Academic advocacy in public health: Disciplinary ‘ duty ’ or political ‘ propaganda ’ ?

The role of ‘ advocacy ’ within public health attracts considerable debate but is rarely the subject of empirical research. This paper reviews the available literature and presents data from qualitative research (interviews and focus groups conducted in the UK in 2011 e 2013) involving 147 professionals (working in academia, the public sector, the third sector and policy settings) concerned with public health in the UK. It seeks to address the following questions: (i) What is public health advocacy and how does it relate to research?; (ii) What role (if any) do professionals concerned with public health feel researchers ought to play in advocacy?; and (iii) For those researchers who do engage in advocacy, what are the risks and challenges and to what extent can these be managed/mitigated? In answering these questions, we argue that two deeply contrasting conceptualisations of ‘ advocacy ’ exist within public health, the most dominant of which ( ‘ representational ’ ) centres on strategies for ‘ selling ’ public health goals to decision-makers and the wider public. This contrasts with an alternative (less widely employed) conceptualisation of advocacy as ‘ facilitational ’ . This approach focuses on working with communities whose voices are often unheard/ignored in policy to enable their views to contribute to debates. We argue that these divergent ways of thinking about advocacy speak to a more fundamental challenge regarding the role of the public in research, policy and practice and the activities that connect these various strands of public health research.


Introduction
The idea that public health, by its very nature, entails advocacy, appears to be ascendant. To take one example, a 2014 letter published in the high-profile medical journal, the Lancet, written by five public health registrars, argued that 'it is the duty of UK public health institutions to advocate strongly for evidence-based measures to improve the health of society' (Tillmann et al.,p.213). Whilst difficult to gauge in quantifiable terms, searching the academic databases Web of Science for "public health" AND (advocacy OR advocat*) in 2007e2016 returns over three times as many hits as the previous decade.
The roots of this idea are long-standing. Rudolph Virchow, a nineteenth century Prussian medic, famously argued that, for medicine to accomplish its aims, 'it must intervene in political and social life' to highlight 'the hindrances that impede the normal social functioning of vital processes, and effect their removal' (Virchow, 1985, p.33). More recently, public health frustrations regarding deaths and other health harms arising from conflicts (Shenoda et al., 2015), weak responses to pandemic disease outbreaks (Timen et al., 2015), failures to tackle health inequalities (Mackenbach, 2011) and cuts to health system financing (Karanikolos et al., 2013) have all been cited as evidence of the need for advocacy. In some contexts (particularly the UK), the emphasis that research funders have begun placing on 'research impact' has provided further impetus for public health researchers to try to achieve political and social influence (Greenhalgh and Fahy, 2015). Yet, there has been little attempt within contemporary public health to examine what advocacy means in practical terms (Horton, 2012), who ought to be undertaking this kind of work or what exactly it involves.
For those who promote the need for public health researchers to engage in advocacy, the relationship between evidence and advocacy is often assumed to be relatively straightforward; once sufficiently robust evidence is available, it can be used to advocate for change (Tabak et al., 2015). Yet, Roberts (2009, p.46) argues that doctors' role as advocates for social justice' has been discouraged in medicine due to the 'heavy weight of conservatism' and tradition in the profession (see also Chapman, 2007). Indeed, the role of advocacy in public health has often proved controversial. Even in tobacco control, where the overlap between research and advocacy has been particularly strong in recent years (Smith, 2013), Richard Doll and Austin Bradford-Hill (who together helped demonstrate the link between smoking and lung cancer (Doll and Hill, 1956)), were extremely cautious about engaging in anything resembling advocacy for much of their academic careers: 'At that time (1945e1960) I held the view, indeed I held it very strongly, that the researcher faced with positive results, as we were, had no part to play in telling the public about those results, and still less in how it should behave. Any education aimed at changing habits must to some degree smack of propaganda even in the best sense of that word … ' (Bradford-Hill, 1971, p.57).
In an era of 'knowledge exchange', 'public engagement' and 'research impact' (Greenhalgh and Fahy, 2015), reservations about simply 'telling' the public about research results seem almost unfathomable. However, other examples continue to court controversy. Editor-in-chief of The Lancet, Richard Horton, has explicitly stated that he aims 'to use science as a political instrument to promote social justice' (Shalan, 2013;unpaginated) and has attracted condemnation for his perceived 'longstanding and wholly inappropriate use of The Lancet as a vehicle for his own extreme political views,' (Professor Sir Mark Pepys, quoted in Wallis Simons, 2014). In response, Horton has argued that politics and health 'go hand-in-hand' (Wallis Simons, 2014;unpaginated).
Against this backdrop, this article draws on existing literature and qualitative data to ask the following questions: (i) What is public health advocacy and how does it relate to public health research?; (ii) What role (if any) do professionals concerned with public health feel researchers ought to play in advocacy?; and (iii) For those researchers who do engage in advocacy, what are the risks and challenges and to what extent can these be managed/ mitigated? In answering these questions, we argue that two deeply contrasting conceptualisations of 'advocacy' exist within public health, the most dominant of which, 'representational', centres on strategies for 'selling' public health goals to decision-makers and the wider public. This way of thinking about advocacy closely resembles 'lobbying'. While valuable, particularly for public health concerns that are currently under-represented in third sector advocacy work (e.g. health inequalities), we argue that this approach to advocacy can be rightly criticised as technocratic and elitist. An alternative, 'facilitational' conceptualisation of advocacy involves taking a more democratic approach to advocacy that centres on listening to, and working with, communities and members of the public whose voices are under-represented in research and policy debates.

Methods
This article is based on (i) a literature review; and (ii) interviews and focus group discussions with individuals involved in public health research, policy and advocacy. For the literature review, we identified existing academic publications concerning public health advocacy by conducting systematic searches of relevant academic databases. Our search terms included 'public health', 'policy', 'advocacy' and 'research' or 'evidence' (for precise search strings and databases, see Web Appendix 1). In assessing relevance, we focused on publications that either defined public health advocacy or commented on the role of research or researchers within it. We did not employ any date or methodological restrictions and included essays and opinion pieces as well as empirical research. However, due to resource limitations, we were only able to include publications written in English. ES first conducted this search in May 2013, and then updated it in February 2014 (as we did not include a historical cut off, the searches were intended to capture all available literature in the included databases published up to and including February 2014). Those publications considered to be relevant on the basis of their title and abstract were downloaded to a Zotero library. Once this had been completed for all five databases, all duplicates were removed. The searches produced 129 relevant publications, which were reviewed in full. In this article we draw on the 28 publications that we felt provided the most insights into the questions posed in the Introduction, above.
In addition, 69 individuals involved in public health research, policy and advocacy in the UK were interviewed between 2011 and 2013 and 90 individuals participated in focus groups, 12 of whom were also interviewees (see Table 1). In total, this article is therefore drawing on the perspectives of 147 individuals. Table 1 provides an overview of the professional affiliation of these individuals. All of the interviews were semi-structured and conducted by KS. The majority took place in a private room where, for the duration of the interview, only the interviewee and the researcher were present (one interview was a joint interview involving two interviewees and two interviews were conducted by telephone, at the request of interviewees). A themed interview schedule was employed which focused questions around public health research, policy, advocacy and knowledge exchange. The interviews varied in length, lasting between 45 and 150 min (most were around 60e80 min).
Potential interviewees were selected on the basis of four criteria: (1) their particular public health concern (most had some interest in health inequalities but, beyond this, we tried to identify individuals concerned with a range of issues, including tobacco, alcohol and obesity); (2) their professional role (we tried to identify individuals working on public health in academic, public sector, third/community sector and private sector settings); (3) their role in research and policy (we tried to include individuals who were primarily undertaking research, individuals primarily undertaking policy work and range of intermediaries, including individuals in knowledge exchange and media roles); and (4) their perspective on appropriate policy responses to health inequalities (here we tried to include individuals who were known to favour more upstream, radical policy responses and those favouring more meso-and micro-level responses, though we found it harder to identify the latter since there does now appear to be a fairly strong consensus that upstream responses are required, at least among those with a specific interest in reducing health inequalities).
Fifteen focus groups were also conducted, all of which lasted around an hour (the topic guide for all focus groups was designed by KS; facilitation was undertaken by a combination of the lead author and colleagues (see acknowledgements), all of whom met collectively to discuss the aims and approach in advance). One was undertaken at a People's Health Assembly in Nottingham in 2012 and this focused explicitly on discussing 'public health advocacy'. In this focus group, which involved 15 participants, all of whom identified themselves as public health advocates, participants were asked to consider how to define public health advocacy, who they believed did (and should) act as public health advocates and what the relationship between research and advocacy ought to look like. The other 14 focus groups were undertaken during a two-day symposium held in Scotland in December 2012 at which participants (researchers, policymakers, civil society campaigners, public health practitioners and research funders) were asked to explore potential future directions for health inequalities research. The 14 focus groups were undertaken in two sets (seven focus groups in the morning and seven in the afternoon) with 76 attendants participating in at least one (most participated in two). Although these 14 focus groups did not explicitly focus on 'public health advocacy', both the potential importance of public health advocacy and the risks facing researchers who advocate were strong features of these discussions.
All focus group discussions took place under the 'Chatham House rule' enabling participants to share the content of discussions with others only on a non-attributable basis. All participants (for focus groups and interviews) were asked to sign written consent forms enabling discussions to be digitally recorded and transcribed, before being analysed (all interviewees were also given an opportunity to check their transcripts for accuracy and anonymity). All anonymised transcripts were coded by KS in NVivo 10 using a thematic coding framework that she developed iteratively, via analysis and re-analysis of the transcripts. The key thematic codes ('parent Nodes' in NVivo) we developed were: academic freedom; academic/research work; achieving policy/social change; advocacy, knowledge exchange and impact; changes in academia; coalitions and fractures in public health debates; credibility (of individuals, research and ideas); idea types; morality and ethics; policy actors & sectors; policy processes; politics; position on health inequalities (causes and solutions); professional careers. The emerging themes and findings were presented to audiences at conferences and workshops that included some of the participants, enabling us to test our interpretations. Participants' feedback largely supported our interpretations but repeatedly drew particular attention to the tension between 'faciltiational' and 'representational' forms of advocacy in public health (without using those terms), leading us to place more emphasis on this aspect of our analysis in this paper. All research was approved by the University of Edinburgh's School of Social & Political Science Ethics Committee in March 2011.

Findings
Here, we draw on the multiple data sources to explain the key findings. The concluding discussion then presents a reflexive analysis of these findings.

Varying definitions of public health advocacy
Most of the literature on public health advocacy is, as one of its most enthusiastic promoters notes (Chapman, 2007), lacking in empirical research; it is, instead, disproportionately composed of editorials and comment pieces. This may explain both the paucity of clear definitions and the lack of consensus where definitions do exist. Only two of the publications we identified which report empirical research on public health advocacy adopt clear definitions (Asbridge, 2004;Carr-Gregg, 1993), while interview and focus group participants similarly provided a range of different accounts. Table 2 provides an overview of common ways of defining advocacy across the literature and qualitative data. The first column of Table 2 categorises definitions by the emphasis: processes, goals (agreeing future-orientated types of change to achieve and working to achieve these) and outcomes (a more retrospective, evaluative means of defining public health advocacy as work that helped achieve actual, health-positive change). Beyond this, three key variances in the definitions emerge. First, while the practical, outcomes-focused conceptualisation of public health advocacy evident in some sources (e.g. Carr-Gregg, 1993;Christoffel, 2000;Rock et al., 2011) implies it is a form of work that can be monitored, perhaps even evaluated, Chapman and Wakefield specifically argue that advocacy differs from traditional public health research work in that it cannot be treated as an intervention and assessed for efficacy (Chapman and Wakefield, 2001). Second, while some definitions focus on achieving upstream systemic or policy changes (Chapman, 2007;Christoffel, 2000;Freudenberg, 2005), others (particularly in North America), focus on altering 'community level conditions' including via health promotion (Avery and Bashir, 2003;Bassett, 2003), while others (e.g. Adshead and Thorpe, 2009 writing from within the UK Department of Health) emphasise achieving individual behavioural change.
Thirdly, as outlined above, there are two discernibly distinct understandings of what public health advocacy entails. In Chapman's (2007) terms, public health advocacy includes (amongst other things) working to place and maintain issues on public and political agendas (and exploiting opportunities to do so), discrediting opponents of public health objectives and working to frame evidence in persuasive ways (e.g. via metaphors or analogies). In other words, 'advocacy' involves strategically 'selling' pre-defined public health objectives to a range of non-academic audiences. Carlisle describes this way of thinking about advocacy as 'representational' (2000) and contrasts it with an alternative, more community orientated 'facilitational' definition of advocacy (e.g. Altman et al., 1994;Avery and Bashir, 2003;Bassett, 2003) which is also evident in Table 2. This way of thinking about advocacy is more akin to Burawoy's (2005) notion of 'public sociology', in which researchers engage in dialogue with members of the public and work collaboratively with organisations representing public interests; it involves working with relevant communities to ensure that voices which might traditionally be ignored are given due regard.
The difficulties of defining the boundaries of public health advocacy are, then, a distinct feature of the literature. To some extent advocacy (as presented by its proponents) seems to be a state of mind, or orientation, as much as a set of activities. There are, nonetheless, some areas of agreement. It seems clear, for a start, that 'advocacy' involves something more than the widely accepted model of researchers working in isolation or even in conjunction with senior policymakers to try to develop evidence-informed policy responses to public health problems (Petticrew et al., 2004). Moreover, when it comes to thinking about the relationship between public health research and advocacy and, specifically, academic advocacy, some broadly consistent types of activities are described, captured in Fig. 1, below. The centre of Fig. 1 depicts ongoing parts of life in the academy that go only a stage further than the 'traditional scientist archetype' described by interviewees in Haynes et al. (2011). Such activities include working within NGOs to wield greater legitimacy and resources (Caira et al., 2003;McGovern, 2007), actively seeking to influence policy development through strong relationships with policymakers (Caira et al., 2003;Haynes et al., 2011) and offering oneself as an 'expert' for commissions or advisory groups (Loue, 2004;Weed, 1994). While these kinds of activities reflect some of the definitions of advocacy discussed above, they are also precisely the activities that proponents of evidence-based policy tend to encourage (e.g. Hunter, 2009;Lavis et al., 2009). The other three groupings in Fig. 1 e planning, conducting and disseminating research -relate to how advocacy might be located across specific projects. With the exception of participatory research, most activities relate to the planning and dissemination stages. Those listed under dissemination are relatively uncontroversial. Some, such as presenting research findings to policymakers and politicians at local, national or international levels (Okonofua et al., 2014;Thornton et al., 2007), sit comfortably with the 'traditional Table 2 Quotes illustrating different emphases in definitions of public health advocacy.
Definition and example from literature review Illustrative quotes from qualitative data Emphasis: Process Using evidence to push for changes in policy and practice (Asbridge, 2004) Researcher (interviewee): "[Blank e anti-poverty campaigning group] is a classic group for using research and they're all academics on the steering group so it's not surprising that you would find that [this organisation] is a very research savvy organisation […] [which also has] really powerful links […] into the policy world." Coalition building (Christoffel, 2000) Focus group participant: "It's this way of using a, it is like creating a movement," Working with the media (Chapman, 2004) Public health practitioner who has held research, policy and advocacy roles: "[Public health advocacy] has a very important policy role and [advocacy organisation] punches well above their weight, I think, because they can produce statements which do get listened to by the media." Helping those who are not heard to have voice (Carlisle, 2000) Focus group participant: "[It's] being a voice for the people who can't speak"

Emphasis: Goal
Countering corporate influence (Chapman, 2007;Freudenberg, 2005) Researcher: "I actually do think that one [of the biggest issues] is corporate policy influences and we [researchers and advocates] have got to get better at countering that" Selling health (Chapman, 2007;Walsh et al., 1993) Focus group participant: "you've got to be confident in the fact that you've got to sell health as this fantastic thing and you've got to sell it any way you can." Emphasis: Outcomes Achieving policy change (Carr-Gregg, 1993) Public health practitioner who has held research, policy and advocacy roles: "In the passage of the bill recently [advocacy organisation] was mentioned a lot […] which is great. That shows that [the advocacy organisation] is having some impact in terms of policy change." Achieving health-positive changes (Christoffel, 2000) Focus group participant: "To me, being a public health advocate -you are changing something within your community," scientist archetype' (Haynes et al., 2011). Others go beyond this, including the production of specific non-academic outputs such as factsheets or toolkits (Hicks et al., 2012) or using personal narratives and visual imagery to 'sell' research findings to media outlets (Chapman, 2007;Rock et al., 2011).
The activities described under 'planning research', such as surveys to demonstrate public support for a measure (Maina et al., 2012), or work to show a particular field is underfunded (and under-researched) (Behague and Storeng, 2008), attract more critical analysis, illustrating some of the tensions of advocacycentric research. To take two examples, Behague and Storeng (2008; see also Storeng and B ehague, 2014) describe researchers searching for a single 'testable' intervention for maternal health to attract attention and funding, with the consequence that wider systemic changes were ignored, while Cruz and Walt (2013) describe Gates Foundation funded researchers experiencing pressure to find evidence to support a particular malaria control intervention in order to aid its policy adoption. These concerns about research being shaped in particular directions in order to have 'real world' impact (and closely relatedly) to advance researchers careers, were also raised by our research participants.

Competing views about the appropriate role of researchers in public health advocacy
Whatever public health advocacy is, our interview and focus group data suggest public health researchers, policymakers and advocates often hold strong opinions about the appropriateness of academics' engagement in it. Focusing first on areas of consensus: everyone we interviewed appeared to believe that public health researchers ought to be contributing to efforts to achieve social and policy changes to the benefit of public health. Most interviewees also recognised that policies were shaped by the advocacy and lobbying activities of different actors and suggested that policies, and advocacy efforts to inform policies, could be strengthened through engagement with research: NGO staff member: "I suppose for me [effective public health advocacy] comes back to having the evidence -we need to be able to back up what we're trying to say." For some, these beliefs translated into an unequivocal view that researchers working in this area faced a fundamental (in some cases, moral) imperative to help achieve change: Focus group participant (advocate): "I honestly don't see the point of doing research unless we can try and implement the good points that come out of that or the lessons that's learned from it …" For others, however, the role of researchers (and specifically academics) in advocacy was more questionable, as the following sections of the paper explore. Some simply argued that professional advocates (e.g. in NGOs) were better equipped/skilled to undertake this kind of work: Academic (interviewee): "I don't know if it's the role of academics, in general, to be … using research for advocacy. […] In terms of social skills, the people who work in NGOs often understand more, however irritating meetings are, that actually being able to get along with people and make relationships with people is actually half the battle," There were also, however, obvious problems with the idea that academics and other researchers ought to leave advocacy to others, particularly when it came to complex and cross-cutting public health concerns for which prominent NGOs appeared lacking (e.g. very few of our participants could name any organisations actively campaigning to reduce health inequalities). This suggests that the idea that researchers can simply feed their research ideas into professional advocacy groups may not always be feasible and, in these cases, there may be more of an impetus for researchers to engage in advocacy more directly.

Understanding and mitigating the risks and challenges of researcher involvement in public health advocacy
For researchers working in public health who do take on advocacy roles, whether that is because they have a strong, evidence-informed commitment to a specific policy solution or because they are concerned about a general lack of awareness around a particular issue, it is clear that risks are involved. In this section, we consider what those risks are, first to academics and the research they undertake, and next to the public health issues of concern, before considering whether and how such risks might be managed.

The risks facing academic advocates
Across the literature and our qualitative data, four distinct concerns emerge. The first (and most common) was a broad consensus around the tension between policymakers' need for clearly defined proposals and the limitations of traditional academic research for developing such proposals: Public health advocate (focus group participant): "Often researchers focus on the problem and explaining the problem and evidencing the problem, which you need, absolutely, but less focused on solutions and less focused on what could be done about it […] and that's actually really difficult to take to politicians and policymakers, because as soon as you take them a problem, they want the solution." Similar comments were made in three different focus groups and, in each case, this prompted an intense discussion about the extent to which it is reasonable (or desirable) to expect academic researchers to move beyond describing research findings to the realm of making policy recommendations (concerns also evident in the interview data). Generally, while almost all of our participants seemed to accept that advocates need specific recommendations to work with, several participants (mostly, though not exclusively, academics) queried whether it was ethical or appropriate for academics to perform this role. The following extract represents one such example: The second, related concern, expressed by a smaller number of interviewees was whether academics engaging in advocacy do so for ideological, rather than empirical, reasons: Academic (interviewee): "I find advocacy a difficult issue because … the most passionate advocates tend to have a very particular view of the world and the way it is and … this may not fit nicely onto what the evidence says is going on." In these cases, participants' concerns were not simply that academic advocates might 'go beyond' the available research evidence but that the research outputs these individuals produced were so heavily shaped by their personal-political positions that this undermined the credibility of their research: Academic (focus group participant): "During [a collaborative research project on health inequalities that interviewee was involved in], they had two groups working with the same data who came out with the opposite conclusions. And surprise, surprise, [Professor X's] group came out saying Nordic welfare states had lower inequality." Carr-Gregg suggests this occurs because advocacy derives from "personal values and self-interest" more than "the findings from science" (Carr-Gregg, 1993: p.36S). In previous work we have argued that it is never possible to separate 'objective research' from researchers' normative, ideological and ethical positions (Smith, 2013), but advocating strongly on the basis of them involves reputational and intellectual risks. For example, Cruz and Walt (2013) argue that, once a researcher has developed and expressed a clear policy position, it can be difficult to take a step-back and consider other points of view, even as evidence evolves. Interestingly, Australian researchers found that when research advocates do not declare their moral positions, this could impact on the effectiveness of their advocacy (Carey and Crammond, 2015).
A third evident concern is that engaging in significant amounts of advocacy work may compromise a researcher's actual and/or perceived independence which, in turn, can reduce both their academic and their policy credibility (Cruz and Walt, 2013;Haynes et al., 2011;Rychetnik and Wise, 2004;Veerman et al., 2006). Rychetnik and Wise (2004, p.253) argue that academics are valued by policymakers precisely because they are 'perceived to be independent of the machinations that occur between competing policy stakeholders, and thus untainted by the politics of policy development. This means that, in working to try to increase the policy and political relevance of their work, academics may actually undermine a core aspect of their value in policy settings; a point several interviewees also made: Academic (interviewee): "I probably would want to strike a note of caution [about academic advocacy] in that … as researchers, there is probably a need to not only be independent but also be perceived to be very independent." The data also make clear that academics who adopt public media profiles risk criticisms and attacks from their peers. This was less apparent for academics involved in tobacco control but most academics working on health inequalities who were identified by themselves and/or other academics as 'advocates' (or as academics who engaged in advocacy work) were criticised on the basis either that their recommendations were not sufficiently well supported by research evidence or that they were egotistical and 'liked being on the telly'. Academics in this group were themselves very conscious of this (see also Pickett and Wilkinson, 2016) and tended to view it as a broader issue facing any academics engaging with the media or public: Academic (interviewee): "I think it's often seen as sort of lightweight, to go outside of academia, to engage with the public, that it's grandstanding or showing off or -I bet all historians don't like Simon Schama.
[…] So I think it's often been seen as a bit egotistical and not quite rigorous to go and do those kinds of things …" Rychetnik and Wise (2004) also note that when researchers adopt a clear policy position, it suggests that the available research is sufficient, limiting a researcher's ability to make a case for further research on this issue. Although none of our research participants made quite this point, several senior academics acknowledged that they benefited from a context in which research and policy answers were viewed as unclear: Academic (interviewee): 'In a way … it suits people like me to have this phase of going round and round in circles because those circles produce demand for research, which means people like me can get funded, so our kind of every-day coping strategy is just to think, 'oh well, we won't complain while we're ahead. We won't … look a gift horse in the mouth.' Steve Fuller (2005) makes a similar point, noting that such a situation not only benefits academics in search of research income but also politicians in search of reasons not to take action.
Finally, in more practical terms, several interviewees complained that the sort of proactive, meticulously planned campaigns undertaken by Chapman (2004) can be extremely time-consuming, which invariably reduces the time available for more traditional academic work (Haynes et al., 2011). A closely related concern is that the focus expended on 'translating' research to connect with community needs rather than academic discourse (Hicks et al., 2012) might make it increasingly difficult for researchers to 'switch mode' and make the theoretical contributions still expected (and rewarded) within the academy. For some, this was an issue because they felt under pressure (for institutional and/or career reasons) to focus on research outputs. For others, the issue was more fundamental. One senior academic, for example, suggested that academics whose work became ever-more policy focused, risked losing 'their links with the scientific stream of work, which actually ensures the quality of what they do for policy.'

The risks of academic advocacy for the public health issues of concern
Many definitions of public health advocacy suggest that a key aim is to attract public and/or policy attention to particular issues. In this context, it might be assumed that working to ensure research findings make their way into the public domain (through, for example, media coverage) can be a useful way of advocating to achieve policy influence (e.g. Veerman et al., 2006). Yet, in discussing an example of a media campaign regarding food instability in poor Canadian families, Rock et al. (2011) describe how unexpected 'off-message' reactions from members of the public meant that public awareness of their research quickly became unhelpful for policy change. Put bluntly, if research prompts a public backlash, policymakers may feel constrained (rather than enabled) by research-informed advocacy.
Although the specific example of media coverage of research attractive negative public debate was not identified as a concern by our own participants, there was quite a lot of debate about whether public health advocacy should focus on trying to 'educate' and 'influence' the public or represent their views, reflecting Carlisle's (2000) distinction between 'representational' and 'faciltiational' forms of advocacy introduced earlier: Knowledge broker (focus group participant): "We are very good at telling people what's good for them, rather than listening to what might be good for them from the bottom up, and if we're going to transform public health in Scotland I think that's the way to go." This related to a broader risk of public health advocacy, identified by two academic interviewees and an interviewee working for an NGO, which was a concern that advocacy by 'expert' researchers and large NGOs has the potential to be elitist and undemocratic. The NGO policy advisor made this point in interview by giving an example of attending a formal boardroom dinner with civil servants and academics who seemed oblivious at the irony of discussing inequality and participatory democracy in that setting.
All this poses a challenge to the idea that researchers and third sector public health organisations should have any more influence than they already do. Instead, the above three participants and the focus group participant below suggested, researchers ought to spend more time listening: Academic (focus group participant): "How good are we at actually knowing, is the research community at knowing and understanding communities and working with communities? […] We just don't -none of us do actually really listen." More practically, both the advice offered by experienced advocates (Chapman, 2004) and empirical research on improving evidence-based policy advice (Petticrew et al., 2004), suggest that viable advocacy campaigns focus on 'winnable issues'. To be effective, this may mean putting more research effort into issues which are 'feasible' or already on the policy agenda, and less on those that seem too challenging for the current policy context: Focus group participant (advocate/activist): "A lot of the arguments we were making [were] about, this is an equity issue and all the social determinants arguments. And when you present those in a public forum, often what seems to happen is a kind of drawing back, because people go, 'well, that's just too huge, you're basically asking us to change the whole structure of the economy […] so I was trying to think […] whether you don't need some kind of two pronged approach. One which is […] to come to policymakers with quite concrete suggestions of things that can be done in the short term …" Yet, trying to agree concrete suggestions may reveal (or deepen) divisions within a research community, with negative consequences for the broader causes: "The search for a single intervention was not only reductionistic, some argued, but contributed to infighting and the constant shifting of proposed vertical interventions … each vying for policy attention. Such dynamics resulted in the splintering of what could be a comprehensive community and facility-based health systems approach into specific targeted subcomponents, or, as one policy expert described, isolated "bits of the jigsaw puzzle." (Behague and Storeng, 2008, p.645; see also Cruz and Walt, 2013) Finally, if researchers lose, or damage, their own credibility by engaging in advocacy work (as described above), then any campaigns associated with those researchers may also be damaged (Veerman et al., 2006).

Can the risks involved in academic advocacy be managed?
So far, we have identified a range of risks and dilemmas which appear to recur for academics who engage in public health advocacy and the causes for which they might advocate. Here, we consider the two main strategies put forward for managing these risks.
The first is training; some risks may simply arise from poorlyexecuted advocacy when, for example, researchers do not have the right skill-sets for advocacy-orientated work (Rychetnik and Wise, 2004). While some participants felt being appropriately skilled at advocacy work was more a matter of personality, others agreed with Chapman's (2007) case for advocacy training: Academic (focus group participant): "I do think it's indefensible that advocacy isn't a core public health competence.
[…] Lots of the pressure for advocacy being integrated into the public health curriculum has come from people like Simon Chapman who've long argued that why is it that we're prepared to go on and make media appearance without any training, but will sit down and rehearse a conference paper that's going to be listened to by 20 people?" From this perspective, this is a risk with the potential to be managed, albeit requiring investment by universities, professional organisations and others involved in public health training.
The other key recommendations for managing the risks of academic advocacy involved working collaboratively: Focus group participant (advocate/activist): "[Public health researchers need to] make alliances and coalitions […] You see like young academics getting nervous of saying things, and you think, for goodness sake […], you're self-censoring yourself. And actually it's not that dangerous if you go together, it's dangerous on your own, you can get hit, but if you go together it's actually a really strong force. And I think this country needs it right now." However, questions regarding the appropriate make up of advocacy coalitions return us to the issue of democracy. Neither the literature reviewed nor our interview data provided any clear suggestions as to the appropriate role of different kinds of actors in advocacy coalitions (particularly regarding leadership/coalition formation) and the dearth of comments regarding the role of the public was particularly noticeable. In the context of concerns about declining public trust in experts (Saltelli and Giampietro, 2015), this is a crucial challenge for those arguing for more public health advocacy.

Concluding discussion
The appropriate relationship between public health research and advocacy, and the role of researchers within this, are longstanding tensions subject to intense contemporary debate. The findings presented here suggest that this debate is marked by confusion as to what advocacy is and more fundamental disagreements. On the one hand, if advocacy is understood in Chapman's (2007) terms, many of the activities involved overlap with those proposed in studies supporting evidence-informed policymaking (Lavis et al., 2009) and research impact. From this perspective, advocacy might be understood as representing a section of work on a continuum between 'ivory tower' academia and fully fledged (often self-identified) advocacy and activism, in which the role of research is secondary to the advocacy aims, as depicted in Fig. 2.
Although our findings suggest that there is an ongoing debate within public health as to the most appropriate (and desirable) location for researchers on this spectrum, we are clearly a long way from the era in which Bradford-Hill (1971) reflected that even basic dissemination might be perceived as 'propaganda'. Indeed, virtually all of our participants were somewhere to the right of Fig. 2 and there was a clear appetite for more discussion and guidance (especially for earlier career researchers) as to how best to manage competing tensions at this end of the spectrum. This included: a desire to better understand how to support advocacy work with research while avoiding the politicisation of research; an aspiration to better understand the skills and resources required for advocacy work; and some discussion as to how such demands might be balanced with more traditional research career demands. In other words, there is a desire to incorporate better understandings of advocacy into public health training.
On the other hand, our findings also identified examples of very different interpretation of advocacy, reflecting Carlisle's (2000) contrast between 'representational' and 'facilitational' advocacy. So, while most participants (and most of the literature) focused on representational advocacy, others made a strong case for more facilitational approaches, involving 'listening' and 'working with' communities. Perhaps the closest our participants came to overcoming this tension was in suggesting that advocacy requires coalitions of different types of actors, though even here the role of the public remained largely undiscussed. For the most part, in both the literature and qualitative data, where the need to 'engage' the public was mentioned it was depicted as a means of increasing support for pre-defined policy or social changes (based on research evidence) rather than a means of opening up the evidence itself for public debate (let alone dispute). This neglects both the widelyargued advantages of publicly-engaged research, such as mobilising 'lay' knowledges for more socially-appropriate and acceptable solutions (Cunningham-Burley, 2006), and the repeatedly identified failings of top-down scientific dissemination (Wynne, 2006). Facilitational advocacy does not, however, provide easy solutions: unless researchers attempt to work in collaboration with the whole population, they will inevitably be amplifying some voices at the expense of others (since it is usually only undertaken for small or specific groups).
The potential overlap between these different approaches is depicted at the centre of Fig. 3. The size of circles in Fig. 3 represents an approximation of the frequency with which this approach emerged in our data, while their relative position represents the degree of overlap between the different concepts. Importantly, none of our participants made suggestions that could be located at the point in which all four circles overlap, where we might hope strategies for overcoming the seemingly fundamental tension between facilitational and representational advocacy to exist. Indeed, there are few obvious ways to address this tension, which essentially concerns the tension between 'selling' evidence-informed health proposals and more democratic ideals. Efforts to date, such as public or 'lay' representation on expert-led decision-making committees, have been criticised for failing to: (i) engage a sufficiently wide range of people (Abelson et al., 2003); and (ii) acknowledge the validity of people's experiential expertise (Stewart, 2016).
If we are to progress beyond this tension, a more dialogic, less top-down conceptualisation of public health advocacy seems worth exploring. Many of the changes implied by public health research demand national-level (often state-led) action; public health advocacy needs to be effective. As such, we find the normative case for both representational and facilitational advocacy compelling (a preference which undoubtedly reflects our own ethical and political commitments as scholars). There is now a pressing need for more research to understand whether the empirical case for facilitational advocacy lives up to the normative one, and this might entail working to better understand links between 'representational' and 'facilitational' forms of public health advocacy, especially as it seems unreasonable to expect the same individuals to be skilled at both. This paper's efforts towards definitional clarity within the field are, we hope, a first step towards a stronger evidence base for these important practices. Chief Scientist's Office (PDF/13/11).