Contributors are representative, as long as they agree: How confirmation logic overrides effort to achieve synthesis in applied health research

Abstract Introduction The paradox of representation in public involvement in research is well recognized, whereby public contributors are seen as either too naïve to meaningfully contribute or too knowledgeable to represent ‘the average patient’. Given the underlying assumption that expertise undermines contributions made, more expert contributors who have significant experience in research can be a primary target of criticism. We conducted a secondary analysis of a case of expert involvement and a case of lived experience, to examine how representation was discussed in each. Methods We analysed a case of a Lived Experience Advisory Panel (LEAP) chosen for direct personal experience of a topic and a case of an expert Patient and Public Involvement (PPI) panel. Secondary analysis was of multiple qualitative data sources, including interviews with the LEAP contributors and researchers, Panel evaluation data and documentary analysis of researcher reports of Panel impacts. Analysis was undertaken collaboratively by the author team of contributors and researchers. Results Data both from interviews with researchers and reported observations by the Panel indicated that representation was a concern for researchers in both cases. Consistent with previous research, this challenge was deployed in response to contributors requesting changes to researcher plans. However, we also observed that when contributor input could be used to support research activity, it was described unequivocally as representative of ‘the patient view’. We describe this as researchers holding a confirmation logic. By contrast, contributor accounts enacted a synthesis logic, which emphasized multiplicity of viewpoints and active dialogue. These logics are incompatible in practice, with the confirmation logic constraining the potential for the synthesis logic to be achieved. Conclusion Researchers tend to enact a confirmation logic that seeks a monophonic patient voice to legitimize decisions. Contributors are therefore limited in their ability to realize a synthesis logic that would actively blend different types of knowledge. These different logics hold different implications regarding representation, with the synthesis logic emphasizing diversity and negotiation, as opposed to the current system in which ‘being representative' is a quality attributed to contributors by researchers. Patient or Public Contribution Patient contributors are study coauthors, partners in analysis and reporting.

which goes beyond involvement to encourage active partnership between researchers and public contributors. However, coproduction lacks a consistent definition and can be a contested activity in health research. 2,3 One common debate concerns the 'representativeness' of public contributors (the individuals who become involved in active or collaborative ways with researchers and the research process, as opposed to being participants in research or recipients of research findings). The 'professionalization paradox' 4 describes the concern that patients or members of the public who become involved in research are required to possess or gain an expert familiarity with the research process, which threatens their ability to reflect genuinely 'lay' viewpoints.
The coproduction literature has previously explored this paradox and critiqued the contradictory nature of the demands placed on contributors (e.g., Peter Beresford's work in Disability studies, 5 and Diana Rose in mental health 6,7 ). Contributors are expected to bring individual and personal experiences yet also transcend them, and to be knowledgeable enough to have an informed opinion on research but not lose their research-naïve public viewpoint. [8][9][10][11] Previous research has further argued that this characteristic of 'being representative' is used as a rhetorical device to reject input when contributors seek to make changes. Analyses in both health research 12 and health commissioning 13 therefore position the debate about 'representativeness' as a classic form of boundary defence employed by researchers against suggested changes, achieved by challenging the legitimacy of contributors' knowledge as lay people. 14 Although boundary defence may offer one understanding of the paradox of representation, there remain concerns in health research that lack of representation is a problem in PPI, which may undermine its impact or relevance. A particular target for the critique of requiring capacity to be representative are expert contributors who contribute across many studies or who have been involved in research over many years. These have been referred to as 'super patients' whose experience and expertise are seen as undermining their ability to represent lay perspectives. 15 An NIHR review identified on the one hand wariness from some researchers about using experienced contributors as their expertise may dilute their experience, while other researchers felt such specialized expertise and commitment was both necessary and beneficial. 16 Similarly in the quality improvement field, there is recognition that patient contributors may require particular skills or confidence to contribute meaningfully alongside professional stakeholders, but corresponding concern that such atypical contributors cannot reflect the concerns of wider patient populations. 17 One possibility which is under-explored in this debate is that both expert and lay or research naïve contributors may be required in different research activities and contexts (an option articulated by Kristina Staley 18 in her response to one of the critiques). Research exploring the perspectives of public contributors themselves has observed that different roles can be performed by contributors within research. 19 In a recent study of PPI across an academic health science network, 20 Barker and colleagues proposed nine distinct roles for contributors, with one group relating to lived experience but other roles themed around citizenship or acting as an outsider. It is likely therefore that questioning whether contributors should be an expert or naïve is an over-simplistic approach, as in practice there is a need for a diversity of roles, expertise and experience that can be performed by different contributors.
A second area of neglect in this debate is the absence of the voice of contributors themselves. Critiques of representation in PPI have largely been written by researchers-and tend to favour their perspectives as opposed to those of public contributors. While not a systematic assessment, we observe in the 16 research papers referenced so far that there was no patient involvement in nine of the studies, patient involvement described in the study but without PPI input in the paper itself in three, 13,16,19 involvement in the study and paper of an academic/ academics who identify as a service-user researcher or equivalent in three 3,9,10 and inclusion of a contributor coauthor alongside academic coauthors in only one. 20 There is a need for more active collaborative analyses between researchers and contributors, exploring representation from both perspectives. Public contributors themselves are reported to be acutely aware of the paradox of representation, for example, reporting a perceived need to deliberately emphasize or underplay their expertise depending on the context, in order for their perspectives to be taken into account. 21 This suggests that efforts to resolve the paradox or progress the debate need to work with contributors who have direct experience in navigating these tensions in practice. This paper addresses this issue by reporting a collective analysis undertaken by a team of researchers (S. E. K., S. D., R. K., R. B.) and contributors (P. W., J. F.).
We report a study conducted within the Greater Manchester

| METHODS
The setting for the study was an applied health research collaboration in the North West of England. For the purposes of this study, we completed a secondary analysis of qualitative data derived from two cases, which were selected from cases within a larger study exploring collaborative practice across the CLAHRC themes. The two particular cases were chosen as both involved data collected about the PPI activity and impact. Description of each case and corresponding data are presented in Table 1.
Qualitative case study research can involve a bricolage of sources and methods, as the definition of the case is emphasized over the data collection method used. 26 Secondary analysis can similarly involve drawing on diverse sources of secondary data to produce 'created assemblages' that structure a comparative analysis. 27 The assemblage is created to address questions that would not be answerable by analysis of a single project. In this study, we defined our cases according to what was being observed, meaning the discussion of public contributors, their role in research and how the discussion of being a representative was part of this, rather than the cases being defined through having equivalent data sources. Most importantly for this study, the cases were chosen because they enabled a comparison relevant to both research question 1, that is, provided an exemplar of lived experience PPI and expert contributor PPI, and research question 2, that is, provided data that enabled analytic interpretation of both contributor and researcher perspectives.

| Analysis
Each member of the analysis team had been involved in the primary data collection in one or both cases (although the contributor coauthors had only been involved to date as participants, rather than as collaborators in study design or data collection), but all of them contributed together to the secondary analysis described here.
The analysis team included P. W. and J. Representation was apparent as a key factor in two contrasting ways across the two cases.
First, representation was brought up explicitly in the LEAP case in the interview data from researcher participants, where it was used to justify a decision to reject contributor input. This is consistent with previous research indicating that representation is employed by researchers as a way to delegitimize contributor input when they seek instrumental change. It demonstrates however that representation critiques are not limited to 'expert' patient groups, but can also be employed when side-lining those with direct lived experience.
Second, and contradicting the questioning of representation as described above, representation appeared to be unequivocally described in relation to the expert Panel as providing 'the patient view' in documentary data from the highlight reports. This suggests that not only was representation only questioned when refuting changes suggested by contributors but that representation is assumed when contributor feedback could be used to express support for researcher decisions.
We present illustrative data of these observations below.
3.1.1 | Use of representation to reject contributor input in the LEAP case Representation was employed by two of the researchers in the LEAP case specifically to question the legitimacy of contributor input when the contributors had requested changes be made. The decision to reject the LEAP input within this narrative is justified through reference to representation: There's the concern that … a PPI group of eight people

| Analytic comparison of the two logics
It is notable in the data that synthesis with researchers is aspirational for the contributors, but not fully realized in practice. This can be seen as due to the more powerful group in this case (the researchers) being able to enact the confirmation logic, which our collaborative analysis group observed to be in opposition to the synthesis logic in three key ways: 1. While the synthesis logic encourages interaction and blending of different perspectives, the confirmation logic seeks validation of existing perspectives.
2. The synthesis logic emphasizes polyphony, with contributors themselves bringing multiple different experiences, while the confirmation logic seeks a monophonic 'patient view'.
3. The synthesis logic emphasizes interaction and negotiation, while the confirmation logic operates as a discretionary decision made by researchers about contributor input without accountability or transparency back to those contributors.
Consequently, we conclude that the synthesis logic is currently predominantly aspirational and is inhibited in practice in cases where the confirmation logic is enforced.

| DISCUSSION
Consistent with previous research, we found that representation is used by researchers as a means to defend against changes initiated by contributors. We further demonstrate that this is not limited to expert contributors but is applied to contributors with specific lived experience. This suggests that the relative expertise and experience of the contributor are secondary to whether the contributor is requesting changes be made. Our findings are therefore consistent with studies suggesting that challenges to legitimacy are most often encountered as a form of defence against instrumental impacts of PPI. 12 Problematically, contributors are representative as long as they agree.In this paper, we have expanded further on this through the analytical observation of the confirmation logic, whereby the representation of a monophonic patient perspective is assumed when it can be used to confirm-justify or support-existing researcher decisions. We acknowledge that our analysis is based on only two cases and that our conclusion of a confirmation logic particularly is based on secondary documentary data reported for other purposes rather than being articulated in direct interviews. We suggest however that researchers in interviews may not have wished to express this logic so openly, and therefore the secondary analysis is a strength of the study. The confirmation logic also has support in other literature on patient involvement. For example, an evaluation of the impacts of James Lind Alliance Priority Setting Partnerships found that 'researchers tended to use a Top 10 priority to strengthen the case for a study they already planned to do', finding no evidence that an existing research topic was changed to accommodate a new priority. 28 In the service improvement context, similarly it was reported that 'involvement was used instrumentally by programme leaders to gain support for change the case for which had already been made, and for service models already developed'. 29 At a systemic level, Montenegro and Cornish have argued that the role of user groups in mental health reform in Chile was driven, and later undermined, by the use of user input for legitimation. 30 We suggest therefore that our proposed confirmation logic has validity beyond the present study and the specific CLAHRC setting, and offers a succinct way of conceptualizing this common observation.The key analytic contribution of this paper has been to contrast this conformation logic with the synthesis logic held by contributors themselves. We observe the following implications.

| Comparing synthesis and confirmation logics in involvement 1: Monophonic versus polyphonic accounts of patient perspectives
First, there is an expressed contradiction between the researcher's preference for a monophonic patient view and the plurality of voices that are considered essential by contributors. The tendency to frame PPI as providing a singular 'patient perspective' has been reported previously by Rowland and colleagues, 31 with the recommendation that clarity is needed about what 'patient voice' is being represented and how, for example, referring to democratic, statistical or symbolic representation, which each have different requirements in terms of who is involved in research and how. 32 If it is indeed a consensus opinion being sought, then researchers would need to adopt approaches that deliberately define and seek views of a representative group. There are established participatory methods, which deliberately seek to engage representative cohorts, for example, citizen juries. 33 We observed in the documentary analysis however there is currently an inconsistency in how contributors were recruited (emphasizing personal experience) and how their input was then framed in reporting (representing collective opinion). Future research should aim to provide a clearer articulation of the purpose of the involvement and the criteria on which representation is to be judged. Alternatively, the polyphonic contributions described by contributors themselves may be considered more appropriate or valuable, depending on the research context. To achieve this multiplicity of viewpoints, more focus is likely needed on supporting diverse contributors to access involvement opportunities, and on providing inclusive spaces, which can effectively elicit and explore such differences.We note that in both of the cases, the onus is on researchers themselves to understand which approach is required and to make efforts to support this, positioning researchers themselves as responsible for transparently articulating what they assume contributors offer. 34 This may be a positive direction to take, given that to date the emphasis appears to be on how contributors themselves handle the paradox, including through rhetorical use of the collective voice 35 and negotiation of their 'symbolic capital' as patients. 21 We suggest that instead researchers themselves could be considered to hold a responsibility not to place paradoxical demands on contributors. This is not however solely the responsibility of individual researchers but should be understood as operating within a wider research context that limits opportunities for flexibility and change and incentivizes an extractive approach to involvement. 36 Within applied health research, Papoulias and Callard identified how both organizational activities and researcher behaviours, such as a focus on deliverables, create spatial and temporal logics that constrain the potential for involvement. 37

| Comparing synthesis and confirmation logics in involvement 3: Exploring how synthesis occurs
Finally, the synthesis logic that we describe is worthy of further investigation, particularly as this logic is consistent with current understandings of and recommendations for the production and coproduction of knowledge in health research (referring to 'mode 2' approaches, which emphasize the need for blending of different ways of knowing to inform health research 45 ). In particular, it will be valuable in the future to better understand the nature of synthesis and its processes and impacts, particularly considering Bakhtin's idea of polyphony, which has been expressed in previous studies of involvement 11,24 and may particularly align with the synthesis logic identified in this study. In contrast to Barker

| Reframing the call for diversity in patient involvement
It should be understood that a focus on synthesis in the way we describe does not negate consideration of whose contributions are included and who may be excluded. This consideration was apparent in the contributors' discussion of the need for diversity of experience, and diversity, rather than representation, may be a productive focus for future work. Across different fields, there is consensus that coproduction efforts should explicitly attend to the need to include different ways of knowing and addressing who has power and legitimacy in these processes. 51 Rather than only questioning whether 'the usual suspects' of white, professional, educated contributors can represent other views, we should also question how and why people who do not match that profile have been excluded to date.

| Strengths and limitations
The data sources were retrospectively analysed, and we are drawing conclusions based on this secondary analysis rather than having explored this actively with participants. We acknowledge a lack of comparative data, particularly in the panel case where researchers were not directly interviewed. However, drawing on diverse forms of available data provided a novel insight into how PPI input was being described in researcher reports. It is not uncommon for case study analysis to draw across diverse naturalistic sources, and it is important to recognize that we do not suggest a cumulative data set was achieved (meaning the data is automatically comparable and can be combined), but rather we judged we had sufficient data to enable a meaningful analytic conversation across the cases that were chosen relevant to the two research questions. 52 While we were a partnership in relation to data analysis and writing this paper (and agree with calls for greater clarity in reporting patient authorship 53 ), the contributors were not involved before this in design or data collection. Researchers should be mindful that there remains a gap regarding studies that involve contributors fully and throughout the research process, from conceptualization to reporting.We were prompted to acknowledge, by an anonymous contributor reviewer, our choice of language, for example, around descriptions of 'logics', and whether this is accessible to lay audiences. This is an important consideration because it can be seen as another layer of exclusion that prevents contributors themselves from engaging with debates, meaning discussion is about them, without them. We explored this as a team and discussed our decision to adopt language and terminology that would be persuasive to our target audience, which in the case of this research, we felt to be the research community. The contributor coauthors described this as a deliberate compromise, that 'we cede the battle [around using more accessible language] but we may win the war, by converting researchers to appreciate our point of view'.
Although our own group found this to be acceptable in this instance, it is an important area for future discussion, and finding mutually acceptable and valuable ways to express findings is yet another area requiring open discussion and negotiation. We also acknowledge this approach is not without risk, as we may inadvertently contribute to the privileging of knowledge that is expressed according to academic norms. We would in fact encourage researchers to recognize that sophisticated analysis of involvement concepts such as representation is often produced outside academic channels by contributors themselves (see e.g., 'The Rep Trap' by David Gilbert, 54 which astutely and critically addresses the challenge of being deemed a 'patient representative').

| CONCLUSION
We began the study with the expectation that the expert Panel may be subject to challenge over-representation in a way the livedexperience LEAP was not. Our findings show that relative expertise is not the deciding factor in whether representation is critiqued or not.
Instead, the driver behind this appears to be the confirmation logic, whereby representation is assumed or revoked by researchers based on whether their own decisions are supported or critiqued. This demonstrates that despite ostensible commitments to equal partnership, researchers continue to hold power over decision-making in research involvement, limiting the potential for contributors to have influence. We observed that contributors, by comparison, hold a synthesis logic, and made suggestions for how this synthesis process can be better understood in the future. We note however that such effort to understand this logic will be irrelevant, if, in practice, it cannot be enacted. Surfacing tensions in approaches to coproduction is essential to move beyond misleading debates about representation.
Involving contributors themselves in these debates is both a necessity for making progress and, we suggest, an ethical responsibility.

AUTHOR CONTRIBUTIONS
All authors were involved in conceptualizing the study. Data was collected by Roman Kislov, Sarah Darley and Sarah E. Knowles. All authors contributed to the data analysis. Sarah E. Knowles prepared the first draft of the paper. All authors commented on and approved the final submission.