Qualitative Secondary Analysis and Social Explanation
by Sarah Irwin and Mandy Winterton
University of Leeds; Edinburgh Napier University
Sociological Research Online, 17 (2) 4
Received: 6 Sep 2011 Accepted: 5 Mar 2012 Published: 31 May 2012
The current paper takes as a focus some issues relating to the possibility for, and effective conduct of, qualitative secondary data analysis. We consider some challenges for the re-use of qualitative research data, relating to researcher distance from the production of primary data, and related constraints on knowledge of the proximate contexts of data production. With others we argue that distance and partial knowledge of proximate contexts may constrain secondary analysis but that its success is contingent on its objectives. So long as data analysis is fit for purpose then secondary analysis is no poor relation to primary analysis. We argue that a set of middle range issues has been relatively neglected in debates about secondary analysis, and that there is much that can be gained from more critical reflection on how salient contexts are conceptualised, and how they are accessed, and assumed, within methodologies and extant data sets. We also argue for more critical reflection on how effective knowledge claims are built. We develop these arguments through a consideration of ESRC Timescapes qualitative data sets with reference to an illustrative analysis of gender, time pressure and work/family commitments. We work across disparate data sets and consider strategies for translating evidence, and engendering meaningful analytic conversation, between them.
Keywords: Qualitative Research Methods, Secondary Analysis, Re-Use, Gender, Time Pressure
Introduction1.1 Light as well as heat has been generated in debates about the value, and practice, of qualitative secondary analysis. In this paper we briefly review these debates and document an argument for a critical secondary analysis, specifically one which engages with a set of under-explored middle range issues relating to context, method and conceptual development and oriented to explanation in qualitative research. We have developed this position in the practice of undertaking secondary analysis of data from ESRC Timescapes, a qualitative longitudinal programme of research comprising seven independent, but linked, primary research projects. We develop our argument with reference to some substantive research questions relating to gender, time pressure and experiences of work life balance, building on evidence across different qualitative data sets.
1.2 Recently there has been some growth in resources geared to facilitating re-use for research and for teaching purposes (e.g. through the UK Data Archive and the ESRC Timescapes Archive). Researchers who undertake re-use and secondary analysis are building on, and extending, an established body of work (e.g. Bishop 2007, Bornat 2005, 2010; Gladstone et al 2007; Savage 2005a, b; Fielding and Fielding 2000; Walkerdine and Lucey 1989). Some have suggested that the practice of secondary analysis could contribute to understanding the generalisability of findings from qualitative studies, and potentially enhance the scope for findings from qualitative research to be cumulative (Fielding 2004, Hammersley 1997). However, the practice of qualitative secondary analysis has been accompanied by extensive discussions of methodological challenges, including within contributions to this journal (e.g. Moore 2007, Hammersley 2010).
1.3 Some concerns are founded on the particularity of the relationship between the primary researcher(s), their research participants and the data collected. The centrality of these relationships to qualitative research has led some to see scope for effective re-use of qualitative data as severely circumscribed due to the non-involvement of the secondary researcher in the process of data generation. Another set of concerns relates to how contextual knowledge influences the interpretation and analysis of data. This is not just about researcher presence but about the accessibility of relevant contextual information to researchers, an issue which is likely to be more problematic for secondary than primary researchers. Context has been construed in quite particular ways in debates about secondary analysis so we highlight the significance of a set of middle range issues pertaining to how research problems are conceptualised and framed. Additionally, there is limited discussion within the literature of the challenges of working across qualitative data sets yet an important question can be asked about whether this practice can enhance the explanatory capacity or reach of qualitative data sets. We discuss these issues in turn. In developing our arguments we will draw on examples of our own secondary analysis of ESRC Timescapes data. We consider the value of a comparative strategy for developing, testing and, potentially, building knowledge claims.
Researcher presence and the scope for effective secondary analysis2.1 Data are inextricable from the contexts in which they are produced. Within qualitative research the contextual and relational conditions of data production are usually very central to conceptualising data, and to its theorisation. This presents challenges to secondary analysts. Qualitative researchers typically gain insights into meanings, experiences and subjective orientations through achieving ‘up close’ relationships with participants. They seek in-depth knowledge of participants, and of the settings and contexts in which data are generated. In this way qualitative researchers typically generate data in which they have a key role and a unique situatedness. The immersion of the researcher in research settings, as with ethnography or in evolving lines of questioning in interview, means that they shape how data are generated in an embodied way. They are implicated throughout the data generating process in ways which often evolve and change. Therefore ‘being there’ (inside the researcher’s head) may be important to a full understanding of the process of data generation in qualitative research and, it is argued by some, to data analysis.
2.2 The question then arises: does direct involvement in the original data production offer unique insights that can never be recovered? Hammersley (1997) proposes that even the same researchers returning to their own previously produced data will experience some sense of dislocation from that original context. It is precisely this kind of experience that led Mauthner, Parry and Backett-Milburn (1998) to conclude that qualitative secondary analysis is subject to limitations primarily because of the subsequent researcher’s distance from the original context in which data were produced. For Mauthner and her colleagues (1998) primary researchers have a specific and privileged relationship to the data they generate, through their relationships with research participants and the immediate context of the research. The implication is that this generates forms of reflexivity, and knowledge and insights which are not available to secondary analysts. In their argument such knowledge is key to the adequacy of subsequent analyses and claims being mounted on the data.
2.3 There are some echoes of the concerns of Mauthner and her colleagues in the arguments developed by Hammersley (1997, 2010). He too highlights the intuitive element that resides in much qualitative research work due to the unique engagement that the original researcher has with the populations they research. Whilst he references ethnographic methods to illustrate the implications of this ‘unique’ relationship he suggests that ‘much the same problem’ arises with other sorts of data (1997:139). The unique engagement of the primary researchers and their subject is perceived as a problem for qualitative secondary analysts, unable to capture that unique relationship to the data. For Hammersley (2010, para 3.4:)
“For one thing, in the process of data collection [primary] researchers generate not only what are written down as data but also implicit understandings and memories of what have seen, heard, and felt, during the data collection process. For instance, not everything that is experienced in the course of participant observation or in-depth interviewing will be or can be written down. Nevertheless, it will often be drawn on tacitly, and perhaps sometimes consciously, in the course of analysis, perhaps playing an important role in making sense of the data that have been recorded”
2.4 However, as we will see, for Hammersley the absence of such ‘head notes’ (Ottenberg 1990, cited by Hammersley 2010) is not necessarily damning for re-use projects. Bishop too (2009) notes the difficulties which arise from primary researchers’ claims to greater veracity in their reading of the data due to their intimate involvement in the data generation process. The research method used in primary data generation is very important to the relevance of, and potential for, secondary analysis. Diverse qualitative research strategies will render the resulting data more or less amenable to secondary analysis. Ethnographic research where analyses are built upon forms of cultural immersion, identification and theory building will mean that data, and the inter-personal relationships from which they stem, are more tightly bound up with the researcher and do not stand as a body of evidence which is independent of the primary researcher. In contrast, a series of semi-structured interviews are more separable, as a body of evidence, and therefore potentially much more readily grasped within secondary analysis. In short, the research method has a significant bearing on what kind of secondary analysis can be undertaken effectively. A concern remains that even in relatively structured qualitative enquiry, the original researchers’ presence, and aspects of proximate context, permeate the data in ways that are not always evident for subsequent users of the data (e.g. Bornat 2005; Hammersley 2010). However, this issue also arises for primary researchers, for whom what aspects of context are perceived as salient may alter anyway in the course of the research (Fielding 2004; Bishop 2006). Further, precisely what context has a bearing on research participants’ views and behaviours is a conceptual matter. Indeed its fuller understanding may be an objective of the research (so it will not all be systematically ‘available to view’ within the data from the outset). As Hammersley (2010) notes there is overlap in issues facing primary and secondary researchers here, although difficulties may be sharper for secondary analysts. Whilst distance may be something of a barrier for qualitative secondary analysis to negotiate, it is not a general obstacle. Distance, also, may open up new avenues of enquiry (see also Van den Berg 2005, Mason 2007, Hinds et al 1997).
2.5 Moore (2006) suggests that Mauthner and her colleagues over-privilege the insights available at the time of research compared to those which can be gained through re-use. Data are constructed in both primary and secondary analysis, never ‘given’, and she argues that secondary analysis is more closely aligned to primary analysis than accepted by Mauthner and her colleagues. “…we can come to understand re-using qualitative data not as the re-use of pre-existing data, but as a new process of recontextualising data” (Moore 2006, para 4; also cf. Broom et al 2009). She therefore argues that we should conceive of context as part of the research undertaking, that primary and secondary strategies are more closely aligned than sometimes presented, and that we should see secondary analysis as being not about analysing pre-existing data but about recontextualising data. This puts it on a more equal footing with primary analysis, as well as reminding users of the continual relevance of the socially shaped nature of data and its interpretation. For Hammersley (2010) this critical engagement has value yet he believes that undue emphasis on the socially and contextually made nature of data runs us into trouble. He argues we should retain a concept of data both as given, and as constructed, and that Moore overstates the similarities between primary and secondary analysis. For Hammersley it is more productive to distinguish between data, as that which is collected or generated in the course of research, and evidence, as the analysed data which provides the grounds for inference and for the descriptive and explanatory claims which are built on the data. This allows us to see that both primary and secondary analysts move between data and evidence. Primary and secondary research have a somewhat different positioning: primary analysts have a privileged relationship to the data they have generated, but do not necessarily have a privileged claim on the arguments which can be made from that data. With Hammersley (2010) we argue that sociological data will support different theoretical understandings, and researchers’ presence at data generation is not the final arbiter of the adequacy of such understandings.
2.6 Hammersley (1997) argues that “…what is crucial for the testing of knowledge claims is not how those claims were produced but their cogency in light of the evidence” (p. 135). We will not benefit from circling within questions of exactly how data were produced, the reflexivity of primary researchers, or the contextual information we have available. This would mimic an auditing of research projects, and wrongly imply that veracity comes from an exact reconstruction (Hammersley 1997). Indeed, there may be a risk that undue emphasis on context and ‘presence’ itself encourages a particular mode of engagement with data, a mode which tends towards description, and a sense that the ‘answers’ reside within the specific data set, rather than in theorisation of how the data articulate evidence to address specific research questions. Many of our research questions will be based on, and draw from, wider bodies of evidence in such a way that forms of verification may be sought from outwith, as well as within, the ‘raw’ data sets. Grounding knowledge claims will often entail stepping outside of the specifics of the data set, and relating it to our theories, to other evidence and so forth. It seems to us that a good deal of the discussion about re-use and secondary analysis has proceeded with very broad brushstrokes. There is no single right or wrong here. The central issue is about adequacy of the grounds on which we build, develop and test knowledge claims. With others we argue that the primary and secondary researcher are differently situated with respect to the available data. However, it is how data are used as evidence in the development, testing and building of knowledge claims which is most important. In the next section we reflect on some further conceptual issues relating to context which, we contend, have been under-explored in the literature on qualitative secondary analysis.
Contexts, methods, concepts3.1 Within the literature on secondary analysis and data contexts there has been discussion about the social and historical contexts of knowledge production. Here writers have examined the ways in which theoretical frameworks impact on how research questions are framed, and on what we can ‘see’ through different ontological and epistemological positions (e.g. Moore 2006; Savage 2005; Gillies and Edwards 2005). These are important issues. There has also been commentary on the practice of supplying project specific and often proximate contextual information for re-users (e.g. Bishop 2006). However, there is little discussion of a set of ‘middle range’ issues relating to the ways in which data are shaped by method and by the project contexts in which is it generated. Whilst this is perhaps a very obvious methodological point it warrants some consideration in the process of turning data to evidence (cf. Hammersley 2010). In order to assess data as evidence we need understand how it ‘stands for’ the phenomena under investigation, and which facets of our research questions are being addressed (Irwin 2008; Mason 2002).
3.2 Firstly, we consider the question of how salient context is conceptualised within research designs. To illustrate we can take some simple examples. We might seek extensive detail of context, including proximate, situational, and cultural contexts and suppose that we thereby increase our chances of understanding the circumstances of our research participants and gain insights into why they think and act in particular ways. Hence the importance for analysts (primary and secondary) of having a good grasp of context. However, we may not grasp this opportunity without a well developed conceptual understanding of how specific contexts hold salience for the phenomena under examination. In discussions about effective secondary analysis, such contexts typically appear to be evidenced by whatever data are recorded regarding participants’ immediate circumstances. Yet we can also understand context to be the environment in which participants move, as this holds salience to our research questions. In turn this may mean many things. It may be about their physical environment, their housing, and neighbourhood. It may be about the economic, social and cultural resources available to them. It may be about important social relationships and ties, reference groups and social networks. For example, if research is framed by a theory of ‘rational’, economic behaviours and decision making, then presumptions about salient contextual information would be very different than in research framed by a theory that beliefs and behaviours are culturally embedded. It is important for secondary, as all, analysts to engage with these ‘middle range’ questions of context. They remain relatively under-discussed in the literature on secondary analysis yet they are crucial for situating data as evidence.
3.3 The second, and overlapping question, is how we access relevant contextual information within our research. This is not just a case of the range of our data collection strategies, but also of understanding how assumptions about salient context are embedded within such strategies. For example, interview based research which explores people’s accounts of their practices may engender particular kinds of account. Morally reflexive accounts may dominate, and provide an accurate reflection of people’s thinking, yet this could be a particular picture of partly pragmatic or circumstantially driven motivations. The interview method offers a particular angle on the complex question of motivation, and of values (Mason 2002). What is deemed relevant shapes what is manifested in data, through research designs and through the ways in which questions are asked, and how they are asked. This issue has been more the concern of those undertaking historical comparative research (e.g. Savage 2005a; Gillies and Edwards 2005). Whilst it is acknowledged as an issue for secondary analysis of contemporary data (Moore 2006) we warrant that it merits more detailed attention. The issue relates to how research designs, methodological strategies and data collection tools access multi-faceted problems in potentially quite particular ways, and how researchers’ initial conceptualisations of such problems lie within the data, and their presumed uses.
3.4 A central strength of qualitative research is its capacity to furnish contextual detail and to enhance understanding of the salience of contextual diversity in lived experience. Theorising context lies close to the heart of many qualitative research projects, where context is often an object of analysis. Its perceived salience shapes how projects are conceived and what data are deemed relevant. For example, it may allow us to challenge simplified behavioural models and reveal the cultural and moral embeddedness of social action (e.g. Duncan and Edwards 1999). The concern with contextual relevancies also lies near the heart of the recent expansion of interest in data generation strategies which move away from established interview and participant observation traditions. Many seek more ‘ground up’ strategies, such as visual methods, walking interviews, relational mapping exercises and other techniques which seek to get closer to people’s lived experiences and ‘real lives’ than may be possible in conventional interview formats (e.g. Emmel and Clark 2009; Mason and Davies 2009).
3.5 Within ESRC Timescapes, the separate primary research projects exemplify designs which build on theorisations of how context is salient to the phenomena under investigation. All made use of a range of data collection strategies designed to access key aspects of context. One example might be how, for some projects, participants were involved by virtue of their positioning within family and inter-generational groupings. Rather than simply ask individuals about inter-generational relationships from their perspective, these projects sought to access the associated relationships and dynamics through a range of techniques. Additionally, the Timescapes empirical projects all contained (diverse) theorisations of ways in which longitudinal data enable understanding of evolving lives. Researching continuity and change offers a potentially very revealing lens on context. Longitudinal data have much light to shed on the salience of context in people’s experiences, beliefs and actions. Many of the Timescapes projects were themselves exploring how the longitudinal perspective allows points of comparison through which contextual relevancies are thrown into relief (e.g. Holland and Thomson 2009; Emmel and Hughes 2009). Making context visible is an important objective and outcome, a rationale for diverse forms of comparative analysis. For secondary analysts too, the longitudinal nature of Timescapes data sets may offer a resource for conceptualising contexts and their salience for participants’ social actions, beliefs, decision making and so on.
3.6 How do these observations have a bearing on the conduct of secondary analysis? They highlight context as a domain for conceptualisation and, potentially, for analysis in its own right. Particular conceptualisations of the nature of the social phenomena and of salient contextual information are embedded in methodological decisions and data collection strategies, and will shape the available data, and encourage particular ‘readings’ of the social phenomena being researched. When we bring data from diverse projects into comparison we can see evidence of the impact of diverse project contexts on the data generated. A brief example serves to illustrate. It relates to people’s reflections on belonging (or not) to a generation. Within Timescapes all projects committed to asking a few questions in common, relating to perceptions of biographical change and turning points, to perceptions of historical change, and to perceptions of belonging to a generation. A preliminary analysis suggests that some members of The Oldest Generation sample identified as members of a wartime generation, and referenced the war as a point of commonality. This reference to a common experience could be seen as offering some support for a notion that generation is a meaningful concept for the oldest Timescapes participants, more so than for participants of other projects for whom it typically appeared more nebulous. However, the Oldest Generation study was centred on life history interviews (although other kinds of data were also collected). It seems possible that such interviews, each focusing on a life lived through most of the 20th century and thus in its historical context, might be more likely to render a particular, affirmative, response to a question on generational membership than a similar question asked in the context of a study focusing on current experiences. This is somewhat speculative, and does not undermine the interesting evidence about perceptions of generation amongst participants. However, we need always to recognise that participants’ accounts make sense within the particular narrative and context in which data are generated.
3.7 As secondary analysts of Timescapes data it was incumbent on us to develop an understanding of the contextual embeddedness of the data we examined. This was very much at the forefront of our minds in undertaking a project working across data sets, a topic to which we now turn.
Bringing qualitative data sets into comparison
Introduction4.1 As secondary analysts we needed to get to grips with research project contexts not of our own making, and to evolve hypotheses we could sensibly explore through data from which we were at ‘a remove’. Many of the methodological strategies we advocate would be as applicable to primary as to secondary analysis, for example, attentiveness to sample specificity and structure, and undertaking internal project analysis in order to situate evidence and thereby develop understanding of underlying social processes. In working across data sets we could not simply set data side by side due to their embeddedness in their project specific contexts. We discuss a process of working across disparate data sets and the need to translate between them in order to allow meaningful analytic ‘conversation’.
4.2 The Timescapes data sets were generated by projects which held points of commonality and connection but which had been independently conceived, designed and run, with their own project teams, aims and objectives and research designs. As members of Timescapes, and as secondary analysts of project data, one of our central aims was to explore the scope for undertaking analyses across data sets.  Therefore one central issue was to conceptualise if and how we could establish meaningful bases for comparison of data across diversely conceived projects. An important consideration was how to conceptualise the specificity of samples. We assumed that we were bringing into conversation data sets with participants who were comparable in terms of their social, economic and demographic characteristics, at least in terms of the study populations from which they were drawn. Here we focus on the issue, discussed above, of working across different study contexts. It is not possible, nor desireable, to isolate data, nor evidence, from the ways in which they were created within each study. For example, when we examine fathers’ accounts of aspects of their experience as fathers, as we will see, even responses to similar areas of questioning bear the marks of the differing project contexts. We therefore need to exercise caution when it comes to working comparatively. We cannot assume that seemingly comparable data are commensurate. To work effectively across the data sets requires strategies for allowing data to ‘talk to each other’ meaningfully. 
4.3 In the example we develop below we explored data from two projects where participants were parents of young children, specifically Timescapes projects: ‘Work and Family Lives: the changing experience of “young” families’ (abbreviated to ‘Work and Family Lives’ and ‘WFL’) and ‘Masculinities, Identities and Risk: Transition in the Lives of Men as Fathers’ (abbreviated to ‘Men as Fathers’ and ‘MaF’). These both carry extensive data on identities and orientations to parenting. The Work and Family Lives Project followed a set of sociological questions about work and family life with a particular interest in children’s as well as parents’ perspectives on managing time, and examined children’s, fathers’ and mothers’ experiences separately, running interviews between 2007 and 2010 in and around Edinburgh. Primary research outputs cover methodological as well as substantive areas (e.g. Harden et al 2010; MacLean et al 2010). Men as Fathers was conceived with an interest in social and social-psychological questions, and men were interviewed about their own identities and experiences as these evolved from before the birth of their first child to when this child was eight. The original part of the project was conducted between 2000 and 2008, in East Anglia, and we were provided with access to this data, which has been supplemented by research run from 2008 onwards in Cardiff. Primary research on the MaF data covers a range of areas (e.g. Shirani et al 2012; Shirani and Henwood 2011; Henwood et al 2010; Shirani and Henwood 2010).
Secondary analysis and working across data sets: an example of gendered perceptions and experiences of time pressure in managing commitments to family life and work
4.4 The substantive area we explored was that of how mothers and fathers with dependent children experienced time pressures, how this related to their paid work and family commitments, and if and how gender differences were manifested in their accounts. Due to space constraint and the focus of this paper we only indicate very briefly how the evidence linked with available literature and other research on work life balance and gender. Our main purpose is to outline a ‘working example’ of a substantive secondary analysis, working across data sets and pointing to strategies for testing out emerging conclusions against the available evidence through building meaningful comparisons.
4.5 As secondary analysts with no involvement in the primary empirical projects we came to the Timescapes project data with a range of broad conceptual questions (see Irwin and Winterton 2011a for an account of how we evolved the research questions). We then sought to refine these on reading available data on specific substantive themes. We explored a number of options for working productively across subsets of Timescapes projects. As part of the analysis on which we report here, we reviewed in-depth the content of two data sets to gain a working knowledge of the data held within, reading the metadata, and interview schedules, and most interview transcripts across different waves from one project (Work and Family Lives) and then across the other project (Men as Fathers). From this we generated some ideas and hypotheses about what the data were manifesting, and what we might usefully research in a meaningful way across the data sets. Two hypotheses were (a) that women manage family time and the scheduling of work and care differently to men, and perceive greater tension in managing demands, which they experience as competing, and (b) that men are more likely to maintain greater control and autonomy over their time scheduling, and less likely to experience work and family commitments as competing. We formulated these questions in this gendered way since gendered differences appeared to be very strong on our initial reading of data, both within Work and Family Lives, where we could compare mothers’ and fathers’ accounts, and in Men as Fathers, where a male gender coding was very evident amongst most fathers’ accounts. However, we need to confront the risks of working with a ‘gender model’ of analysis (cf. Feldberg and Glenn 1979, see also Henwood et al 2008). In our subsequent more in-depth analysis of the individual data sets we explored this gendered patterning in more depth, and we also sought to identify exceptions to the general picture which emerged from our early analyses. Rather than assume gender is a fundamental categorising device we wanted to gain a better understanding of the contexts in which particular gendered accounts are manifest and if, and how, they vary.
4.6 We framed our analysis with reference to questions about continuity and change in gender, work and care arrangements. Whilst there have been important changes in women’s position with respect to employment, and the material resourcing of families over recent decades (Irwin 1995, 2005), there have been continuities in the unequal arrangements whereby women carry much more extensive domestic responsibilities than do men. We know from an array of studies (both quantitative and qualitative) that women tend to carry the weight of work life balance pressures (Kan et al 2001, Rose 2011, Bianchi and Milkie 2010, Forsberg 2009, Crompton and Lyonette 2006, Craig 2006; Vincent et al 2004). This is the wider context in which we were interested in gendered experiences of time pressure. Our strategy for developing an analysis was to work in detail with one data set, evolve a line of enquiry and see how far we could test and develop the above hypotheses, and then consider how this might translate into a commensurate analysis of the second data set, in which we could also explore (suitably translated) hypotheses. We developed an inductive approach exploring data from the Work and Family Lives project, reading and re-reading cases and building an understanding of how the phenomenon of interest (experiences of time pressure) were manifest across the sample and how individuals were situated in this respect and relative to one another.
4.7 Starting from readings of the transcripts of all parents with a co-resident partner, and extending to single parents within the study, we read available transcripts of all of the WFL study participants. The analysis was based not on answers to a specific question but on an assessment of experiences of time pressure, and the linked managing of day to day scheduling of work and care commitments, as manifested in participants’ accounts. We undertook readings of Wave 1 and Wave 3 transcript material, where parents were interviewed on a one to one basis. Our analysis suggested that those who were mothers and had extensive paid work (or, in one case, study) commitments were the ones most likely to talk about managing their work and family commitments in terms of compromise, conflict and spreading themselves thinly. Such anxieties were not absent from (some) male accounts but it was a more common motif amongst working women. For example, one mother who worked full time said:
It’s really hard. It’s a lot harder than I thought it was going to be. It’s great fun, but it’s much harder and I think it’s becoming… well, it’s obviously becoming increasingly harder as we’ve had the baby as well, and we don’t have family support round about. … So yes. It’s two extremes. It’s really, really good fun and it’s very kind of noisy, and lots going on. But on the other hand it’s just exhausting
4.8 Another mother, talking about her work identity and how she manages it refers to the juggling, a theme of many women’s accounts in the WFL project:
But the juggling, constant juggling, is always there. It’s just this background noise in your headWomen in part time employment, and men in general were less likely to foreground such challenges within their accounts.
4.9 We sought to work comparatively within the project data, making use of the diversity of household types in the data set. There was some diversity of household types. Of 14 families participating, 9 were heterosexual couple, and 5 were lone mother, families. All parents were in paid work excepting two fathers who had retired early for health related reasons (see Maclean et al 2000 for details of the sample). There is not, therefore, the numerical breadth which would allow a comparative analysis in which we could isolate differing contextual influences on experiences. However, we could see evidence of some interesting commonalities in women’s experiences of work life stress, manifested across different contexts. We next consider cases where we might expect working women would experience less time pressure due to their partners’ circumstances. Two fathers had retired due to ill health and taken on significant care duties, whilst their partners undertook paid work. Whilst not strictly a role reversal situation this offered some opportunity to explore working mothers’ experiences of time pressure in such contexts. In one case the couple had one child, supportive family locally, and the mother described some flexibility in her work. In the other case, the mother, Fiona, manifested an experience of time pressure similar to women in dual worker households. Fiona and her husband John had three children, she worked four days a week as a project manager in the public sector and he, having retired early through ill health, took on significant domestic and childcare responsibilities. We might hypothesise she would be less likely to experience conflicting pressures on her time. John’s intermittent recurrence of illness placed Fiona under more stress than if this were a straightforward role reversal situation. Nevertheless, interestingly when she was reflecting on her role as a worker and mother, her account echoed those of other working women with working partners. So whilst she talked about the fact that she was not in a traditional mum’s role and her husband did a “lot of the houseworky things”, she said being a mum was “..kind of competing demands all the time” and that “..wherever I am , I’m always thinking that I should be doing something else”. She went on: “..I think, for me, its always feeling like you’re not devoting quite what you should to whatever task that it should be. And so …, it is that, just you feel you’re spreading yourself thinly”. Fiona’s account of “plate spinning” appeared to strongly echo accounts of other full time working mothers with working partners. This is in keeping with wider evidence which points to a pattern where working mothers retain extensive responsibility for managing care and the domestic domain, and experience conflict and compromise more extensively than men. In the data to hand there are hints that this obtained even in circumstances where we might hypothesise otherwise: specifically where the male partner did extensive practical caring work. On our readings, the evidence related well to extant evidence on experiences of gender and time pressure. For example, women across diverse circumstances appeared more likely to manage the work of work life balance than did men, and generally appeared more prone to experiencing pressure, particularly when they held extensive paid work commitments.
4.10 We then wanted to bring into conversation evidence from WFL with data from the Men as Fathers project. We earlier emphasised that we need exercise caution in bringing together data from differently conceived projects, as the data reflect their project context, design and method. In ‘Men as Fathers’, men were oriented to issues of their own personal, masculine identities as these evolved through their experiences of being a father. The design of Men as Fathers meant that the way the male participants were oriented: by the study design, by the lines of questioning, and by the fact that within their families it was only they who were interviewed, all press towards accounts in which we might expect men to come over as, if not more self centred at least more self oriented than participants in WFL. As such, work life balance difficulties (perhaps as they saw them being experienced by their partners if not by themselves) might be underplayed by these fathers. For these sorts of reasons we need to conceptualise what kind of evidence enables an effective comparison with data in WFL (recalling also that we have only men to choose within MaF). An important move here was again to initially undertake an internal project analysis across particular cases. We began with a deductive way of entering into the project data set, starting with cases which looked potentially ‘productive’ with respect to our questions, based on project metadata (supplied to us by the team) about household divisions of labour. However, having followed this more deductive strategy we then read ‘outwards’, to give confidence we were interpreting the evidence in a way which was consistent with other cases. In WFL we had explored diversity amongst working women and, as part of this, taken an interest in how the experiences of mothers in less conventional circumstances (where their partners took on extensive practical care commitments) compared with others. As a reverse mirror image in MaF we took particular interest in men in less conventional circumstances (in the data set, and more generally), specifically those who worked extensively and desired extensive practical hands on care of their young children (we might see them therefore as being in circumstances more typical of working mothers). We brought these into comparison with other men within the MaF data set. In effect we sought to translate our questions, and emergent hypotheses, to the second project context, as dissimilarity in project designs, samples and the data meant that we could not simply ask identical questions across them.
4.11 As a reference point, we start with the example of Kenny, a participant within a conventional division of labour. In an early interview when his first child was a baby, and he was unemployed, he had envisaged a more progressive division of labour in the future than in fact came to pass. When the child was 8 (and another aged 6) he and his wife manifested a conventional division of labour in which he was the main breadwinner as a self employed chiropodist building his own practice. Consistent with his primary earner situation he offered a male coded account. To illustrate, in answer to a question about being an involved caring father, looking back over 8 years, he said:
“I mean I don’t know what percentage out of the hundred that I’d hoped for, Michelle’d probably say its less than she’d hoped for. I would probably say it’s about as good as I can manage”
4.12 The quote encapsulates a more general sense conveyed in the interview that ‘as good as I can manage’ refers to a division of labour with which he feels comfortable. He conveyed no evidence of unease with their work family balance situation. We can then compare this quite unproblematic rendering of work/family balance with the perspectives of two working fathers who appeared committed, unusually, to an ongoing and extensive hands on caring commitment to their children. Bruce was a school teacher and head of a department within school, his partner worked full time as director of her own company. He desired extensive practical caring involvement:
Int: do you feel you have a different role to (your partner)
Bruce: I think it is interchangeable as far as the sort of caring is concerned, I mean I think that probably we are not the average couple…… I think we are interchangeable and.. I mean its hilarious but it is that I am more of a mum and she is more of a dad. ….there is almost a role reversal in the traditional roles
4.13 Bruce worked full time and undertook an extensive caring role so we might ask if he was positioned like working mothers and experienced work family conflict. As it turns out his commitments were facilitated by extra, bought in support, since the family had a nanny. Interestingly, also, he talked not of the conflicting pressures which recur in women’s accounts in WFL (and elsewhere), nor with their guilt over spreading themselves thin and the need to compromise, but saw this compromise as his to make, an aspect of his authority:
Actually I have re-organised my working patterns I suppose, and I am also now much less worried about missing minor deadlines. You know in some ways I am rather more robust and about things at work…. … it would be wrong to say that I don’t care about it but I don’t let things get to me ..I used to be very assiduous.. It had to be right and now you know well you know if something works well fine and that will do
4.14 In the example of Bruce then we see evidence of how a potential time pressure ‘tension’ was perceived and experienced very differently than was the case amongst mothers in Work and Family Lives, in part here due to the bought in support, and in part (we suspect) facilitated by his work identity. Commensurate authority at work amongst women does not appear to generate commensurate authority or autonomy in respect of work life balance issues.
4.15 Our third example from ‘Men as Fathers’ is of Malcolm, a man who also desired extensive practical commitment with his children but who, unlike Bruce, had limited control over his own time. He therefore was situated in a circumstance more similar to many women. He was a prison officer and worked shifts. He talked of chores and domestic planning (“tomorrow I’ll start planning the picnic and everything”) suggesting a contribution to household management and a level of involvement which was relatively unusual for men in the MaF data. Additionally he offered an account in which he appeared more conflicted and sounded more like WFL mothers in his sense of compromise and unease about whether he was effective in managing his time across work and family life:
Int: ….when I say time to you what do you think of?Interestingly then, in this case of a man who worked full time and experienced limited control or autonomy over his work time, and desired an extensive hands on care parenting role, he also had a much more marked experience of work / family conflict.
Malcolm: Not enough, not enough,… one thing I’ve not got enough…. Its not so bad at the moment but I never used to have enough time to complete me work and I’d come home and I’d look at it and think there’s not enough time to have something to eat before I go to bed
4.16 Overall the data as we have explored them here suggested that typically women and men with young children experience time very differently. Where women were working extensively they also tended to manage the work of work life balance and provide accounts of their experiences of compromise and juggling. This was the case when partners were in full time work. However, we also saw it evidenced where a male partner did a lot of practical care work. In the example within MaF of a man desiring (and offering) extensive practical domestic involvement, and having limited autonomy over work time, we see an account which reflects his experience of conflict and compromise. His account echoes those much more typical of women (as reported in WFL and also in the qualitative studies cited earlier). The data offer some evidence about the asymmetry and inequality in which women and men are situated, reflected both in gendered experiences of time pressure within conventional divisions of labour, but also within minority circumstances of non-conventional divisions of labour. It evidences the very deep seated nature of structural drivers of gendered inequalities in experiences of time pressure in family life. The evidence here is based on numerically small samples, although it is consistent with what we would expect from extant research and evidence. However, our wider purpose is a methodological one, and here we hope that the approach we have taken illustrates scope for working across diversely constituted project data, and exemplifies a strategy for doing so.
4.17 In sum, the Timescapes primary projects held different designs, disciplinary contexts and samples, and different sets of questions which were being explored, although there were some commonalities of focus in the domain we wished to analyse. We developed research questions informed by extant evidence and research on gender, work and care and experiences of time pressure. The central methodological question we sought to address was if we could allow a meaningful analytic conversation across differently constituted data sets, taking into account the very different contexts in which data were generated. We therefore sought to translate our research questions across the project data, so these questions had integrity within the project contexts but were also commensurate, allowing us to bring project data into comparison.
Conclusion5.1 In this paper we have reviewed debates about secondary analysis, focusing on distinctions made between primary and secondary research, on issues of context, and on strategies for building and testing conceptual understanding of social processes. With Hammersley (2010) we have argued that rather than debate the respective roles of primary and secondary researchers in conducting valid analyses, a more productively drawn distinction is between data and evidence. Primary researchers have more privileged knowledge of primary data but both primary and secondary analysts will construct data as evidence in the service of some empirically grounded set of arguments and knowledge claims. How effectively such arguments are made can be judged against the usual criteria of social scientific explanatory adequacy. Presence at the point of data generation is not a final arbiter. In addition there are risks that overplaying the significance of proximate context relative to other salient factors may privilege description over explanation. In the literature on secondary analysis and context there have been discussions oriented to practical issues in documenting immediate contextual factors, as well as discussions regarding historically and paradigmatically embedded research contexts, through which current researchers can ‘excavate’ or reconstruct data from different theoretical positions. We have suggested that there is a relative dearth of discussion of a set of middle range issues relating to context and method. Theorising and analysing context, then, should be part of a critical secondary analysis.
5.2 We have also argued the value of a comparative strategy in undertaking secondary analyses. This has been part of our own remit, and may be an aim of others working with more than one data set and/or seeking to bring secondary data into conversation with their own primary data. We have developed the argument through an analysis of gender and parents’ experiences of time pressure, drawing on data from two Timescapes projects. We explored the risks of comparison in the absence of a sound understanding of project contexts and methods, and of how these shape the resulting data. Within each data set we sought an understanding of internal diversity as it related to our research questions, and we tested our hypotheses about experiences of time pressure amongst those in majority, ‘conventional’ circumstances as well as those in generally less conventional circumstances within both data sets. We sought to conceptualise the kinds of evidence which would support or falsify our evolving understandings. In our examples we used a strategy of exploring typical and atypical experiences as these were meaningful within data sets, and in relation to wider evidence on extant patterns and processes. We undertook the detailed analyses initially with Work and Family Lives data, and then explored how we might translate our hypotheses in order to generate questions of the Men as Fathers data set. We could not ask the same questions of a wholly different sample and project data set, but rather we asked commensurate questions, enabling meaningful analytic conversation across data sets. We sought to develop and test emerging ideas and concepts within and across data sets. Grounding knowledge claims entails stepping outside of the specifics of the data, and relating them to our theories, and to other evidence. Rather than dwell with the question of the possibility of effective secondary analysis, we have sought to describe our position in more positive terms, and orient towards questions of adequacy in the grounds on which secondary analysts build and test knowledge claims.
AcknowledgementsWe would like to thank Joanna Bornat, Nick Emmel, Ros Edwards, Libby Bishop and Bren Neale for their helpful comments on an early version of this paper. In addition we are grateful to the Timescapes 'Work and Family Lives' and 'Men as Fathers' project teams and, in particular, Jeni Harden and Karen Henwood, for their comments on an earlier draft of our analyses relating to their project data. The analyses here are our own and our interpretations of that data are not necessarily shared by the primary project teams.
Notes1Timescapes was funded by the ESRC: RES 347-25-0003, and ran from 2007 - 2011. Details of the seven primary research projects which were part of Timescapes are available on the Timescapes website. The Timescapes Secondary Analysis Project was funded as part of Timescapes, and ran through 2010 and 2011. The project was run by Sarah Irwin, and Mandy Winterton worked on the project as a Research Fellow, for 18 months. The Timescapes Archive of qualitative longitudinal data is available for secondary users. For further information visit <http://www.timescapes.leeds.ac.uk>.
2For a description of the range of work we undertook please visit: <http://www.timescapes.leeds.ac.uk/secondary-analysis/timescapes-secondary-analysis-project.php>. For an account of our strategies and practices in the doing of secondary analysis across a range of substantive areas see Irwin and Winterton 2011b.
3This project was directed by Professor Joanna Bornat at the Open University. We are grateful to Joanna and her team for providing us with access to The Oldest Generation data, and for having a dedicated project meeting with us in winter 2010-11.
4There is an echo here of the approach of meta-ethnography (Noblit and Hare 1988) where analysts seek to accumulate knowledge by drawing together findings of diverse studies. These cannot simply be pooled, but require a process of translation so that deductions about equivalence can be made, for example where common processes might be identified. Meta-ethnography (like research syntheses more widely) work with other researchers’ findings and not with the fieldwork data itself. (For a review see Dixon-Woods et al 2004).
5Undertaking secondary analysis when the projects were still ‘live’ generated challenges for both primary and secondary analysts. The projects were at different stages in archiving their data, and team members had not completed their own analyses and writing even though the present authors, as Timescapes secondary analysts, were undertaking concurrent analyses of the data teams had supplied to us as part of the collective Timescapes commitment. We are grateful to the teams for supplying us with their primary project data.
6This assumption is reasonable given the project sampling strategies. There is a separate Scottish and English context for the two projects discussed here, but we do not think that the data sets hold a systematic difference with respect to the issues of gender and time pressure we explore. There are other likely dimensions of sample specificity which remain hidden, for example relating to project-specific self selection issues.
7Identifying strategies for theory building and testing was difficult given the relatively small numbers of people interviewed in the data sets available to us. Given their longitudinal design the amount of data there was nevertheless extensive.
8This project was directed by Professor Kathryn Backett-Milburn at the University of Edinburgh. We are grateful to Kathryn and her team for providing us with access to the Work and Family Lives data, and for having a dedicated project meeting with us in winter 2010-11.
9This project was directed by Professor Karen Henwood at the University of Cardiff. We are grateful to Karen and her team for providing us with access to Men as Fathers data from interviews conducted in East Anglia from 2000-2008, contextual information regarding the work arrangements of participants and their partners, and for having a dedicated project meeting with us in winter 2010-11.
ReferencesBIANCHI, S.M. and Milkie, M.A. (2010) ‘Work and family research in the first decade of the 21st century’, Journal of Marriage and Family, 72: 705-25. [doi:://dx.doi.org/10.1111/j.1741-3737.2010.00726.x]
BISHOP, E. (2009) ‘Ethical sharing and reuse of qualitative data’, Australian Journal of Social Issues 44 : 255-72.
BISHOP, L. (2006) ‘A proposal for archiving context for secondary analysis’, Methodological Innovations Online 1 .
BORNAT, J. (2010) ‘Remembering and reworking emotions. The reanalysis of emotion in an interview’, Oral History 38 : 43-52.
BORNAT, J. (2005) ‘Recycling the evidence: Different approaches to the reanalysis of gerontological data’, Forum: Qualitative Social Research, 6 : Art 42.
BROOM, A., Cheshire, L. and Emmison, M. (2009) ‘Qualitative researchers’ understandings of their practice and the implications for data archiving and sharing’, Sociology 43 : 1163-81. [doi:://dx.doi.org/10.1177/0038038509345704]
CRAIG, L. (2006) Does father care mean fathers share? A comparison of how mothers and fathers in intact families spend time with children, Gender and Society 20 : 259-81. [doi:://dx.doi.org/10.1177/0891243205285212]
CROMPTON, R. and Lyonette, C. (2006) Work-life balance in Europe, Acta Sociologica 49 : 379-93
DIXON-WOODS, M., Agarwal, S., Young, B., Jones, . and Sutton, A. (2004) ‘Integrative approaches to qualitative and quantitative evidence’, London: The Health Development Agency.
DUNCAN, S. and Edwards, R. (1999) ‘Lone mothers, paid work and gendered moral rationalities’, Bassingstoke: Macmillan.
EMMEL, N.D. and Clark, A. (2009) ‘The methods used in Connected Lives: investigating networks, neighbourhoods, and communities’, ESRC National Centre for Research Methods. NCRM Working Paper Series 06/09.
EMMEL, N.D. and Hughes, K. (2009) ‘Small –N access cases to refine theories of social exclusion and access to socially excluded individuals and groups’, in Byrne, D. and Ragin, C. (eds.) The Sage handbook of Case-Centred Methods. Sage. London [doi:://dx.doi.org/10.4135/9781446249413.n19]
FIELDING, N. (2004) Getting the most from archived qualitative data: epistemological, practical and professional obstacles, International Journal of Social Research Methodology 7 : 97-104. [doi:://dx.doi.org/10.1080/13645570310001640699]
FIELDING, N.G., and Fielding, J.L. (2000) ‘Resistance and adaptation to criminal identity: Using secondary analysis to evaluate classic studies of crime and deviance’, Sociology 34 : 671-689.
GILLIES, V. and Edwards, R. (2005) ‘Secondary analysis in exploring family and social change: addressing the issue of context’, Forum: Qualitative Social Research 6 : Art. 44.
GLADSTONE, B., Volpe, T., and Boydell, K. (2007) ‘Issues Encountered in a Qualitative Secondary Analysis of Help-Seeking in the Prodrome to Psychosis’, Journal of Behavioral Health Services and Research, 34 : 431 – 442. [doi:://dx.doi.org/10.1007/s11414-007-9079-x]
HARDEN, J., Backett-Milburn, K., Hill, M. and MacLean, A. (2010) ‘Oh, what a tangled web we weave: experiences of doing ‘multiple perspectives’ research in families’, International Journal of Social Research Methodology 13 : 441-452. [doi:://dx.doi.org/10.1080/13645571003650979]
HAMMERSLEY, M (2010) ‘Can we re-use qualitative data via secondary analysis? Notes on some terminological and substantive issues’, Sociological Research Online 15 , at <http://www.socresonline.org.uk/15/1/5.html
HENWOOD, K., Shirani, F. and Coltart, C. (2010) ‘Fathers and financial risk-taking during the economic downturn: insights from a QLL study of men’s identities-in-the-making’, Twenty first Century Society, Journal of the Academy of Social Sciences (Special issue: Lived Experiences through econonomic downturn in Britain) 5 : 137-148
HENWOOD, K., Parkhill, K. and Pidgeon, N. (2008) ‘Science, technology and risk perception: from gender differences to effects made by gender’, Journal of Equal Opportunities International 27 : 662-676. [doi:://dx.doi.org/10.1108/02610150810916730]
HINDS, P., Vogel, R., and Clarke-Steffen, L. (1997) The possibilities and pitfalls of doing secondary analysis of a qualitative data set, Qualitative Health Research 7 : 408-424. [doi:://dx.doi.org/10.1177/104973239700700306]
IRWIN, S. (2008) ‘Data analysis and interpretation. Emergent issues in linking qualitative and quantitative evidence’, in Hesse-Biber, S. and Leavy, P. (eds) Handbook of emergent methods, New York: Guilford Press.
IRWIN, S. (2005) Reshaping social life, London: Routledge
IRWIN, S. (1995) Rights of passage. Social change and the transition from youth to adulthood, London: UCL Press.
IRWIN, S. and Winterton, M. (2011b) Qualitative secondary analysis in practice: an extended guide (with reference to concepts, contexts and knowledge claims) Timescapes Working paper 7, available at <http://www.timescapes.leeds.ac.uk/resources-for-ql-research/publications.php>
IRWIN, S. and Winterton, M. (2011a) ‘Timescapes data and secondary analysis: working across the projects’, Timescapes Working Paper 5, <http://www.timescapes.leeds.ac.uk/resources-for-ql-research/publications.php>
KAN, M.Y., Sullivan, O. and Gershuny, J. (2011) Gender convergence in domestic work: discerning the effects of interactional and institutional barriers from large-scale data, Sociology 45 : 234-51. [doi:://dx.doi.org/10.1177/0038038510394014]
MACLEAN, A., Harden, J. and Backett-Milburn, K. (2010) ‘Financial trajectories: how parents and children discussed the impact of the recession’, Twenty first Century Society, Journal of the Academy of Social Sciences (Special issue: Lived Experiences through econonomic downturn in Britain) 5 : 159-70.
MASON, J. (2007) ‘“Re-using” qualitative data: on the merits of an investigative epistemology’, Sociological Research Online, 12, 3. Available at <http://www.socresonline.org.uk/12/3/3.html>. [doi:://dx.doi.org/10.5153/sro.1507]
MASON, J. (2002) ‘Qualitative interviewing: asking, listening and interpreting’, in May, T. (ed) Qualitative research in Action, London: Sage.
Mauthner, N.S., Parry, O. and Backett-Milburn, K. (1998) ‘The data are out there, or are they? Implications for archiving and revisiting qualitative data’, Sociology 32 : 733 – 745 [doi:://dx.doi.org/10.1177/0038038598032004006]
MOORE, N. (2006) ‘The contexts of data: broadening perspectives in the (re)use of qualitative data’, Methodological Innovations Online 1, 2.
NOBLIT, G.W. and Hare, R. D. (1988) Meta-ethnography: synthesizing qualitative studies, London: Sage.
ROSE, J. (2011) ‘Finding time to balance: perceptions of time pressure and work-life balance’, paper presented to Community, Work and Family IV International Conference, University of Tampere, Finland, 19-21 May 2011.
SAVAGE, M. (2005b) ‘Revisiting classic qualitative studies’, Forum: Qualitative Social Research 6 : art 31.
SHIRANI, F., Henwood, K. and Coltart, C. (2012) ‘Meeting the challenge of intensive parenting culture: gender, risk management and the moral parent’, Sociology 46 : 25-40. [doi:://dx.doi.org/10.1177/0038038511416169]
SHIRANI, F. and Henwood, K. (2011) ‘Taking one day at a time: temporal experiences in the context of unexpected life course transitions’, Time and Society 20 : 49-68. [doi:://dx.doi.org/10.1177/0961463X10374906]
SHIRANI, F. and Henwood, K. (2010) ‘Continuity and change in a qualitative longitudinal study of fatherhood: relevance without responsibility’, International Journal of Social Research Methodology, First published on: 16 March 2010 (iFirst).
VAN DEN BERG, H (2005) ‘Reanalyzing qualitative interviews from different angles: the risk of decontextualizion and other problems of sharing qualitative data’, Forum: Qualitative Social Research, 6, (1), art 30, <http://www.qualitative-research.net/index.php/fqs/article/viewArticle/499/1074>.
VINCENT, C., Ball, S.J. and Pietikainen, S. (2004) Metropolitan mothers: mothers, mothering and paid work, Women’s Studies International Forum 27: 571-87 [doi:://dx.doi.org/10.1016/j.wsif.2004.09.011]
WALKERDINE, V. and Lucey, H. (1989) Democracy in the kitchen. Regulating mothers and socializing daughters, London: Virago Press.