First steps in qualitative secondary analysis: experiences of engaging with the primary research team

ABSTRACT

Engaging with primary researchers during qualitative secondary analysis is a practice much recommended but rarely written about. In this article, I reflect on my experience of crossing an imagined boundary between the discrete textual dataset and its creators, of acknowledging and engaging with those researchers who invested in constructing the data, some of whom are still actively working with it. Focusing on four rationales for engaging with primary researchers - orientation, navigation, interpretation and the ethics of ownership - I describe the opportunities, tensions and dilemmas that presented themselves as I entered into dialogue with members of the primary research team. As the encouragement of rapid archiving practices and open qualitative research present new possibilities in working with contemporary ‘living’ archives, so it also raises new methodological and ethical considerations regarding engagement with and relationality towards primary researchers still active in their fields.

KEYWORDS:

• Qualitative secondary analysis
• reflexivity
• ethics
• qualitative archives
• open qualitative research

Introduction

In this article, I reflect on my experience of engaging with members of a research team whose data I was seeking to use in a Qualitative Secondary Analysis (QSA) project. I encountered a number of ethical and methodological issues that, whilst common to any QSA study, were heightened because the archived data set in question is very recent. Many of the original research team are still very much active in the substantive area of research and some are themselves still working with the data. Structuring these reflections thematically around four rationales for engaging with primary researchers – orientation, navigation, interpretation and the ethics of ownership – I describe the opportunities, tensions and dilemmas that presented themselves as I entered into direct dialogue with a number of the researchers who created the archived data.

The ethics of QSA in relation to the rights and protection of participants, and arguments for and against its methodological merits, have been much discussed (see, for example, Broom et al., 2009, Chauvette et al., 2019, Hughes & Tarrant, 2020, Irwin, 2013, Irwin & Winterton, 2011, Mason, 2007, Moore, 2007). One key point of contention has been the extent to which first-hand connection to the field site and moment of data generation are requisites of analytic insight. Another is the duty of care to participants who have engaged in a reciprocal relationship of trust and disclosure with a particular researcher (or project team) and who cannot realistically anticipate the ways in which their data may be reused and reinterpreted in some unknown future project. However, the expectation for publicly funded researchers (in the UK at least) to make their data openly available at the earliest opportunity,¹ increasing encouragement from journals to make qualitative datasets available alongside publications (Tsai et al., 2016), and the launch of research council funding calls targeted explicitly at data reuse,² suggest that these debates have been resolved in the minds of funders, or at least rendered moot.

Nonetheless, funder endorsement to utilise archived qualitative datasets constructed by others, particularly in cases where one has no prior connection or claim, leaves unresolved quandaries for researchers seeking to conduct QSA, in terms of their engagement with and relationality towards the data set’s originators. As Weller (2022) observes, increasing formalisation of archiving processes creates an air of separation between data originators and potential re-users:

Collaboration is generally not a requirement or expectation of many data sharing policies. The emphasis on open access via centralized repositories and processes imposes a formalized and perhaps more distanced means of data sharing, rather than through personal and professional collaborations … The general assumption is that re-users will work with archived material independently of the original researchers/teams.

(Weller, 2022, p. 10)

Perhaps due to such assumptions of disconnect between originators and reusers, this relational dimension of QSA practice when using ‘other people’s data’ appears to have been rarely written about in the methodological literature to date (though see Coltart et al, 2013, Irwin et al, 2012, Weller, 2019, 2022). I share my reflections here as a contribution to this nascent area of discussion.

As Heaton (2004) notes in her foundational text, researchers conducting secondary analyses stand in different relationships to the data. Secondary analysis may involve only the original researcher(s), a mix of original study team members and new ‘outsider’ researchers, or only new researchers with no previous involvement in the original study. This latter scenario represents my position in relation to the project discussed here, and at the time of Heaton’s assessment of the QSA field, was the least common, representing only 14% (9/65) of the secondary analysis studies reviewed (Heaton, 2004, pp. 36–37). This position seems to have persisted, with a more recent review by Ruggiano and Perry (2019) finding that 84% of secondary analyses involved primary researchers. In this article, I describe my experiences of approaching an archived dataset as an outsider researcher, with no prior connection to the project or primary team.

The remainder of the article is structured as follows. I begin by outlining the primary study and the secondary analysis project I sought to conduct. I then describe my experiences of making contact with three members of the primary team, in the course of developing the research funding proposal that would underpin the secondary analysis project. Following this, I situate my experiences of engagement with the primary team within the broader literature on QSA, focusing in particular on the rationales that are proposed for involving primary researchers in qualitative secondary analysis projects, and reflecting on the ways in which these have played out in relation to my project to date. Drawing on themes within the extant literature and my own motives for engaging with the primary team, I group these rationales under four headings: orientation, navigation, interpretation and the ethics of ownership. The article ends with a brief discussion and conclusion.

The project

The dataset which I sought to utilise is the Welfare Conditionality Project (Dwyer et al., 2019).³ Welfare Conditionality (henceforth WelCond) was a large-scale longitudinal qualitative project, funded by the UK Economic and Social Research Council (ESRC), which ran from 2013 to 2019. WelCond investigated people’s experiences of sanctions and support in the UK welfare system.⁴ The dataset is now held within the Timescapes Archive at the University of Leeds.⁵ The WelCond study involved a large team of seven investigators and six researchers, most of whom were involved in carrying out research interviews, as well as project managers who were involved in preparing the data for archive. The study recruited 481 participants, with three waves of semi-structured qualitative interviews at approximately 12-month intervals, totalling over 1000 transcripts. Participants were sampled according to nine variables of interest: experience of antisocial behaviour orders, disabled people, ex-offenders, homelessness, jobseeking, lone parents, migrant status, social housing residents, and people in receipt of the social security benefit Universal Credit. A publication by a subgroup of the primary team (Dwyer et al., 2020) revealed that around half of the WelCond sample (including participants across all nine groups) talked about experiences of mental health problems during their interviews.

In early 2021, having recently taken up a new position within a research programme focused on work, welfare and mental health, I began to consider the possibility of a secondary analysis project drawing on the WelCond dataset to explore conceptualisations of mental health within the welfare system. Whilst the primary team had analysed the experiences of the subgroup of participants with mental health problems in relation to their experiences of sanctions and support, it had not been their intention to interrogate the fine-grained detail of narrative accounts of mental distress. I saw an opportunity here to ask a new question of the data – to explore more conceptually the lived experiences of distress among a diverse range of people accessing the UK welfare system. Echoing Fielding and Fielding’s (2000, p. 680) position on their secondary analysis of Cohen and Taylor’s (1972) prisoner study, this is not to suggest that the primary team ‘got it wrong’, but rather that there is scope for multiple analytic lenses on rich qualitative data sets such as WelCond, and value to be obtained from probing what more can be gleaned, beyond the substantive focus or practical capacity of the primary study. As described by Irwin and Winterton (2011), I would be ‘prioritising a concept or issue that was present in the original data but was not the analytical focus at that time’ (p. 4).

Whilst secondary analysis may take a similar methodological approach to the original study (Fielding & Fielding, 2000), my own intention was to use a rather different analytic method from the thematic analysis utilised in the primary study. The focus of the secondary analysis is on the use of language and narrative in the subjective interpretation of psychological distress, as expressed by WelCond participants in the course of relaying their substantive experiences of engaging with the welfare system. The QSA will come at the data from a different angle, focusing the analytic lens on words, phrases, conceptual frames and identity positions used by participants to describe and interpret their distress, rather than taking mental ill health as a variable against which to consider the substantive themes of sanctions and support.

At the time of writing, the secondary analysis project had recently been awarded funding under the ESRC’s Secondary Data Analysis Initiative, but had not yet formally commenced. The reflections that follow are thus based predominantly on communications I had with the primary team during the process of developing the funding proposal, as I shaped the secondary analytic project’s research design. The experience of conducting the analysis itself will be the focus of future publications.

Initiating contact with the primary research team

Although I knew several members of the primary research team by academic reputation, I had only ever had passing contact with a few of them and we had never been involved in any direct collaboration. Having gained access to the WelCond data through application to the Timescapes Archive, the dataset felt very much a self-contained entity, with which I had assumed it possible to interact on an independent basis.

However, as I was reading around the methodological literature, one of the things that struck me was the advice and entreaties to make contact with the researchers who had originally generated the data. For example, Tarrant (2017) talks of the numerous benefits gained from ‘on-going discussions with the research teams’ (p. 604), including greater understanding of the context of data production, evaluating the potential of the dataset to address the research questions, obtaining information not contained in publications and advice on sampling cases. More generally, Neale and Bishop (2012b) note that, ‘the process of bringing the originating team into communication with potential secondary users [is] potentially productive for both’ (p. 63).

My first reaction to this advice was that getting in touch with the primary researchers sounded extremely daunting, and something that I would be very apprehensive about. I worried that this would be an imposition on the primary researchers, who had moved onto other projects and were busy with their current concerns. In the case of the Principal Investigator (PI), I knew he was now approaching retirement, and may have no interest in turning thoughts back to this data. In short, my first thought was that I would be a nuisance, were I to make approaches to the primary researchers and ask for information about the study or their perspectives on my proposal.

As a requirement of the Timescapes Archive application process, I had already had to seek written approval from the primary study PI, because the data are of a sensitive nature. However, beyond that, I had naively assumed that using archived data would be a very simple, solitary and boundaried task; that I would be able to treat the data as ‘freestanding’ (Coltart et al., 2013, p. 281) and that my intellectual QSA endeavour would be comfortably lacking in any interactional or relational element!

However, a pragmatic reason for contacting the primary team soon presented itself. As noted above, my intention was to focus on the experiences of the subset of approximately 200 WelCond participants who had spoken about experiences of mental health problems during their interviews. Unfortunately, I discovered that there was nothing in the archived project metadata from which I could readily identify this subset. I wondered if I would simply have to trawl through all 1,081 transcripts to reconstruct the mental health subsample.

I had almost resigned myself to doing so. But then thought I would take a chance on asking a former WelCond Researcher, with whom I had recently become acquainted through other channels, to see if she had access to the subsample details. We had struck up a good professional correspondence around shared interests, and asking for her help felt like a slightly less formidable prospect than reapproaching the PI or ‘cold calling’ one of the Co-Investigators. Yet, the idea of getting in touch to ask about the data still felt uncomfortable and somewhat of an imposition.

As it turned out, the Researcher no longer had personal accesses to information on the mental health subsample, having now moved on to a different institution and project grant (on which, see Weller, 2022) but said that the PI may do, and that she was happy to contact him on my behalf. I emailed back, saying that I was nervous to approach the PI directly, and that her offer of a mediated introduction would be great! This turned out to be really effective; I heard back from the PI within a couple of hours and, although there were limits to the amount of additional detail he was able to share with me, he was extremely helpful in directing me to a way in which I could much more easily identify the relevant subsample.

A couple of months after my contact with the primary study PI and Researcher, I noticed a Twitter post from a PhD student who was also using the WelCond dataset. I got in touch with him for an informal chat to compare notes on our experience of working with the data so far. This was really interesting and helpful in itself, to talk with someone who was also at a similar early stage of feeling their way through unfamiliar data. About a week after this, I received an email out of the blue from one of the WelCond Co-Investigators, who – I hadn’t realised – was the PhD student’s supervisor. I was very surprised and frankly honoured to hear from her, as a senior researcher in my field whose work I had long held in high esteem. The Co-Investigator sent a very warm and enthusiastic message about the work I was trying to develop, and we have gone on to have multiple exchanges via email, videocall and telephone, over the ensuing months, as I describe further below.

Rationales for engaging with primary researchers

In the following subsections, I discuss four thematic rationales or motivations for making contact with primary researchers, which appear in the extant methodological literature and which likewise emerged as considerations in my own project. These are: orientation, navigation, interpretation and the ethics of ownership.

Orientation

Orientation to the primary study – gaining a thorough understanding of the structure of archived data, the composition of the study sample, and the context in which archived data were generated – is one of the main reasons that methodological literature advises making contact with primary research teams (Irwin & Winterton, 2011). Irwin et al. (2012) stress that ‘an awareness of the manifold ways in which context shapes the data which are produced is particularly crucial for secondary analysts, who will have a limited “feel” for proximate contexts of primary data production’ (pp. 74–75). Hinds et al. (1997) contend, in relation to the validity of QSA, that ‘the sensitivity of researchers to the context of the primary study … requires the researchers on the primary study team to adequately sensitize the researchers conducting the secondary analysis’ (p. 414, emphasis added). Evans and Thane (2006) provide an engaging account of written correspondence with a primary researcher to enhance their understanding of the process and context of data collection in an earlier sociohistorical era. They report that this addressed their ‘need to understand the nature of and the reasons for these [methodological] differences in order to contextualize his work’ (Evans & Thane, 2006, p. 79).

In my case, there were a number of reasons why orientation did not initially present as a strong motive for contacting the primary team. Firstly, the WelCond project is extremely well documented and curated. The project website contains a huge amount of methodological and substantive detail about the study, and I had followed the project and its outputs quite closely over its duration, so felt familiar with its rationale and design. The archive is also very comprehensive, with project recruitment materials, interview guides and detailed sample metadata contained, and the complete dataset transcribed and available in electronic form. The timeframes, recruitment sources and geographical locations of the interviews are all documented, and the original project’s aims and objectives can readily be understood from published reports. In contrast to the experience of earlier secondary analysts (e.g. Fielding & Fielding, 2000), the fact that the WelCond data was archived after the shift towards archiving as the expectation in projects funded by UK research councils⁶ meant that the dataset had been proactively and deliberately curated and was thus in much better shape for the prospective secondary analyst, compared to earlier times.⁷ To use Fielding and Fielding and Fielding’s (2000) phrase ‘recovery of the context’ of the primary data was in this case quite straightforward (p. 671).

In addition, the nature and structure of the data felt very familiar to me. WelCond used semi-structured interviews, was addressing a substantive topic very close to my own longstanding area of research interest, and interviews were conducted by a large team of researchers – echoing the type of large-scale team-based applied qualitative research on which I had ‘cut my teeth’ over many years. I was used to working with the transcripts of semi-structured interviews conducted by other research team colleagues, on topics of health, employment and the UK welfare system, and so the structure and content of the material within the archive felt very accessible to me, in a general sense. As James (2012) notes, there are ‘clear parallels between the experiences of the secondary data user and team researchers who analyse interviews that they themselves have not carried out … both are delving into archival memories lodged by unknown others; and for both, the subjective, contextual knowledge is missing’ (p. 566). As James indicates, familiarity with the general style of data does not, of course, negate the lack of embodied experience of ‘being there’, but it has been argued that in some applied qualitative research fields, it is relatively rare for researchers to have extended relational connection with project participants and teams frequently work with interviews they have not personally conducted (Mozersky et al., 2022).

For these reasons, orientation to the dataset was not a primary driver of my contacting the WelCond team, and my sense of familiarity and ease with getting to grips with this type of data likely underlay my initial confidence in proceeding without such contact. It may be that, as archiving becomes a more embedded and systematic element of the process and practice of qualitative research, the reconstruction of context becomes a lesser challenge for secondary analysts. In addition, the contemporary nature of the WelCond dataset certainly made for an easier task of grasping the temporal policy climate, as well as locating members of the primary team for further consultation.

Navigation

I have considered navigation of the archived data as a separate theme to orientation, navigation referring to secondary analysts finding their way around inside the dataset, with respect to sampling and focusing research themes. In my case, sampling was the initial motive for contacting the primary team, beyond the necessary preliminary approach to gain permission to access the archive as a whole. As noted above, I sought the help of the primary study PI (via an approach to a WelCond Researcher) to locate the subset of interviews that included descriptions of mental distress. This was extremely helpful as a labour-saving intervention, and I was able to quickly proceed with constructing a subsample and mapping demographic variables to this subset of participants.

As a slight aside, in the process of later refining the research funding proposal, my attention was brought to the emergent methods of text mining as a tool within large-scale QSA (Edwards et al., 2020). Having initially established the focal dataset via the primary study PI’s assistance, I later found it a useful complement to run a number of keyword searches of the full dataset (using terms relevant to mental distress), utilising the ’breadth and depth’ method developed by Edwards, Davidson and colleagues (Davidson et al., 2019, Edwards et al., 2020). This text mining approach thus offers secondary analysts another productive way of navigating within archived datasets, particularly where primary researchers may not be accessible.

Returning to my communications with the Co-Investigator, in her initial email she extended the following offer:

Feel free to give me a shout once you’re getting into the secondary analysis. I’m happy to meet to help you make sense of how the data is organised … or to talk through decisions like narrowing down the sample etc.

I took up this offer and we went on to arrange an online meeting, where she shared much more background to the WelCond methodology, including the participant classification codes the team had used over the multiple waves of the longitudinal study, and further explained the different variables and descriptors in the metadata. We also discussed her own ongoing analysis of subsets of the data. We continued to have a series of email exchanges, in which she offered valuable additional insights into the way the mental health subsample had been determined (drawing also on the expertise of another primary team colleague who had had closest involvement with this part of the analysis), and the strengths and limitations of the approach used. Despite the comprehensiveness of cataloguing within the WelCond dataset, these additional, informal conversations were incredibly helpful in enhancing my understanding of the dataset and generating further reflections on sampling strategy. Thus, whilst I perhaps could have ‘gone it alone’, these exchanges showed the value of engagement with the primary team in enriching and deepening methodological thinking at the research design stage.

However, my conversations with the Co-Investigator also generated some cause for pause and reflection, relating to the integration of divergent interests across primary and secondary studies. The WelCond study had a major focus on claimant experiences of benefit sanctions in the welfare system and it was through this lens that the Co-Investigator initially offered possibilities on sample construction (for example, focusing on those who had been sanctioned multiple times). My research questions for the planned QSA were coming at the data from quite a different angle than the original study. I wanted to look at mental health conceptualisations and illness representations, using linguistic and narrative approaches. It had not occurred to me that the substantive topic of benefit sanctions might shape my approach to the data, and I was unsure of the fit with my planned methodology. In thinking through the Co-Investigator’s suggestions as to ways in which I might approach the dataset – how to sample and which participant cases might prove most fruitful – questions arose regarding how the research interests and priorities of secondary analysts sit in relation to those of the primary team. The fact that one’s research questions and analytic lens as a secondary analyst may or may not mesh closely with those of the primary team emerges as a potential tension – one that contains both methodological and relational dimensions, and needs to be sensitively navigated. For the secondary analyst, to what extent is it necessary, advisable or desirable to involve primary researchers in shaping the design of QSA studies? For the primary researcher, how does it feel to have an outsider secondary analyst pick up their data and poke at it from a totally different angle, potentially disinterested in those topics and themes that were so central to their original endeavour?

Interpretation

A third potential reason for engaging with the primary research team is to invite their collaboration or assistance in the interpretation of data during secondary analysis. There are again both methodological and ethical aspects to the question of how far primary researchers could or should be involved in sharing in the interpretation and co-production of the analysis. Coltart et al. (2013) suggest that there is a place for ‘receptiveness to feedback from researchers with greater proximate knowledge’ (p. 285) – i.e. the primary research team. In their review of QSA studies, Ruggiano and Perry (2019) note that some secondary analysts reported involving primary researchers as a means of ‘increasing rigour’ (p. 91). At the same time, Neale and Bishop (2012a) argue that secondary analysts, ‘need to be free to ask their own research questions of the data, engage in distinctive modes of analysis, and produce their own interpretations, even where these run counter to primary understandings’ (pp. 5–6).

On a more pragmatic level, Weller (2019) and colleagues found that ‘sustained collaboration … was not something that the original researchers necessarily wanted, expected or could accommodate’. Echoing the findings of Weller’s more systematic consultation with primary researchers, I was aware from informal contacts that some members of the WelCond team had moved on to other projects, other endeavours and were no longer closely connected (practically or emotionally) to the data. What had been all-consuming for a period of their life may no longer be. Irwin et al. (2012, p. 71) note that there is a difference between consulting primary researchers and collaborating with them in dialogic reinterpretation of data. In a similar vein, Hinds et al. (1997) observe that ‘the range of involvement could be from noninvolvment other than providing some forms of data to assuming the full responsibilities of collaborator and co-author’ (p. 413). As Weller (2019) cautions, collaboration over the longer term ‘relies largely on the goodwill of colleagues [and] could result in exploitation’ if ongoing involvement of primary researchers is not supported through additional funding or formalised agreements.

As my QSA project has not begun in earnest at time of writing, is not yet possible to reflect on the experience of secondary analysis itself, or the potential role of any primary researchers from this point onwards (as noted, the reflections contained within this article relate to the proposal development phase). However, at the point of preparing the research proposal, I felt unsure about how much I might seek to engage any of the primary researchers in interpreting or critiquing my later analysis. For me, this was a pragmatic as well as an epistemological question: Would they have time? Would they be interested? But also, if I was honest, how open was I to having my new angle on the data potentially moulded by the primary researchers’ concerns or opened up to criticism or doubt? At the same time, how far could I approach my proposed project as standalone and detached from the ongoing presence in the field of those who generated and archived the data, only a few years ago? Was it possible to treat the data as freestanding, detached from the insights and contextual knowledge of those who undertook the primary study? How aggrieved might they be if I didn’t ask for their input at all? This led me to ponder further questions of whether primary teams necessarily need to approve, endorse or agree with the findings of a secondary analyst. Application to the Timescapes Archive requires an outline of the proposed use(s) to which the data will be put. Where data are of a sensitive nature, prior approval is also required from the primary study PI, who also reviews the outline proposal. As such, the PI has some gatekeeping power over who is permitted to access the data and how they reuse it. However, there is no requirement to consult with any wider members of the primary team, who may often have had a closer involvement in generating the data, analysing it and building relationships with participants. This brings us to the final consideration, the ethics of ownership.

Ethics of ownership

As the archiving and reuse of qualitative data become more widespread and standardised, Weller (2022) has foregrounded further ethical issues relating to the ‘detachment of intellectual knowledge and outputs from the producers of that knowledge’ (p. 3). These concerns go beyond long-debated matters of contextual understanding and interpretation, to the ethics of ownership and emotional labour. In formalising practices of qualitative data archiving, research becomes a ‘public good’ separated from the material, emotional and relational investments of those who (co-)created it (Weller, 2022, p. 3).⁸ Yet, as noted by Weller and Edwards (2022), ‘a sense of ownership can emerge from the emotional and intellectual connections fostered through the temporal resources, commitments, and care researchers invest in the relationships that are central to the production of qualitative data’. Similarly, Hughes and Tarrant (2020) observe that, ‘to re-use those data means re-using the intellectual labour of the original researchers … Questions of endeavour, interpretation and labour including emotional investments [are] raised’ (p. 40).

As I proceeded through the development of my QSA proposal, I felt a growing personal conflict about a sense of profiting from others’ work. Having worked on qualitative longitudinal studies in the past, I knew how much effort goes into recruiting and retaining participants and generating multiple waves of interview data. It seemed almost too good to be true that I could – within a couple of weeks – apply for and download an enormous dataset of hugely rich material, essentially on a plate. It might be argued that this is what researchers agree to when they deposit their data in an archive, but regardless of the requirement from funders, and the fact that clearly the project team had prepared and signed over the material to the Timescapes Archive, it seemed to me that this still needed some pause for thought. My anxieties were heightened because of the ‘living’ research team and the newness of the dataset. The data were barely a few years old and most of the researchers involved were still very much active in this field of policy research. To simply parachute in and start profiting from ‘their’ material – through my own research grant and associated publications – did not sit quite comfortably with me.

Moreover, I became aware that some members of the team were themselves still in the process of further analysis of the dataset, through deeper interrogation of certain themes or through novel theoretical approaches to the data (which Heaton, 2004, terms supplementary and supra analysis, respectively). Thus, as highlighted by Weller (2022, p. 7), notwithstanding principles of ‘reasonable first use’, the open-ended and non-linear process of qualitative analysis means that there is evident potential for secondary analysts to begin drawing upon the labours of others before the data originators have concluded their own scholarly work. The Timescapes Archive was indeed always intended to be a ‘living archive’ whereby both primary and secondary researchers were encouraged to (re)engage with longitudinal data whilst projects were still in progress (Neale & Bishop, 2012a, b). To ensure that this could be achieved ethically and safely, Timescapes was founded on an ethos of a ‘stakeholder approach’ to archiving and re-use. The stakeholder approach seeks to enable ‘an appropriate balance between the needs and requirements of research participants, primary researchers, secondary users, funders, and the wider public’, enabling access whilst minimising risk (Neale & Bishop, 2012a, p. 1). The themes of the present paper clearly speak to this concern for stakeholder ethics, in particular those pertaining to relationships between primary researchers and secondary users. Weller (2019) describes how a concern to acknowledge the emotional labour and connection of the primary researchers to their archived data drove her team’s approach from the outset: ‘Even though the original teams had archived their data for the purpose of re-use, we ought, in our negotiations about the secondary analysis of their material, to be sensitive about such long-term connections and the emotional investment made by the researchers’. Hence Weller and colleagues proactively engaged the primary team in their plans at the project development stage and beyond.

I confess that such considerations occurred to me in an ad hoc and retrospective way, having seized initially on the exciting fact of the dataset being available for use, and only latterly beginning to reflect on what this meant for my relationship and ethical ‘habits of care’ (Weller, 2022) towards the primary team. My contact with the primary researchers has indeed been ‘emergent, iterative and unexpected’ (Weller, 2022, p. 11).

To my relief, the language used by the Co-Investigator in her initial email correspondence and our later telephone and online conversations have indicated no sense of protectiveness or hesitance about me undertaking this work. Indeed, there was explicit support from the outset that my proposed project was ‘very much needed’ and ‘sounds absolutely brilliant’. My concerns about profiting from others’ labour seemed to be unfounded.

During a session of the Timescapes 10 Festival (September 2022), I had the opportunity to ask the Co-Investigator directly how she felt about people using the WelCond data. She described how she had initially experienced a sense of ‘panic’ as people started to make requests to access the data, because the team still had not written all the things they wanted to write, from the primary project. However, over time it had become apparent that there was no practical possibility that they would ever be able to write everything that they wanted to, so she had come to an acceptance – a kind of resignation and ‘letting go’ of the hope of getting it all done, and likewise ‘letting go’ of the data in some respects. She said that there remained so much untapped potential in the dataset that it was better that somebody was doing the work, even if not them. In subsequent correspondence, the PI expressed a similar sentiment, that he was pleased that others like myself were able to make further use of the anonymised data within the archive.

Thompson (2004) offers similar reassurance in his reflection that seeing other researchers flock to make use of his data was ‘an enormous source of satisfaction’ (p. 84) and led to many more publications than his primary team alone could have achieved. He suggests that primary researchers stand to gain ‘great pleasure and pride in the long run’ (ibid. p. 84) from offering their materials for re-use. However, I think I would have felt less conflicted had I been working with an archive of material that was (like Thompson’s) over 20 years old at the point of deposit, when the people who had invested their labour in creating the data were not still very much in the process of enacting and building their own research profiles and careers. On ethical questions of ownership, I found Coltart et al.’s (2013) article particularly helpful, in thinking through these issues of seeking feedback, positioning oneself as an ally and not a competitor, and acknowledging the practical, emotional and intellectual investments of the primary research team.

More recently, Weller (2022) has offered detailed proposals on the ethics of care in data sharing, which acknowledge not only ownership of the substantive epistemic materials generated during the study, but also the (unequal) investments of emotional, ethical and intellectual labour made by primary research team members in preparing data for archiving and re-use. The WelCond team comprised numerous researchers and administrators with varying roles and different durations of involvement with the study. The team members I had most contact with during this process of proposal development were the PI, one Co-Investigator and one Researcher, all of whom had conducted some of the qualitative interviews. As was relayed to me by one of these individuals, on very large team-based projects such as WelCond, researchers will inevitably have different levels of emotional investment in the qualitative interview data. Their feelings about the archived dataset may vary from deep attachment to relative nonchalance, if their role in the project was brief, transitional or fairly minor in their overall career trajectory, and these feelings may change over time. The short-term nature of academic research contracts in the UK (and beyond) also means that some primary researchers may not have the opportunity to develop the deep and invested relationship to their data that they would like to.

Discussion

A complete lack of involvement in the primary research study poses certain practical and methodological challenges to secondary analysts. As noted by Weller (2019), having had some involvement in primary studies can both help to better understand the origins of the data, and also facilitate relationships with primary researchers when combining datasets for secondary analysis. I did not have this advantage in the present study, and broaching contact with lead members of the primary team initially demanded some courage. My apprehension had two dimensions. On the one hand, I was concerned about encroaching on the time and goodwill of the primary team, in assisting with navigating the dataset. On the other hand were wider concerns about the ethics of ownership and profiting from the labour of others.

When the research team is still live – by which I mean both living and ‘live’ as in still working with the data – there are additional considerations to when using a historic archive. As recognised by Coltart et al. (2013), ‘concurrent primary and secondary analysis magnifies a number of issues; particularly the ethical boundaries between researcher, participant and secondary user, which are often relevant but less acute for the analysis of historical datasets’ (p. 289). My early experience in working with the WelCond dataset very much brought these considerations to the fore. There is a need to think about how your research will be received by the primary research team and to explicitly acknowledge their labour – the practical and emotional investment they have made – and their generosity in making this data available. This again speaks to the Timescapes principle of stakeholder ethics. As Neale & Bishop (2012a) note, ‘bringing [primary and secondary] researchers into conversation with each other may help to foster an ethic of mutual care and respect’ (p. 1). I would now always consider it a mark of courtesy to make contact with the primary team, thank them and let them know your plans, at the very least. That said, I am conscious as I write this that, although I have entered into an open dialogue with selected senior members of the primary research team, there are many other researchers who invested significant labour in the WelCond project, with whom I have had no contact. These researchers are likely unaware (until now, perhaps) that I am beginning to engage deeply with their past work, spending hours immersed in their interviews and their voice, and building a sort of ‘parasocial’ academic relationship as I come to know their craft skills and habits of practice without them necessarily having any awareness of this.

To sum up, I am pleased to say that my initial apprehension about being a nuisance to the primary team appeared to be unfounded. People who had finished their involvement in the project, people who were still working with the data and people who were on the verge of retiring altogether were all willing to help me out with access and information, to the best of their abilities. There has also seemed to be no contention around this notion of profiting from their work. Quite the opposite – they have seemed delighted that somebody else is going to attempt to make further use of their rich dataset, and I have had a number of ensuing contacts with members of the team that have been both supportive and generative to my thinking as the project develops. What initially felt daunting now feels like a valued set of ongoing conversations, and my apprehension has dissipated as the primary researchers with whom I made contact have continued to be enthusiastic about my secondary analysis endeavours. Thus far, I have not perceived any signals that I have become an inconvenience or am making unreasonable or intrusive bids for their attention. Yet I remain mindful of the limits of their capacity for collaboration and communication, and of the need for care and consideration in the requests I make of them.

All of the above points to important questions about relationships between primary and secondary analysts, including dimensions of ownership, control, epistemic authority and gatekeeping powers. The primary team will be immersed in their knowledge of the dataset and the connections they built with the participants and the settings. They have embodied experience of doing those interviews, co-constructing that data, and the primary privilege of first sense-making. My journey into QSA has brought me to recognise that beyond the practicalities of access and orientation to the structure of an archived dataset, there are relational aspects to secondary data use, including careful acknowledgement of the investment of the primary team, and – in the course of consulting with primary researchers and benefiting from their deep knowledge of datasets – the negotiation of how integrated or divergent analytic interests will be respectfully navigated. Whilst my own initial anxieties centred around burdening the primary team with requests on their time and exploiting their labour for my own ends, the concerns raised in this paper are not far removed from broader considerations of power and hierarchy in research. Differences in academic position and career stage may also bring concerns for the secondary analyst who feels unsure of their footing when drawing on the work of more senior or renowned primary researchers.

Conclusion

My experience of engaging with primary researchers when designing a QSA study has been a journey from initial reluctance to even initiate contact (borne of a combination of introversion and fear of intrusion), through my tentative (and readily granted) requests for further information, to an open, ongoing and supportive dialogue with key members of the WelCond team. In further conversations, we have discussed the complex emotions raised for both primary and secondary researcher, when thinking through the interpersonal dimensions of working simultaneously on – but with very different positionalities and relationships to – a ‘living archive’. As ethical and methodological debates about QSA evolve, the voices of those who have deposited their research labours in contemporary archives also need to be heard. More publications from primary researchers reflecting on their experiences of seeing their data reused by familiar and unfamiliar secondary analysts would be welcome.

The relationship of secondary analysts to the original project will shape the extent to which engaging with the primary team constitutes a step into the unknown. In many cases, secondary analysis is conducted by primary researchers, sometimes in collaboration with new researchers (Heaton, 2004, Ruggiano & Perry, 2019, Weller and Edwards, 2022). In my case, there is no current plan to conduct analysis collaboratively with members of the primary team. Thus it remains to be seen how far and in what ways our correspondence evolves as the project continues. However, as the encouragement of rapid archiving practices and ‘open qualitative research’⁹ edges us towards a new era of possibility in working with contemporary archives and living research teams, I hope that my experience with the WelCond project offers a useful contribution to this advancing area of methodological debate and development.

Acknowledgments

The author gratefully acknowledges the Welfare Conditionality Project team who have made their data available via the Timescapes Archive [data set identifier: https://doi.org/10.23635/13]. Particular thanks are conveyed to Dr Katy Jones, Professor Sharon Wright, Professor Peter Dwyer, Dr Alasdair Stewart, Dr Thomas Rochow and Cassandra Lovelock. Three anonymous reviewers provided extremely helpful feedback which improved earlier drafts of this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the ESRC Centre for Society and Mental Health at King’s College London [ES/S012567/1] and the ESRC Secondary Data Analysis Initiative [ES/X002101/1]. The views expressed are those of the author and not necessarily those of the ESRC or King’s College London.

Notes on contributors

Annie Irvine

Annie Irvine is a qualitative researcher with expertise in mental health, employment and welfare systems. Currently at the ESRC Centre for Society and Mental Health, she is developing a programme of qualitative research on transitions between welfare and employment for people with experience of mental health problems, in the context of conditional welfare and a precarious labour market. Annie prioritises research that has policy and practice relevance, whilst also exploring social conceptualisations of mental health.

Notes

1. https://www.ukri.org/manage-your-award/publishing-your-research-findings/making-your-research-data-open/

2. https://www.ukri.org/opportunity/esrc-responsive-mode-secondary-data-analysis-round-one/

3. By naming the project here, the identity of the primary research team becomes visible. I have obtained approval from those who are quoted or directly referenced in this article, to name the project and, in doing so, to lay open the possibility that their identity might be inferred by readers.

4. Extensive reporting of the primary study can be found at: www.welfareconditionality.ac.uk

5. The Timescapes Initiative was an ESRC-funded project launched in 2007 with the aim to advance qualitative longitudinal research and practices of qualitative data archiving and re-use. Eight projects were conducted under the original initiative and formed the initial archive content. Further datasets, including WelCond, have since been added to the archive. Researchers from original Timescapes projects have worked collaboratively across their datasets, combined secondary data with new primary data, and developed a wealth of resources and guidance for secondary analysts. See: https://timescapes-archive.leeds.ac.uk/.

6. See: https://www.ukri.org/about-us/policies-standards-and-data/good-research-resource-hub/open-research/

7. The Timescapes Initiative has had a significant role in shaping and improving these practices. Indeed, the WelCond anonymisation protocol explicitly acknowledges the assistance and expertise of the Timescapes team in designing their archiving process.

8. Relating to this point, Heaton (2004, p. 12) distinguishes formal from informal data sharing, the latter involving primary researchers directly sharing their data collections via special request or private networks. With the introduction of the GDPR and the move towards formalised standards of data archiving and gatekeeping, one might presume informal data sharing is on the decline – a question that would be interesting to explore.

9. https://www.sheffield.ac.uk/ihuman/our-work/human-futures/fostering-cultures-open-qualitative-research