As suggested in section 3.2, the reciprocity described above is central to feminist
approaches. Indeed, the qualitative interview has been described as "the principal means by which feminists have sought to achieve the active involvement of their respondents in the construction of data about their lives" (Graham, 1984, p. 112).
Contrasting the difference between survey research and more qualitative techniques, Sexton maintains that:
"Unfortunately, the abundance o f statistics and generalisations about1work and its discontents’ gives us little real understanding o f how women lead their daily work lives, experience their jobs, or perceive work-related issues. Personal documents are needed, individual and group portraits o f workers, slices o f real working life, statements by the women themselves - the handwoven fabric o f their daily work lives" (Sexton, 1982, p. 5). Feminists, argues Reinharz (1992) find such techniques valuable in so far as they provide insights into respondents’ perspectives, and give respondents the opportunity to speak in their own words. This is particularly relevant in the case of women entrepreneurs who, as suggested in the literature review, have been largely silenced or ignored, in both theoretical debates as well as more popular fora. Linked with this is the issue of control, described above, and the extent to which the qualitative interview challenges traditional power relationships in the interview process, constructing the respondent as an active- meaning maker, rather than as a subject of study. At the same time feminists (and no doubt other social researchers) appreciate the flexibility afforded by the qualitative
interview. This has been particularly important in this study, where the method has given me the freedom to introduce new questions and to follow up important issues as they emerge. Far from masking diversity, the qualitative interview has provided me with a way of elucidating the richness and complexity of the sample.
3.3.2 Reliability and Validity
Central to debates on quantitative methods are issues concerning reliability and validity; while such issues are equally important in the qualitative context, the precise ways in which reliability and validity are assessed differ from one research context to the other. Reliability
As regards reliability, King (1994) argues that unlike quantitative research which seeks to ensure that the measures used by one researcher will produce similar results when administered by a different researcher, in qualitative research "the interviewer's sensitivity to 'subjective' aspects of his or her relationship with the interviewee is an essential part of the research process" (p. 31).
Reliability, in the qualitative context, concerns the researcher's interpretation of the data, and the extent to which this interpretation reflects what it was that the respondent was trying to say. King suggests that reliability can be assured, first, if researchers explicitly acknowledge their prejudices and assumptions, and "make a conscious effort to set these aside. They should allow themselves to be surprised by the findings" (p. 31).
Furthermore, during the coding stage, inter-rater comparisons can be used to avoid idiosyncratic responses. Likewise Fisher et. al. (1986) extol the virtues of co-research in ensuring reliability in qualitative research. I certainly see the value of collaborative research, but am unsure of its suitability for PhD research, which is by its very nature a solitary activity. As a PhD student I find discussion and consultation with supervisors as well as other colleagues extremely useful, and seek it wherever possible. However, more formalised collaboration with colleagues, in the sense described by King, and others, is more difficult.
On the other hand, what is possible is to check on the accuracy of one's understandings
with the respondents themselves. I would argue that during the interview itself it is important that the researcher explore key concepts and ideas to avoid making incorrect assumptions about the respondent's meanings. For example, as will be explained in subsequent chapters, respondents in this study made frequent reference to the terms
"entrepreneur"; "freedom"; independence" and "support". However, what became obvious after just a few interviews was that they were using these terms in very different ways. Indeed, in the words of one respondent:
"Laura [her business partner, who I also interviewed] and I both talk about freedom and control. But what we mean by freedom and control is completely different"
(Personnel officer to training/development manager o f agency seeking to promote employment o f people with disabilities).
This was an illuminating comment, drawing my attention to the diversity in respondents' understandings, and the extent to which I had assumed a consensus. At the same time I realised that listening to women's accounts, I was interpreting their meanings through my own conceptual (ideological) lens. Fortunately this was recognised quite early on in the process of data collection, and I was able to approach subsequent interviews with greater awareness and insight.
Validity
As regards the issue of validity, King (1994) explains that, just as in quantitative research where "a valid instrument is one which actually measures what it claims to measure", a qualitative study "is valid if it truly examines the topic which it claims to have examined" (p. 31). Thus, validity in qualitative work concerns "the truthfulness of the data" - notoriously difficult to establish when one is working in the realm of perceptions and meanings. Here again King, and other social researchers make a case for collaborative research (Reason and Rowan, 1981; Fisher et. al., 1986).
In addition, Reason and Rowan (1981) emphasise the value of "respondent validation", suggesting that researchers take their tentative findings back to respondents for further clarification and elaboration. This approach, however, has been criticised by a number of writers (Bloor, 1978; Bryman, 1988; Silverman, 1993). Indeed, Bryman maintains that
"such feedback cannot be taken as direct validation or refutation... [but] should be treated as another source of data" (1988, p. 78-79). In the case of this study, respondent validation was not deemed appropriate. Whilst respondents were willing to be
interviewed and welcomed a report on the findings, none was interested in further participation in the study.
Reason and Rowan (1981) identify a number of further guidelines for dealing with the issue of validity in qualitative research. Most relevant to this particular context are their emphasis on "high quality awareness", and the importance of "actively seeking
contradictions in the data", a point reiterated by Silverman (1993). The first of these issues was raised in the discussion of reliability above, while the second two will be considered in greater detail in the sections on data collection and analysis below.
One further point which I feel is worthy of note. As suggested above, this research, far
from aiming to uncover some objective "reality" about "what really happened”, focuses
on women's perceptions of their past experiences. The world of the study is that of interpretation; its aim is to explore the ways in which respondents made sense of that
experience at the time o f interview. The question of validity, therefore, must consider
the data, and the conclusions drawn from them, in these terms.