_KK_090914.pdf
9
Questionnaire for cassava farmers in Ghana Questionnaire_Cassa va_660_copies.pdf
10
Listing of households form HOUSEHOLDS LISTING FORM_140_copies.pdf
11
Trip Report – Interviews and Focus group discussions in Thies region Mission report_Thies_20-22_05_14_KK.pdf
12
Trip-Report – Follow-up of data collection in Thies Mission report_Thies_17-22_07_14_KK.pdf
13
Trip –Report. Planning and launching of surveys in Ghana
Mission report_Ghana_27 July - 24 Aug_2014_KK.pd
Page | 25
Annex 2. List of Millet Household Survey Communities in Senegal Code
Communauté rurale
Community
Type of community
Type of survey community 1=Replacement community for initial communities selected; 2=Rep for control communities, 3=Rep. for demo communities; 4=Rep for FOB community.
0=Control 1= Demo 2=FBO 3=Demo+FBO
Initially selected survey communities 1
MERINA DAKHAR
BATTAL1 (NDIAYENE)
2
1
2
MERINA DAKHAR
TELLY
1
1
3
MERINA DAKHAR
THILOR
0
1
4
MERINA DAKHAR
DIOKOUL TABY
1
1
5
MERINA DAKHAR
KEUR DIADIE (KEUR MA)
0
1
6
MERINA DAKHAR
KEUR MEDOUNE GNING
3
1
7
MERINA DAKHAR
LOUKOUCK CISS
0
1
8
MERINA DAKHAR
LOYENE MBAR
0
1
9
MERINA DAKHAR
NDAWENE
1
1
10
MERINA DAKHAR
NDIAYE DIAMBANIANE
2
1
11
MERINA DAKHAR
NGANCAL
2
1
12
MERINA DAKHAR
SATTE1
1
1
13
MERINA DAKHAR
TABBY FALL
3
1
14
KOUL
NGUER1
2
1
15
KOUL
KHANDANE
3
1
16
KOUL
KOUL
1
1
17
KOUL
LAMDIANE
0
1
18
KOUL
NDEUKOU LAMANE KEUR
3
1
19
KOUL
NDIA NDEUKOU
1
1
20
KOUL
NGANDECK
0
1
21
KOUL
NGAYE NGAYE NDIABAL
2
1
22
KOUL
RISSO1
3
1
23
KOUL
THIARENE
3
1
24
KOUL
DIAMA THIENDOU
2
1
Page | 26
Code
CommunautĂŠ rurale
Community
Type of community
Type of survey community 1=Replacement community for initial communities selected; 2=Rep for control communities, 3=Rep. for demo communities; 4=Rep for FOB community.
0=Control 1= Demo 2=FBO 3=Demo+FBO
Replacement communities for control communities 25
MERINA DAKHAR
KELLE
0
2
26
MERINA DAKHAR
MACKA SARR
0
2
27
MERINA DAKHAR
NGUIGUISE MOR
0
2
28
KOUL
NDIAYENE LACK
0
2
Replacement communities for demo communities 29
KOUL
FASS (FASS DIACKSAO)
1
3
30
KOUL
NDOMBOLOKH
1
3
Remplacement communities for FBO communities 31
MERINA DAKHAR
NDIASS
2
4
32
MERINA DAKHAR
NGADIAGA
2
4
33
KOUL
SAKHE
2
4
Page | 27
Annex 3. List of Cassava Household Survey Communities in Ghana Region
District
Communities
Ashanti
Afigya Krabre
Abedease-Kyekyewere
1
Treatment
Amoako
2
Control
Bonkrong
3
Treatment
Gyansa
4
Foase
5
Control Treatment
Afrancho
6
Treatment
Nweneso No. 1
7
Control
Nsoatre
8
Kwatire
9
Treatment Control
Awisa Subinso
10 11
Control Treatment
Nkonsia
12
Treatment
Wuramumuso
13
Mehame
14
Control Treatment
Sekyere Central Atwima Kwanwoma
Brong Ahafo
Sunyani West Wenchi
Asutifi
Community code
Type of community
Page | 28