RESEARCH DESIGN AND METHODS
4.2 Research design
I used an additive mixed methods design. A mixed methods design has an advantage of using qualitative and quantitative procedures in data collection and analysis (King, Keohane et al. 1994; Creswell 2002). I carried out phenomenological interviews and a questionnaire survey in sequence. The mixed-method design is suitable because each method reveals different aspects of empirical reality (Patton 2002). Additive multiple methods aim for comprehensive coverage (Bloor and Wood 2006) of the subject under study. This triangulation enabled me to enhance methodological rigour and provided cross-data validity checks (Patton 2002; Bloor and Wood 2006).
There is a view that a researcher‘s epistemology is the theory of knowledge which serves to decide how the social phenomena are studied (Mason 1996 ; Creswell
2002; Groenewald 2004). My epistemological position in this study was that data on the structures and strategies used by adults to cope with DL were contained in their perspectives as participants. Bracketing (epoche) in phenomenological research is setting aside pre-judgements and opening the research interview with an unbiased, receptive presence (Moustakas 1994). I engaged past and current students in collecting the data. Phenomenological approach strove to understand the subjective world of the participants while the questionnaire allowed me to determine relationships among themes and cross-check the truth of the findings (Scott and Usher 1999; Sofaer 1999; Snyder 2006).
Every method has its limitations and no single method can be trusted to adequately solve the problem of rival factors. Hence I used a mixed methods design because of its ability to compensate the weaknesses of one research approach by the strengths of another (Patton 2002).
I conducted a questionnaire survey to complement the interviews. I preferred the survey because surveys collect data about larger populations than other methods (Scott and Usher 1999). I was able to collect data from ZOU‘s ten regional centres in Zimbabwe. Besides, surveys maximize the flow of valid and reliable information while minimizing distortions of what the respondents know (Verma and Mallick 1999). Data from the use of the questionnaire were used to augment conclusions from interview data. A qualitative research conducted in isolation from a numeric approach could be used as a rhetorical basis for retaining existing prejudice. Hence I used a combination of approaches to facilitate a clear way of deciding between competing conclusions (Gorard 2001).
4.2.1 The Sample
My population comprised distance education learners and staff because they are the
key stakeholders in revealing essentials of successful DL. My sampling frame included B.Ed. students and staff at ZOU.
In phenomenological research the phenomenon dictates the method and the type of participants (Groenewald 2004). Hence, I used a purposive sampling procedure to identify participants who had the characteristics that were important and would maximize the possibility of examining a wide spectrum of experiences in my study (Locke, Spirduso et al. 1989). They had pertinent experiences relating to support structures and strategies used by adult learners in coping with DL. I chose purposive sampling because it is considered as the most important non-probability sampling to identify the primary participants. I also based my selection of participants on those who had experiences relating to the phenomenon under study and purpose of my study (Groenewald 2004;Leedy 2010). I did this because additional participants would be selected on the basis of analysed data from the earlier cases (Bloor and Wood 2006).
Long interviews with up to 10 people are sufficient to reach saturation for a phenomenological study (Creswell 2002; Groenewald 2004). I had 12 interviewees. I chose participants because they were knowledgeable and informative (Leedy 1997) about the subject of my study. Purposive sampling is flexible and it helps the researcher to explore the dimensional range along which properties of the emerging concepts vary. That maximizes similarities and differences among information obtained ( Lincoln and Guba 1985 ; Dooley 1995; Ertner 1996; Patton 2002 ).
With the help of the regional programme coordinators, I selected 12 B.Ed. former and current student interviewees from Harare, Mashonaland Central, East and West Regional Centres. The 12 included seven female and five male participants. Five of these were from Harare, 4 from Mashonaland Central, 2 from Mashonaland East and 1 from Mashonaland West regional centres respectively. Among the participants
included 8 who had completed the B.Ed. programme but 2 of these had delayed completion. Another 2 interviewees were in fourth year due to delayed completion as they had previously temporarily withdrawn from the programme. One interviewee was in third year and the last one was doing second year. The second year student was a current member of the ZOU student representative council (SRC). I deliberately chose this interviewee to feel the ―pulse‖ of the SRC at ZOU.
To complement my interview findings, I selected a random sample (Scott and Usher 1999) of 100 B. Ed. students, from the 10 regional centres of ZOU to complete a questionnaire. The distribution of 97 returns out of 100 participants in the questionnaire survey comprised: 34 from Harare, 14 from Mashonaland Central, 13 from Masvingo, 9 each from Bulawayo and Midlands, 7 from Mashonaland East, 3 each from Mashonaland West, Manicaland and Matabeleland North and 2 from Matabeleland South Regional Centres. Random sampling was used because it is free of the systematic bias that may stem from choices made by the researcher. It enables the analyst to estimate the probability of any finding occurring by chance. In addition, the sampling error can only be used to estimate confidence intervals where random sampling is used in the sample selection(Gorard 2001).