In psychosocial studies today the most used methodological approach is to employ either biographical narrative interviews or clinical vignettes and explore single cases in their richness and complexity in order to see how specific societal conditions are processed, adapted to or protested against by the person in question (see, for instance, Roseneil 2006; Layton 2010; Rudberg and Nielsen 2012; Hollway 2015). I have chosen another approach in this book: to look at the whole sample and find the commonalities in patterns of feelings between the singular cases, and then try to see the connection between these patterns and societal conditions that formed them and on which they also may have a transformatory effect.
As argued with Eric Fromm in Chap. 2, analysing the individual case and analysing patterns are not mutually exclusive alternatives, but the choice of focus depends on the purpose of the analysis. Both approaches have their limitations: the single-case approach may lose sight of the general patterns that in their aggregated form may become a social force, while the whole-sample approach does not explain the variation and may lose out on the individual context and specificity. However, since the single cases are the foundation of the patterns of meanings and feelings, they must first be summarised and analysed in their individual specificity in a whole-sample approach. Thus, the patterns that emerge in the analyses in Chaps. 5, 6 and 7 were preceded by thorough work on the level of the singular cases. But in order to make visible the shared patterns, much of the individual context and variation had to be excluded in the subsequent process. Any found pattern can always become more specified by highlighting the differences within; however, categories are analytical constructs and their usefulness will depend on the theoretical potentials of the category in combination with the research question rather than the degree to which they grasp empirical variation in any absolute sense.
So what is a pattern in a whole-sample analysis? A pattern in the context of this study means similarities across individual cases along a certain dimensions of meaning, like feelings about parents, bodies and sexuality, ways of organising work and care, ways to reflect on gender, and the ways in which all these things may interact. A pattern may exist even if it does not cover everything or is contextual-dependent. This is obvious in quantitative studies, but in my view the same is the case for patterns of meaning in qualitative studies (Nielsen 1995). It may be equally important to see the similarities in the seemingly different as seeing the differences in the seemingly alike. As illustrated by chaos theory, levels of order and disorder in complex and non-linear systems are dependent on the scale employed: in a close view, patterns may disappear, but they may re-emerge when seen from a greater distance and vice versa (Kamminga 1990). In this way identifying patterns is both dependent on regularities in the studied object itself and a view from the outside that makes the patterns visible. Here I take a critical-hermeneutic position, trying to avoid both the extreme naturalist and extreme constructivist positions. The object of study has its own ontology, but the scale or the approach of the researcher also conditions the result of the analysis (Nielsen 1995, 1999).
In our analyses of the narratives of the three generations, much of the work has been about finding the distances from which significant and meaningful generational patterns emerge. Sometimes a pattern becomes visible because there are marked quantitative differences between the generations. An example of this is the mother-daughter relationship, where nobody in the oldest generation and very few in the youngest generation have negative descriptions of their mothers, compared to the middle generation, where this is the case with more than half of the informants. However, since we are working with a qualitative sample, significant quantitative differences do not in themselves give evidence of general patterns, as they may be purely accidental. Thus, also a numerically based pattern must be argued to make sense in connection with other dimensions of the data or find support in relation to what has been found in comparable studies. In the case of the shifting mother-daughter relationships, the found pattern makes sense with regard to both the structural changes in women’s lives in these three generations and the ways in which they engage in heterosexual relationships. These things also fit with bits and pieces of other studies, even if these other studies do not describe all three generations. Since numbers in themselves are not the main argument in qualitative studies, it is consequently possible to argue conversely that a pattern can also be significant in cases where it is only found in a limited number of informants. An example of this is the way some of the men in the oldest generation talk with compassion about their too-hardworking mothers. Less than half of them mention this. However, seen in relation to the invisibility of the mothers in the narratives of the other men and compared to the generally elaborated description of fathers, the connection between femininity and weakness emerges as a generational pattern among the oldest men. The arguments almost all the men have about not wanting their wives to work makes this connection even clearer, and this pattern is not seen in the younger generations. Since frequencies are only—and not even necessarily—part of identifying patterns of meaning, I have abstained, also for the sake of readability, from operating with exact numbers in my analyses, but instead have used rough amounts like ‘few’, ‘many’, ‘a majority’ and so on. When it comes to comparing details in the empirical Chaps. 5, 6 and 7 with other studies, I have mainly used other Nordic studies, whereas the whole ‘gestalt’ in Chap. 8 is compared with selected generational studies from Britain. The many ‘see also...’ references are used to indicate when a found pattern of our analyses matches findings in comparable studies.