CHAPTER 4: THEORETICAL AND CONCEPTUAL BACKGROUND: THE LIMITATIONS OF MORAL PHILOSOPHY AND THE NEED FOR A LIMITATIONS OF MORAL PHILOSOPHY AND THE NEED FOR A
5.4.4. Sampling, Approach and Access
The Global Ethics Observatory78 overseen by UNESCO, maintains a database of all experts involved in programmes of ethics including planning for pandemic influenza. I drew upon this database to create my own dataset of prospective interviewees by obtaining their names and addresses. Identifying their role or expertise was rather complicated but was resolved after a few email exchanges. Through this process I managed to identify four experts from Ghana and three from Malawi, all of whom were involved in pandemic influenza planning. In search of more interviewees, I also wrote to the Ministry of Health in Malawi and the Ghana Health Service requesting a draft of their pandemic plans or its contributors. A few people were identified and contacted by email and phone and asked whether they could participate in the study. The number of participants found through this process was inadequate to reach a
78 http://www.unesco.org/shs/ethics/geo/user/ (Accessed: May 22, 2013).
sample size necessary for my research however; a few more were identified at a pandemic preparedness conference in South Africa and this enabled more access. Potential interviewees at the conference were asked to recommend colleagues who I could contact about my research. I received an overwhelming response: sometimes one person would recommend between three and five people. Robson (1994) recommends this kind of snowballing for use when there are potential difficulties in identifying respondents.
All potential interviewees were contacted with an official letter (appendix 4), information sheet (appendix 5), and consent form (appendix 6) either by email or fax. This was followed by a confirmatory telephone call to ensure I had a reply. This proved an effective strategy. In Malawi and Ghana it is acceptable to contact someone provided one has a convincing agenda, even without making a telephone appointment. Telephone calls were made to confirm whether the person was willing to participate in this research and ask if I could include them in my database of interviewees. Similarly, I arranged a specific date, time and location for conducting interviews via telephone.
The pilot for this research activity was planned to be carried out in Zimbabwe, aiming to assess the feasibility of the sampling strategy and whether it was possible to find informants and later implement interviews. However, due to the social and economic instability and possible disturbances arising from political unrest in Zimbabwe at the time of fieldwork, I substituted it with Malawi. The pilot was conducted in Malawi and this country was chosen on the basis that it provided opportunities for clear definition of the focus of the research, allowing me to conduct large-scale research in the same country. Malawi was ideal as a pilot and advantageous as it allowed for more in-depth analysis of the study feasibility. Again, the stability of the country and the availability of the pandemic plan, and related data, made it an interesting country to use as a pilot study.
A purposive sampling strategy was adopted, focusing on individuals, groups and settings that could provide relevant information on the topic (Darlington and Scott, 2002). This strategy allowed decisions about the sampling to be reviewed during the research process and stopped recruiting of more participants when data saturation was reached. A total of 70 participants were identified through the snowballing sampling method and stored in the database of interviewees before I entered the field. I interviewed 46 participants during fieldwork which was sufficient to achieve theoretical data saturation. I made a profile of each interviewed
participant depicting demographic data and details of the interview, such as where and when it was carried out (see appendix 7). A logbook detailing the general feeling of the interview was kept separate and confidential.
5.4.5. Interviewing
This section discusses general issues of interviewing together with the practicalities of my fieldwork, particularly in organizing and carrying out interviews. Prior to interviewing, preparation was done in terms of developing sets of questions and themes to follow. I tried to be as clear as possible in my questioning technique to avoid leading questions and ensure I generated meaningful data. There were general guidelines I established for interviewing that can be found in the information sheet. For example, before the interview I introduced myself, explained the interview purpose to the policymakers and asked whether it was convenient to conduct an interview. I also requested permission to take notes while assuring them that any identifiable personal details would remain anonymous and confidential. In interviews I had two digital recorders, a notepad, a pen and an interview guide. I started the interview with a general open question. For example, I asked policymakers how they had responded to pandemic influenza at the national level.
Considering the complexity and fluidity of the interview topic, I had decided earlier not to adhere rigidly to the list of questions or script; rather I used the guide to ensure the conversation with the respondent covered all the topics of my research enquiry. The interview schedule in appendix 3 is simply a guide and does not suggest that I asked all the questions in a single sitting. Lots of questions in the guide allowed flexibility, choice and guided what to ask. Administering research questions systematically as a routine would negatively impact on the interview process as it would be impersonal and intimidating on the part of the respondent. As Robson (1994) notes, interviews are a kind of conversation that demand a non-threatening strategy to ensure interviewees talk in a free and open manner similar to their ordinary conversational style. Abandoning the structured approach presented in the guide was helpful in shaping the interview focus, as I was able to ask follow-up questions to the answers the respondent provided. This approach allowed respondents to say what was important and express it in their own words.
Before conducting the main interviews, a pilot interview was completed with ten policymakers. Not only were these pilots useful in guiding the best means of engagement
using a focused approach when posing questions, they also helped me avoid repeating questions and issues the respondent had already addressed. Conversations free from repetition are more robust and interesting for the interviewees. The purpose of pilot interviews was to give advance warning about any potential failures in research design and uncover problematic interviewing issues in the large-scale study protocol. For example, as far as the trial run was concerned, the pilot served to work out the “estimated time” needed to complete the interview schedule. Initially I timed the questions as taking an hour and a half to complete but the time schedule almost doubled during the pilot interviews. During the fieldwork I found that sufficient time with the interviewees was essential and that there was, therefore, a need to adjust the schedule.
46 interviews took place in Malawi (22) and Ghana (24) representing over 50 hours of time-recorded interviews. Interview duration varied but the minimum time of “interview proper”
was around 48 minutes and the maximum time around 145 minutes. The interview proper is the time I asked the first opening question and ended with the last question. The mean duration of a recorded interview was 72 minutes; this may contain conversation unrelated to the research. The minimum time in the interview was affected by logistical problems. For example, in many cases, experts were constantly busy due to the nature of their jobs; this ultimately demanded extensive travel and sometimes working from the field. Where possible I had to catch up with respondents in their field of work and this involved me travelling as well. Some busy policymakers only allowed me an hour to talk to them during their lunch break, an opportunity I could not refuse to ask few questions. Though time was of the essence and prolonged interviews enabled me to collect a huge dataset, it was also important to consider in-depth data by asking a few questions rather than breadth data to cover all the themes.
The alternate way of working around this problem was to pose questions tactfully, eliciting valuable data in the shortest time period possible. My concern was to take note of the discourses and forms of power that shaped the words articulated within the responses. I was interested in in-depth data with explanations that gave meaning to the research, as opposed to breadth data that provided frequency. Finding a range of responses is achieved by asking questions, and according to Baker and Edwards (2012), this is important within qualitative research in order to build a convincing analytical narrative based on ‘richness, complexity and detail’ rather than on statistical logic. The time duration of the completed interview may
have implications on the quantification of the quantitative responses, such as how many said this or that. Quantification of interview responses in this study may have little relevance to overall analysis of the findings but remains one area of critique of the study reliability and credibility. Quantifying qualitative data, which involves turning the data from words into numbers or percentages, is an acceptable approach in qualitative research. However, the frequency of the number of responses, although important for organising data, does not say very much about where to pay attention in the analysis. What it does do is reveal data saturation – a counting process that was applied in this thesis to determine when to stop interviewing.
The interview locations varied according to agreed arrangements. These locations were generally safe and quiet so as to avoid disturbance. The interviews were mainly conducted in the interviewee’s office, cafes and hotel lobbies. The majority of the interviews were held at the interviewee’s convenience and preferably during their lunch hour or after work.
Interestingly, two Ghanaians and a Malawian policymaker were interviewed in their own homes. The issue of privacy was considered and it was decided that there was no violation; in fact the interviewee was more comfortable and confident at home, giving rise to substantive synergies within the conversation. The interview locations generally did not appear to affect the quality of interviews.
All interviewees completed the profile form which contained their personal data and the role they played in PRPI. All identifying features in the profile form and interview script associated with the personality of the interviewee had to be removed before archiving the material for the next 7 years in a locked cabinet (personal communication from my supervisors). De-identifying information in interview data that is co-produced between the interviewer and interviewee has raised debate. For example, Babbie (1992) argues that personalities within interview data should not be removed from conversation but rather accepted as a fact that enables future replicability. For this reason, figure 5 and 6 are charts of organizations representing the policymakers I interviewed.
To give a proper description of the interviewing process I must discuss the tool that was used to collect the data and how this relates to the validity of future findings. The robustness of the data collection tool, using qualitative methods such as interviews, remains a vexed issue. My own view is that validity of qualitative research rests in the eye of the readers and, most
importantly, their scrutiny of the design of the methodology, such as the choice of collecting data and the process of analysing the data. Patton (2001) supports the notion of a researcher's involvement and immersion in the research by suggesting that situations are subject to change and a qualitative researcher should, therefore, be present throughout to record an event before and after the change occurs and ensure trustworthy, credible findings. In this understanding, trust between reader and researcher is of paramount importance and must rest not only on honesty but a clear description of the research process. As described above, the researcher (author) in this study was actively involved in the design of the research questions and, to some extent, the review of questions during the data collection process. The questions in the interview were informed by research questions and subsequently the literature pertaining to ethics of planning for, and responding to, pandemic influenza. I used a similar interview guide for both Ghana and Malawi. The purpose was to find variations that existed between these two countries. At the end of the interview, the researcher and the respondent exchanged contacts. I also asked for documentary data. In Ghana, I obtained a newly completed plan for pandemic influenza which was still to appear in the public domain at the time of writing.