Data Collection

According to Yin (2009), the data for a case study should come from six sources: documents, archival records, interviews, direct observation, participant observation and physical artefacts. The logic behind using multiple sources of evidence is that it enhances data triangulation. Moreover, the benefit of having multiple sources of evidence in the case study approach can be maximised by maintaining a case study database, on which all the evidence (data) is stored in a systematic way (Yin, 2009). Thus the database gives the critical reader an opportunity to go back to the raw data as and when required, resulting in higher reliability in the case study approach (Yin, 2009).

108

Data collection in the study took place in two stages. In the first stage, data from the chosen MFIs was collected by conducting in-depth interviews with key officers. The data collected from the first stage was used to refine the conceptual framework and research questions. In the second stage, data was collected from both the MFI officers and owner managers of the microenterprises.

Since the study is exploratory in nature and aims to yield insights into the co- production concept in the microfinance context in Sri Lanka, in-depth interviews using a semi-structured questionnaire was the appropriate method for data collection. Thus, in-depth interviews were used as the main data collection tool. According to Martin et al. (2009), the goal of the in-depth interview method is to yield insights into less researched concepts that can guide theory development and/or future research and hence can be empirically verified in subsequent research. Moreover, Patton (2001) states that the semi-structured questionnaire (interview) allows researcher to gather data in more detail and enable a conversational situational interview. Semi-structured interviews also allow the researcher to collect a wide variety of information while covering the topic of interest (Valk, 2007).

For the study, primary and secondary data were collected. According to Patton (2001), the collection of primary data is often costly and time consuming. This data is tailored to a specific interest purpose and is originated by a researcher, whereas secondary data is often quick and less expensive and is already collected for purposes other than the researcher’s. The primary data was collected through in-depth interviews and observations. The secondary data for this study was gathered through company records, websites (i.e. archival records), product brochures, and government institutions/reports. For example, the websites of six cases were visited in order to collect information pertaining to the organisational structures (i.e. linkages) of cases (MFIs) which have a bearing on co-production. Moreover, some MFIs provided their annual reports, from which BDS performance figures (e.g. Portfolio at Risk (PAR) figures, and number of employments generated) were obtained. These performance figures are useful to examine BDS co-production outcomes. According to Yin (2003, 2009), data from different sources enhances data triangulation.

109

Asking questions in semi-structured interviews

This section explains how a semi-structured questionnaire is designed by using the question order and different types of questions.

Question order

Question order is important in semi-structured interviews as surveys (Weinberg, 1996). The order helps to establish rapport with the respondents. It is always better to move from the non-threatening to the threatening; that is, to ask easy questions first. Asking demographic questions, such as ones about qualifications, at the beginning can sometimes be threatening, so it is better to include these at the end of the questionnaire. However, in some instances (i.e. obtaining one’s philosophical beliefs) asking demographic questions such as qualifications at the beginning would be less threatening (Leech, 2002).

Types of question

There are certain types of question that should not be asked in a semi-structured questionnaire; for example, leading questions, loaded questions and double-barrelled ones (Leech, 2002). One should always try to ask open-ended questions as much as possible. Different types of questions asked in a semi-structured questionnaire are given below.

Grand tour questions

As the name suggests, grand tour questions ask respondents to give a verbal tour of something they know well. The advantage of these questions is that they allow the respondent to talk in a fairly focused way (Spradley, 1979, 1988).

An example of a grand tour question would be “can you describe your business?” (see annex 4; question 1 in the owner manager questionnaire).

Example questions

Example questions are similar to grand tour questions but are more specific. They try to identify a single act or an event experienced by the respondent and ask for an example (Spradley, 1979, 1988).

110

E.g. Can you explain how you have used the knowledge and skills received through training and counselling in your business and household activities giving some examples? (see annex 4; question 24 of the owner manager questionnaire).

Hypothetical interaction questions

These questions allow respondents to put themselves into hypothetical situations and give their opinion (Spradley, 1979, 1988). For example, “if you had to pay for training provided by an MFI, would you pay to attend training?” (see question 18 of the owner manager questionnaire).

Prompts

Prompts are very important in semi-structured interviews and serve two purposes: 1. they keep people talking, and 2. they rescue you when the responses are not clear (Leech, 2002).

There are two types of prompts (McCracken, 1988):

1. Planned prompts: prompts that are formally included in the interview protocol.

E.g. “what factors encourage clients to come for counselling?”

Economic reasons Time, location

(See question 8 of the counsellor questionnaire)

2. Informal prompts: these are unscripted prompts that may be nothing more than the reassuring noises and clarification that interviewer and respondent make during an interview.

Respondent: “Training was quite confusing” Interviewer: “Confusing, why?”

The researcher used a combination of grand tour questions, example questions, hypothetical questions and prompts in the interviews. Five different semi-structured questionnaires were used to collect data from the MFI officials (i.e. microfinance manager, BDS manager, counsellor, trainer) and owner managers. 1-2 hours were spent with each respondent to collect the data (see annex 4). The questions included were based on the research aim, questions and objectives. Before the final face-to-face

111

interviews took place, the researcher visited potential MFIs and interviewed key personnel (initial data collection; see annex 3 for the initial study report), read company reports and visited websites in order to gain insight. A similar process was also followed by other studies (e.g. Johnsen and Ford, 2006).

A brief of the research study was sent to the chosen MFI officers and owner managers, and appointments were made prior to the interview dates. Officers were visited at their respective MFI offices and owner managers at their business premises for the interviews. This method was appropriate, as the researcher was able to make additional observations as well as collect verbal information. It was therefore possible to establish whether there was a discrepancy between what the respondents said and what really happened. The in-depth interviews were audio-taped, transcribed verbatim and then analysed. Annex 5 shows a sample transcript.

112

Table 10: Study fieldwork

MFI-1 MFI-2 MFI-3 MFI-4 MFI-5 MFI-6 External

Trainer

BDS consultant

Dates of data collection (in-depth interviews) 05/09/13 19/09/13 10/09/13 20/09/13 26/09/13 01/10/13 09/10/13 16/10/13 13/10/13 06/09/13 13/11/13 14/11/13 21/11/13 27/11/13 28/11/13 26/09/13 25/10/13 06/11/13 03/12/13 05/12/13 06/12/13 Number of people interviewed 7 (4 MFI officers and 3 OM*) 7 (4 MFI officers and 3 OM) 8 (5 MFI officers and 3 OM) 7 (4 MFI officers and 3 OM) 9 (6 MFI officers and 3 OM) 11 (8 MFI officers and 3 OM) 1 officer 1 officer Duration of interviews (in hours) 7 9 8 7 9 14 1 1

Data and data collection

methods Recorded face-to-face interviews, made observations, took pictures

113

Research ethics

All researchers should consider ethical issues when conducting a study. According to Denscombe (2004), a researcher needs to consider the following three ethical principals when carrying out a study:

1. Participants’ interest should be protected.

2. Researchers should avoid deception and misinterpretation. 3. Participants should provide their informed consent.

Following Principal 1, the researcher sent a letter to all the participants before the data collection, informing them about the objective of the study, their role in the data collection and ensuring how their interests would be protected. When setting up the interviews the researcher made sure they were held at a convenient location and at times convenient to the participants. Thus all the interviews were held at participants’ offices and business premises. The participants were informed that the data collected would only be used for the study and that their identities would not be disclosed.

Principal 2 expects the researcher to be open and honest in the interviewing process and its aftermath. For example, before the data collection began every participant was briefed about the objective of the study and their role in it. Transcriptions were also sent to the participants to verify accuracy. Moreover, the participants were allowed to ask questions and to obtain a copy of the results. The researcher updated the progress of the study to participants so that they felt respected and were willing to provide feedback.

Principal 3 is general agreement from participants that they are participating in the study willingly. The researcher obtained this informed consent from all of them. The participants were assured of their right to anonymity. Thus code numbers were used in the study to represent participants. They were also informed that they were participating in the study voluntarily and could withdraw at any time. They could refuse to answer any uncomfortable questions.

In addition, as a postgraduate research student of Bournemouth University, the researcher had to pass an examination on research ethics and needed to obtain ethical clearance for the study from the university.

114

Transcription and translation

The majority of the interviews were conducted in the Sinhalese language, except for a few which were conducted in English. All the interviews were audio-recorded and the Sinhalese dialogues were translated into English. This was a time consuming exercise, but nevertheless the researcher himself transcribed all the interview data since he wanted to remain close to it. After the data had been transcribed, the researcher consulted with a bilingual colleague to make sure the transcriptions had been translated into English words and perspectives appropriately. Once the data had been transcribed into text the analysis was made. The following sections outline the data analysis procedures.