Data collection

One of the strengths of case study design is that it allows for flexibility in the data collection process and the opportunity to deal with a wide variety of sources and evidence. Thus, through triangulation of data sources, the phenomenon under study can be investigated from multiple perspectives (Yin 2014).

This study collects evidence from multiple sources in order to develop a holistic description of the case partnership and its influence on the Ghanaian smallholder cocoa farmers (Easton 2010). To seek triangulation and alternative explanations, primary and secondary data was collected from a variety of independent sources ranging from peer reviewed journal articles, published books, organisational reports and websites, semi-structured interviews and field observations. Primary data was collected during the field study, which took place in Ghana from 28 September until 27 October 2015. The field study in the cocoa sector in Ghana was conducted in order to gain first hand understanding of the case partnership, the local contextual factors and collect data through interactive interviews with key informants. In addition to the information collected from organisational reports and the semi-structured interviews,

Page 30 of 108

observations from field visits to the cocoa communities in the Fanteakwa and Suhum/Kraboa districts are also part of the data collection methods. Especially, observations on the living conditions, infrastructure and farming equipment in the cocoa communities are useful information collected during the field visits. The multi-method design was chosen in order to create a more complete, holistic and contextual portrait of the case partnership and its influence on the Ghanaian smallholder cocoa farmers. According to Sayer (1992, p. 246), testing in intensive research designs is done through corroboration to assure that the results apply to the different individuals studied as part of the case. By investigating the single case using different sources of evidence, the study can corroborate results using information gathered from all parties relevant to the partnership and contextual observations made in the field.

4.3.1 INTERACTIVE INTERVIEWS

Interactive interviews are a typical intensive research method as they allow the investigator to gather information that is relevant to the different entities. While the questions are informed by theory and observations are theory-laden, such interviews allow the researcher to adapt to different respondents and learn from their interpretations and opinions (Sayer 1992). In preparation for the field study in Ghana, organisational project documents, websites and progress reports were surveyed in order to identify relevant interviewees. Through communication with an employee at UNDP Ghana further insight into the partnership was gained and contact with the project coordinator and the other partners was made. Prior to the arrival in Ghana, initial interviews with UNDP and the project coordinator were organised while the remaining of the 25 semi-structured interviews conducted during the field study were organised as the research progressed and new information was gathered. The two initial interviews were used to acquire knowledge of the structure of the partnership and discover relations to other relevant entities. Additional secondary data sources such as baseline studies and yearly work plans were also received from these interviewees. Based on this initial information, interviews with representatives from Cocobod, Mondelẽz, cocoa farmer co-operatives, and other local stakeholders were organised and carried out.

The interviews can be divided into two main groups. The first group of interviewees represents the case partnership and consist of interviews with UNDP (2), the ESP project coordinator (2), Cocobod (2), and Mondelẽz (1). The other group of interviewees represents local stakeholders and consists of interviews with the Department of Co-operatives (1), Fairtrade Africa (1),

Page 31 of 108

Cocoa Farmer Co-operative leaders (8), and individual cocoa farmers (7). Finally, a semi-structured interview with a local expert in the cocoa sector at the University of Ghana was also conducted¹.

The interviewees were selected based on their connection to the CL program. The first group of interviewees all play key decision-making roles in the CL projects and were therefore relevant to interview in order to collect information on the partnership and the different CL interventions. The farmer co-operative leaders were selected as interviewees because they represent the individual cocoa farmers as beneficiaries of the CL program and typically have been involved in the CL program for a long time. In addition to the co-operative leaders, interviews with individual farmers were made in order to get information from the grassroots on the constraints to upgrading in the GVC. Interviews with cocoa farmers that are not engaged in the CL program were also conducted in order to compare their responses with those of the CL farmers. Finally, the interview with the professor at the University of Ghana serves as an additional and independent source of reference, which also provides historical evidence.

Semi-structured interviews were chosen as the main channel of data collection, because they give the researcher the opportunity to cover the main themes discovered in the relevant literature while allowing the interviewees to share their own ideas and thoughts (Willis 2014, in Desai & Potter 2014). Thirteen individual interview guides were formulated based on the research question and the analytical framework. Even though, most of the guides are similarly constructed individual adaptation was necessary as the interviewees have different relations to the case partnership and are embedded in different contexts². The interview guides were useful tools to structure the interviews and explore key themes with respondents. They also made it easier for the interpreter to get to know the basic pattern of the interviews and formulate his questions to the cocoa farmers who did not speak English. For the interviews with government officials, the company manager and Union leaders the guides also gave a certain degree of formality, which is recommended in order to encourage involvement of the interviewees (Willis 20014, in Desai & Potter 2014).

1 See Appendix 1 for table of interviewees.

2 See Appendix 2 for examples of interview guides.

Page 32 of 108

4.3.2 RESEARCH CONTEXT AND DATA RECORDING

Due to the presence of double hermeneutics in any social science the researcher must continuously be aware of and examine the different meanings and interpretations that arise from context dependency (Sayer 1992). Because the data collection took place in a DC context, which in many aspects differs from the researcher’s own context, the researcher has sought to pay special attention to the interpretations of concepts and the context dependency of meaning. For this purpose, clarifying questions were asked during the interactive interviews if statements or interpretations seemed unclear. Furthermore, the semi-structured nature of the interviews made it possible to examine the interviewees own meanings and understandings of different concepts during the interviews. This was important as not all respondents were fully proficient in English and often used different concepts when referring to stakeholders and practices.

The majority of the interviews have been directly recorded and transcribed while only two interviews with representatives from governmental institutions were not directly recorded³. Since the interviews were often conducted in very noisy environments, notes taken during and after the interviews supplement the direct recordings. Together, these two recordings should ensure the most accurate interview material and allow for examinations of different meanings to limit the risk of making wrong interpretations. All interviews were conducted at the interviewees’ office or in case of the cocoa farmers at their homes. This provided an insight into the environment of the interviewees, but also meant that most interviews were subject to interruptions from colleagues, family members or neighbours.

4.3.3 TRANSLATION

While the majority of the interviews were conducted in English and most of the interviewees were very proficient in this language a note on the use of interpretation during the field visits to the cocoa communities is due. As the field visits were organised very last moment, is was not possible to find a research assistant to help with the translation. Hence, I asked the union president in the Fanteakwa District if he could help me find a local person, who was not involved in the CL program to help with the translation. As a result, a local English-speaking farmer with first-hand knowledge of the local languages and the district became my translator

3 See enclosed CD-ROM for sound recordings and transcriptions of interviews.

Page 33 of 108

for two days. He also showed me the cocoa farms in the forest and his local expertise made up for the fact that his English skills were not perfect (Bujra 2014, in Desai & Potter 2014). While we only had very short time to establish a good personal relation and mutual trust, I emphasised the importance of direct translations and asked him not to filter out details even if he did not find them important (Bujra 2014, in Desai & Potter 2014). However, since I did not have any other sources of translation to use in the transcription of these interviews, the full accounts of the interviewees might not be given in all the translated cases. This can conflict with the aim of examining interpretations and different meanings of concepts used by respondents.