Ritchie and Spencer (1994) suggest interpretation has six distinct phases: defining concepts, mapping the nature of phenomena, creating typologies of behaviour in the study group, finding associations between characteristics and behaviour, providing explanations and developing strategies/recommendations. These six stages were followed fairly closely for each of the eleven categories used in the coding, and the results should be transparent in the findings chapters. In the findings chapters, the eleven coding categories appear as chapter or section headings, along with several of the sub-categories used.
A key part of producing research work is the conversion of a raw mass of data into ordered, comprehensible findings; the work of interpretation. As
transcribing, coding, analysing and writing-up went forward, ideas evolved and developed which confirmed some of the findings of previous studies but also often went far beyond them. It became clear that the women in the study were highly skilled financial managers, that access to personal leisure spending was closely linked to women’s material deprivation levels, that emotions about money were almost as important as its day-to-day management, that the
household incomes of many of the women in the sample were simply inadequate to meet their material needs, that male financial irresponsibility, real or
perceived, shaped the lives of many of the interviewees, that even when women had complete control of their family’s finances they were commonly at the bottom of the hierarchy of household resource allocation, that the interviewees consistently valued paid work highly and that gender beliefs did underpin how family finances were organised. All of these significant findings are explored in
the subsequent chapters. Without the slow and careful process of analysing the interviews, which took over six months, such findings would not have emerged so clearly. Nvivo8 proved to be a crucial tool in this process but, as Ritchie and Lewis suggested (2003), it could not substitute for fully immersing myself in the data and reflecting on it carefully over a long period. Although time-consuming, the work of transcribing the interviews myself and devising my own coding for them proved an invaluable part of the interpretation process.
An essential part of this process was ensuring that the research findings presented were an accurate interpretation of what had been said. The right balance had to be struck between the subjective and the objective; interviewee accounts which were subjective and personalised still served to cast light on an objective reality (section 5.5.2). But my own account of what I found out through analysing the interviews was also a subjective, personalised one. Clear criteria were therefore needed to help judge whether my interpretations were more objective than subjective. The criteria used were reliability, validity and generalisability (Cresswell 2009).
Reliability asks whether the research findings are reproducible by a different researcher applying the same methods to the same study group. A lot has already been said about how my gender and attitudes and the nature of the sample may have affected the responses given (sections 5.6 and 6.2.2). However, I am confident that these results are broadly reproducible, even if conducted on a like-researching-like basis by a mother living on a lower income. Overall, the quality of the interviews was high, beyond initial expectations. It has already been stated that rapport with the interviewees was good, answers flowed freely, and I have no reason to suppose they are not a genuine portrayal of the lives of the women interviewed. Because of the structure of the
interview, there was a high degree of triangulation on material deprivation between quantitative and qualitative questions, and answers proved fairly consistent across the interview. In some cases, interviewees became more honest as the interview went on, moving away from a desire to portray
themselves in a positive light as the rapport grew, and where this happened it has been noted. Although making no claims to scientific accuracy and not based on a wholly representative sample of the area’s population, I believe that the
interviews give an accurate portrayal of life for mothers in low income working households in the east end of Newcastle.
Validity asks whether the findings are an accurate and true representation of what was said in each interview. Listening to the tapes of the interviews, transcribing them myself, and then re-reading them several times gave me a strong familiarity with each interview and a powerful sense of the life of each individual interviewee. When undertaking the analysis I became aware that there was a danger of seeing the interview responses only thematically; a set of thematically ordered quotes divorced from the context in which they were made. I therefore consciously attempted to analyse the interview data on a case-by-case basis as well as thematically, and to place each coded quote into the broader situation of the woman who made it. In order to do this, I drew up a table which contained a summary of all the data from the interviews, ordered by theme on one axis and by individual case on the other. This allowed separate pieces of data to be placed into the holistic context of a woman’s life. In the findings chapter, this attempt to keep each case distinct has been maintained, and I hope this gives a flavour of the individual lives of those interviewed as well as allowing thematic analysis. Throughout the process of analysis I returned to the full transcripts of each interview regularly, and in the findings chapters a considerable amount of data is quoted verbatim from the transcripts. I am confident this gives a true picture of what was actually said by each
interviewee.
Generalisability asks whether the results can be applied to new settings, peoples or samples. Drawing on established feminist practice, positivistic concepts of ‘scientific rigour’ were rejected in this PhD (section 5.5.2). It is not a scientific document, but rather one which presents the experiences and beliefs of the mothers who were interviewed. This does not mean that these experiences and beliefs are not typical of women of this group; I believe that they are. Many of my findings confirm those of earlier studies. Where they do not, changing
attitudes and technology over time often provide a clear explanation. They also fit well with many conversations I have had over the years with lower income friends and neighbours. Presenting my work as part of Oxfam’s ReGender
front-line practitioners from across the country. However, it is debatable how generalisable the findings are (Arksey and Knight 1999). This is because, firstly, the fact that interviewees focused on certain things while ignoring others means that this is not a comprehensive picture of their views; the answers given are only broadly representative of the views of the interviewee; a ‘taste’ of deeper and more complex understandings and experiences. Secondly, Duncan and Smith have shown that significant regional differences exist in attitudes to paid work amongst women (2002), with communities only a few miles apart having
distinctly different attitudes and behaviours, rooted in different historical and cultural factors. Thirdly, differences in experiences and beliefs will obviously occur between each of the women interviewed. Perhaps these are linked to age, housing tenure, relationship history, family size or educational attainment, or perhaps not; as no attempt was made to control the interview sample on any of these criteria, this study is not able to draw any such conclusions. Suggestive links, where they have emerged, will be described in the concluding chapter. Overall, it does seem likely that many of the findings of this study would prove true for women in other parts of the country and in similar, if not identical, situations. But it is important to bear in mind that there will be differences too. The policy recommendations made are therefore the ones that would best serve the needs of the sample group, and they claim no wider authority, but they would also certainly bring benefits to many other women across the country.
6.4 Conclusion
This chapter has described and evaluated how the research for this project was carried out. In order to answer these research questions set out in the previous chapter (section 5.1), an interview schedule was developed dealing with six specific areas of the lives of mothers living in low income working households; household income; material deprivation; the experience of life on a low income; the reasons for the way money was distributed within the household; the
emotional consequences of that distribution; the belief systems held by the interviewee. Criteria were developed for the study group; women with children under eighteen who were married or living as married, with a full time worker in the household and an income of less than £25,000 a year. Interviewees were recruited in the east end of Newcastle upon Tyne using the author’s personal
connections with Sure Start and the East End Community Development Alliance. The east end of Newcastle has particular characteristics which may have shaped the experience of the women in the study group, particularly a high level of deprivation, low incomes, low employment and a strong sense of local identity and culture. Seventeen women were interviewed in their own homes or in Sure Start settings. It proved difficult to recruit women but the quality of the
interviews was high, enabling the research questions to be answered fully. The sample was not statistically representative of the east end, with an absence of ethnic minorities, too few women who worked full time and too few young mothers. However, the mix of family types, family composition, broad spread of employment types and housing tenures all suggest that this group of women were broadly representative of the mothers in lower income working households in the area. Interviews were transcribed by the author and analysed using
NVivo8. The interview schedule itself was used as the basis for this coding. Findings were checked and re-checked for reliability, validity and
generalisability in an on-going process of reflexivity. Those findings are set out and discussed in the next four chapters.