The politics of being an ‘expert’: A critical realist auto-ethnography

The work of ‘experts’ with policy advisory panels plays an important part in the making of illicit drug and other policies. This article explores what is involved in this work. It uses critical realist auto-ethnography of the author’s experience over five years of working with the UK’s Advisory Council on the Misuse of Drugs and the House of Commons Health and Social Care Committee. It analyses: how some people become recognised as a ‘suitable’ expert through relational networks of esteem, while others are excluded; how bureaucratic processes and scientific modes of discourse select some types of information rather than others for the creation of acceptable evidence; and how agenda-setting and self-censorship can reinforce the exclusion of other knowledges, further narrowing the range of people and ideas that shape evidence for policy


Introduction
The current coronavirus pandemic has brought public attention to the use of scientific evidence in informing policy (Stevens, 2020a).Such evidence is made rather than simply discovered (Rhodes & Lancaster, 2019).This production of evidence mobilises 'epistemic authority' to inform and bolster 'political authority' through the 'politics of expertise' (Benveniste, 1972;Geuss, 2001;Rich, 2004;Scheel & Ustek-Spilda, 2019).In this article, I focus on what it means to be involved in this processand so on the politics of being an 'expert' -by applying critical realist auto-ethnography to my own experiences of working in this role.
Panels and committees are prime venues for the work of experts in creating knowledge for policy (Jasanoff, 1990).One British example is the Advisory Council on the Misuse of Drugs (ACMD).This is a standing committee of experts which has a statutory mandate to advise ministers on 'measures (whether or not involving alteration of the law) which in the opinion of the Council ought to be taken for preventing the misuse of such drugs or dealing with social problems connected with their misuse' (Misuse of Drugs Act 1971, Section 1, italics added).
The ACMD reports to the Home Secretary.She is the government minister who leads the Home Office (the UK equivalent of other countries' ministry of the interior), which has lead responsibility for UK drug policy.In the UK, the uses of drugs are controlled by their being placed in a class of the Misuse of Drugs Act 1971 (Class A, B or C), which determines the maximum penalties for offences related to each substance.These drugs are also placed in a schedule of the Misuse of Drugs Regulations 2001 (from 1 to 5).This determines whether they can be used for medical purposes, and the restrictions applied to their medical use.Schedule 1 substances (such as cannabis prior to 2018) are those deemed to have no medical purpose.The ACMD advises the Home Secretary on the appropriate class and schedule for each substance, although the final decision rests with her and, ultimately, Parliament.
Increasingly, drug policy advice has also been sought from time-limited groups commissioned to report on specific issues (Black, 2016(Black, , 2020;;Lancaster, Duke, & Ritter, 2015;Monaghan, Wincup, & Wicker et al., 2018;The New Psychoactive Substances Expert Panel, 2014;Recovery Oriented Drug Treatment Expert Group, 2013;).Parliamentary bodies, such as select committees and other crossparty groups, also co-opt external advisers to assist their deliberations.A recent example is the Health and Social Care Committee's (2019) inquiry on drug policy.
The explicit role of such panels, in drug policy and other fields, is to 'organise existing knowledge' in order to inform policy decisions (Hokanson, Ellstrand, & Raybould, 2018).To do this, they rely on the work of 'experts', by which I mean people who have been recognised as having a specific and advanced level of knowledge of the field.For Bourdieu (1975), this is a recognition of both 'technical capacities and social power', which is always open to challenge.Following Bourdieu, I use the term 'field' in this article to denote a sphere of intellectual or other practices in which participants share some assumptions about the 'rules of the game' (Bourdieu & Wacquant, 1992).
Existing research on the use of evidence in policy is largely based on interviews with researchers and policymakers, and on review of policy documents (Décieux, 2020;Masood, Kothari, & Regan, 2020).Many studies refer to reports of expert panels, but relatively few focus on how such panels work.Exceptions are provided in studies of drug policymaking by Taylor (2016) and Monaghan et al. (2018).Taylor shows how discussions within the ACMD were crucial to the 're-medicalisation' of cannabis in the period 1972 to 1982.This was a prelude to later decisions on cannabis, including its reclassifications from Class B to C in 2004 and back to B in 2009.In 2018, cannabis-based medical products (CBMPs) were moved from Schedule 1 to 2. Each of these changes was informed -if not determined -by reports from the ACMD (2002,2005,2008,2018).The ACMD did recommend the 2018 rescheduling to Schedule 2, but in 2009 the government rejected its 2008 advice to keep cannabis in Class C. Monaghan et al. (2018) discuss the role of other panels in drug policy developments in the last ten years.Both Taylor (2016) and Monaghan et al. (2018) rely on documents as their data.This article, in contrast, adds to the emerging sub-discipline of policy ethnography (Blaustein, 2015;Shore & Wright, 1997;Stratford & Wals, 2020), using participant observation within the policy process.It is auto-ethnographic, in that I 'retroactively and selectively' give an account of my own experience of the field (Ellis, Adams, & Bochner, 2011).It is theoretically informed by the combined 'ontological realism' and 'epistemological relativism' of critical realism (Archer et al., 2017).The research question I address is: what does being an 'expert' for a policy advice panel involve?This is a contribution to the collective effort to 'transform' understanding of the 'role of professional advisory systems and individuals', which was recently called for by Oliver and Boaz (2019, p. 4), as well as to the broader attempt to understand the role of expert knowledge in the 'unstable' politics of science (Loader & Sparks, 2011).

Theory, method and ethics
Auto-ethnography is more often used in feminist standpoint research than in the study of policymaking (Ettorre, 2017).This often draws on interpretivist epistemology and constructivist ontology.I have not found a previous example of auto-ethnography being combined with critical realist theory.My analysis is ontologically realist because it presupposes the existence of real events that actually occur, and so are available for observation and analysis.It assumes that these events are caused by real processes which operate even when they are not being observed (Bhaskar, 1975;Sayer, 2000).It is epistemologically relativist in that it recognises that empirical observations of these actual events are mediated by the social and cultural practices in which the observer is entangled (Archer et al., 2017).The critical realism adopted here precludes the use of Foucauldian or other post-structuralist approaches with different ontological presuppositions (Foucault, 1979).It also differentiates this analysis from Jasanoff's (1990) and other more strongly constructionist ways to study policy advice work, such as the 'evidence-making interventions' approach of Rhodes and Lancaster (2019).This article, and many of those cited here, have nevertheless benefited from such insights on how power and knowledge intertwine.
I also draw on Bourdieu's 'reflexive sociology', of which three characteristics are self-reflection, selfawareness and accountability (Bourdieu & Wacquant, 1992).So I acknowledge that my own ethicopolitical preferences affected how I influenced, observed and interpreted these events.My preferences are broadly in line with left libertarian thinking (Edgley, Stickley, Wright, & Repper, 2012), based on a rational commitment to the ethics of freedom and reciprocity (Gewirth, 1978;Stevens, 2011a).Despite some differences in detail, Bourdieu's reflexive sociology is compatible with critical realism through a shared commitment to both realism and critique, as well as an insistence on the importance of positionality in influencing both social action and interpretation (Bourdieu & Wacquant, 1992;Decoteau, 2016).
In this article, I make myself accountable to the field and the data by presenting examples of the events that inspired these reflections.I do not assert that my experiences or understandings of the work of policy advisory panels is universal and stable across different times, places or panels.But nor do I accept that my analysis wholly constructs its objects.My approach therefore differs from those influenced by some interpretations of the 'ontological turn' of Law (2004) and Mol (2002), and especially from the idea that 'realities do not exist outside the time and place of specific practices that enact them' (Scheel & Ustek-Spilda, 2019, p. 668).There are real causal processes at play here, and they produce actual events and documents that are available for observation and analysis (Elder-Vass, 2012).
I have worked for several years with drug policy advisory panels and committees.I joined the ACMD in the spring of 2014 after successfully responding to an open call for applications.I was a member for five years, taking part in the general deliberations of the Council, and also leading -as working group co-chair and lead author -the ACMD's reports on Reducing opioid-related deaths in the UK (ACMD, 2016b) and Custody-community transitions (ACMD, 2019b).I also acted as co-opted special adviser to the House of Commons Health and Social Care Committee for its 2019 report on drug policy.My university was paid for my work with this select committee.The Home Office does not pay ACMD members or their employers for their time, although it does reimburse incurred expenses (e.g. for travel to meetings).
I have also appeared as an expert before the Europe committee of the House of Lords, the Home Affairs Committee and the Scottish Affairs Committee of the House of Commons, and the Drugs, Alcohol & Justice Cross-Party Parliamentary Group, as well as the European Union's Horizontal Working Party on Drugs and the Pompidou Group of the Council of Europe.
I resigned from the ACMD in September 2019, due to my concerns over political vetting and exclusion of well-qualified applicants (as descried below).My involvement with these committees gave me experience of two modes of academic advice to the policy process: direct participation as a member of a scientific advisory committee; and acting as a scientific adviser to a panel of politicians.My analysis is also influenced by my previous experience of working with civil servants inside the policy process (Stevens, 2011c).
During the discussions of these panels, I took notes and received minutes.In reading and re-reading through these documents, I decided that the themes and reflections discussed in this article are those which are most pertinent to my research question.I do not pretend to have produced a replicable analysis.Others may interpret these example and events in different ways, or use my analysis to inform more reproducible, hypothetico-deductive research on the making of evidence and policy.
This study presents unusual ethical issues, as it was not planned in advance and is autoethnographic.I only decided that there was something worth analysing after the process of providing policy advice that I describe here.So I did not seek ethical approval in advance of these experiences.However, I do abide by the Statement of ethics of the British Society of Criminology (2015) in writing this article.
The main research subject of this article is myself, so informed consent is less of an issue than it would be if I were reporting primarily on the actions of others.The British Society of Criminology's Statement of ethics does allow for covert observation (without informed consent of participants) where seeking informed consent would create reactivity, or is otherwise not feasible to achieve, as long as this is justified and independently reviewed.Here, the seeking of retrospective informed consent from participants would effectively hand them the power to decide whether or not this research should be published.Given that one of the main issues raised by this article is the way that power operates in excluding some forms of knowledge from public discourse, it would be rather odd to enable people in positions of power to exclude this research from publication.
Apart from my own experiences and reflections, the source material -in the form of committee reports and minutes -is either already in the public domain or open to Freedom of Information requests.I have not provided any information here that would identify individuals to anyone but the person themselves, or others who were present at the incident described.So there is very little prospect for breaching confidentiality or harming people who were present at the incidents I discuss in this article.I have made it difficult for readers to decipher who is involved, in both my anonymised notes and the anonymised accounts in this article.The research involves no other intervention than my acting as normal in a policy advisory role.
Readers may worry that my writing of this article is motivated by resentment around the issues that led me to resign from the ACMD (as described below).They may also suspect, with some justification, that I was becoming increasingly frustrated by the government's failure to implement ACMD recommendations to which I had contributed (Stevens, 2019).There is perhaps another article to be written about the 'emotional labour' (Grandey, 2014) of providing advice in areas where decisions can make the difference between life and death.I certainly experienced such frustrations, but I expected them -based on my previous understanding that ACMD recommendations are often not implemented (Stevens & Measham, 2014).Prior to the incidents of political vetting described below, my frustrations had been with government ministers, not with the workings of the ACMD itself.I would like to stress here that I hold the members of the Council in high regard as experts in their field who work diligently in their efforts to improve drug policy.The possibility of unconscious bias on my part remains, which is why I have tried to provide evidence in the accounts below which readers can use to judge whether my interpretations are justified.

Being a 'suitable' expert
'Evidence use in policy and practice is significantly shaped by who is included, the conversations they have, how they are connected (or not), and the dynamics of their relationships' (Oliver & Faul, 2018, p. 369).So who gets to be an 'expert' for a policy panel, and how do these people relate to each other?One way to answer these questions is to report on how I got to be recognised as an expert by processes of selection for policy panels, and on how some other people did not.

Recruitment
When applying for my role on the ACMD, I had to present an application including a CV.I made sure to include a long list of peer-reviewed publications, as well as my professorial title.I expected that those selecting candidates would be looking for markers of scientific esteem, which I used in 'performing trustworthiness' through 'credibility work' (Baumberg Geiger, in press).As Baumberg Geiger reports, social researchers know that we cannot take it for granted that our expertise will be trusted as legitimate.We therefore engage in various practices which may help to differentiate us from pure advocates and so gain trust.In my interview, for example, it may have helped that I decided to wear my most sober suit and tie.I was also able to mention some other aspects of public service (including several years of volunteering on a lifeboat crew) in my application.This was discussed in the interview, as a member of the selection panel had a family interest in safety at sea.My overall impression was that the selection process found people who could be trusted to add to the scientific breadth and credibility of the Council, while not 'frightening the horses' by providing views too far outside those constituting the status quo.
I should also note that I share a white, male, upper/middle-class, middle-aged socio-demographic status with many of those who were already on the ACMD.These familiar characteristics may have facilitated the trust required for my selection.At one meeting, a senior member remarked on how diverse the Council's membership was.Looking round the mostly male members, with no women of colour, nobody of visibly African heritage, few audibly working class or regional accents, and no one who could be described as young without flattery, I was inclined to disagree.I doubt this lack of diversity was the product of anyone's conscious intention.It may reflect processes of recruitment and promotion within academia, which are themselves influenced by structural inequalities of class, race and gender (Bhopal, 2019;Pietri, Johnson, Ozgumus, & Young, 2018).This reinforces the lingering assumption that acknowledged expertise belongs to the stereotypical 'man of science' (Lancaster, Treloar, & Ritter, 2017, p. 284).
Recruitment to work with the Health and Social Care Committee was rather less formal, with no openly advertised application process.Rather, I was invited to a meeting of the committee to discuss its plans for an inquiry, along with a few other academics and professionals in the field.I knew most of them from previous research conferences and advisory events.During that meeting, I expressed my genuine admiration for a recent report by the committee (Health and Social Care Committee, 2018) and talked about my work with the ACMD (2016b) on reducing opioid-related deaths, which was a major concern at the time.During the meeting, other invitees mentioned this work, and also referred to me when answering questions about the decriminalisation of drugs (a subject I have worked on for several years (Hughes & Stevens, 2010, 2012;Stevens, 2011b)).This display of me as someone already recognised as an expert, and also my reassuring the committee that I was supportive of their previous work, may have helped the committee to decide to invite me to become one of two special advisers for its inquiry.

Exclusion from becoming an expert
Selection processes exclude as well as include people from being an officially recognised expert.This is usually invisible to the public, and even to others in the field.Occasionally, information slips out.Recent examples include the deliberate exclusion of Niamh Eastwood (a lawyer who is the executive director of the charity Release) and Graham Parsons (chief pharmacist at the charity Turning Point) from the ACMD.We only know about this because they both made public the results of 'subject access requests' under data protection law.Both applied to join the ACMD and were considered appointable by the selection panel.They were subsequently denied appointment after vetting of their Twitter accounts showed occasions where they had disagreed with or criticised ministers.The subject access requests revealed that the Home Office was also collating information about applicants' expressed views on topics unrelated to drug policy, including Brexit and the Windrush scandal of 2018, in which British citizens who were born in the Caribbean were unlawfully deported by the Home Office.
When I first heard about this political vetting, I asked for assurances that it would not continue to be used to exclude potential ACMD members.I argued that it contravened the 'working protocol' under which the ACMD operates (Home Office, 2011).This is supposed to guarantee the independence of the Council.Such assurances were not provided, so I resigned.In media reporting of these events, a 'government spokesperson' was quoted as stating: It is important that candidates who are considered for these roles undergo appropriate checks to ensure they are suitable to hold these vital public positions.(Busby, 2019, italics added) Eastwood and Parsons were apparently not considered 'suitable'.This is not because they were not adequately qualified as experts (this had previously been accepted by the selection panel), but because their opinions might clash with those of ministers, or cause them political embarrassment.
Also absent from these advisory panels are people with their own lived experience of problematic substance use.The working protocol includes a list of professionals that the government expects to include in the ACMD (including chemists, pharmacists, pharmacologists, social scientists, as well as specialists in law, enforcement, education and treatment).It does not specify that there should be any 'experts-by-experience' (Monaghan et al., 2018), who have themselves experienced harms or pleasures from drugs.When asked, these people tend to have diverse views on drug policy, usually different from government policy, with detailed knowledge of potential advantages and drawbacks of specific policy options (Askew & Bone, 2019;Greer & Ritter, 2019, 2020;Leonard & Windle, 2020).
When people who have avowedly used illicit drugs are invited to contribute to the discussions of policy panels, they tend to come with particular experiences and aims.That is, they tend to be people who have put drug use behind them, who now see it as essentially problematic, and who support a greater focus on abstinence in drug policy and treatment.People who see drug use as pleasurable or beneficial, or who wish to speak about the merits of their own opioid substitution therapy, were never heard from directly in the discussions of the policy panels with which I worked.
The 'legacies of deep distrust of drug users', which valentine et al. (2020, p. 14) observed in Australia, were also present in the discussions in which I participated.This is despite the opportunity that their involvement could provide, to inform policy with their intimate knowledge of drug policy effects, rather than relying on 'causal inference at a distance' (Stevens, 2020b) by technical experts with little personal experience of illicit drug use.
These processes of selection and exclusion suggest that recognised experts for policy panels tend to be members of existing networks of people with similar social backgrounds.Our paths tend to cross repeatedly, as we climb to the narrow peak of our specialised careers (I have often heard advice to academics that you have to develop a 'niche' if you want to progress).The circulations of advisers and policymakers within relatively closed 'policy constellations' (Stevens & Zampini, 2018) is conducive to an affective 'politics of familiarity' (Verón, 2019).Through these repeated contacts, we build relationships of trust, affection and mutual esteem.This can easily slip into an informal system of 'clientelism', in which favours are exchanged and secrets are kept that reinforce the inclusion of some types of people and of some forms of knowledge.During my time working with policy panels, I did not observe any activities I would consider corrupt.The network of influence is more subtle than that.I did observe the sharing of secret information and people doing each other favours that would both help the work of the panels and have positive effects on our careers.
Along with the exclusion of suitably qualified but potentially dissenting applicants on political grounds, these relational networks can narrow the range of the evidence that is created.Political vetting of candidates may deter some people from expressing contrary positions even before they join a panel, if they ever hope to be selected as an expert.

Selection of acceptable evidence
One of the most fascinating aspects of working on a policy advisory panel is seeing what gets said.This is structured both by formal institutional processes and by the implicit 'rules of the game' (Bourdieu & Wacquant, 1992).

Institutionalising evidence-making
Policy advisory panels are hierarchically structured and bureaucratically organised.They are led -at least nominally -by a chair person.In the case of many scientific advisory panels, including the ACMD, the chair is selected by a government minister.The chairs of parliamentary select committees are chosen by fellow MPs.The chair is supported in their work by a secretariat, staffed by civil servants.For the ACMD, these are employees of the Home Office, which also holds lead responsibility for government drug policy.This imposes a certain level of dependency by the chair on the ministry which they are advising.
The secretariat function for select committees is provided by parliamentary staff.This provides a greater degree of independence from the relevant arm of government, although committees of MPs are more subject to the party politics of their members.The parliamentary secretariat provides the select committee with recommendations on whom to invite as special advisers and oral witnesses.Members of the secretariat also draft the committee's report, in collaboration with the chair and members.I was invited, as a special adviser, to comment on drafts of the report.ACMD reports, on the other hand, are drafted by Council members, with support from civil servants and external specialists who may be co-opted onto working groups for specific reports.
In practice, the secretariats played an important role in shaping the work of both the panels with which I was most closely involved.When I chaired the ACMD working group on custody-community transitions, members of the secretariat wrote the agendas for the meetings.They provided very useful chair's briefing notes on what they expected to be agreed under each section of that agenda.Home Office civil servants also take the minutes of ACMD meetings, which provides an opportunity to shape how discussions and decisions are recorded.I have never had reason to challenge the accuracy of the minutes, but there were occasions when sensitive issues were glossed over.An example is my discussion with the minister which I report a few paragraphs below.

What counts as 'scientific' evidence?
Bureaucracies can influence the decisions of policy panels and so can the mode of discourse within them.Most of the members of the ACMD have a background in the natural and clinical sciences, which are often seen as being further up the 'hierarchy of the sciences' (Cole, 1983) than sociology or criminology.They imported into our discussions presuppositions which many social scientists would describe as positivist.Policy advice is often called on when the government faces a new uncertainty -for example, when a new substance or practice appears in illicit drug markets.The preferred mode of deliberation under uncertainty in the natural and clinical sciences is to set up an experiment or randomised trial, even though such methods may not be effective in identifying 'what works' in public and health policy (Deaton & Cartwright, 2018).This type of evidence was very rarely available in our discussions.It would have taken too long to create, even if the phenomena under consideration were amenable to experimentation, which they often were not.
A consistent recommendation of all the reports that I worked on was that there should be more investment in research on the topic.The only one of these recommendations that I saw being taken into practice was that on researching the clinical effects of CBMP (ACMD, 2018; National Institute for Health Research, 2020).Recommended research on the effects of drug policy on drug-related deaths was, for example, not funded (ACMD, 2016b;Stevens, 2019).Neither was the ACMD's recommendation that there be an evaluation of the Psychoactive Substances Act 2016 that was 'independent and adequately resourced' (ACMD, 2015a).There was a 'review' of the effects of the Act, but it was done within the Home Office and with no funding for primary data collection (Home Office, 2018).
The absence of sufficient evidence about the effects of drug policy is a frequent finding of policy reviews in this area (e.g.HM Government, 2017).Without such unambiguous information, discussions within the ACMD tended to focus on laboratory-based analyses of the molecular structures of potentially psychoactive substances, the neurotransmitter receptors they were likely to bind to, and their effects in animal models.In my very first ACMD meeting, I observed such a discussion relating to previously unclassified substances in the tryptamine group.The advice that the chair subsequently sent to the minister was: These are highly potent hallucinogens which act on the 5HT2A receptor, in the same way as LSD.The ACMD therefore recommends that the tryptamines covered by the proposed expanded generic definition in this report, are controlled under the Misuse of Drugs Act (1971) as Class A substances. (ACMD, 2014, p. 1) Ministers followed this advice, as they usually do when the ACMD recommends tighter controls (Stevens & Measham, 2014).In the meeting, I had queried why the recommendation would be to place these substances in Class A. I mentioned that a previous research project, involving several members of the ACMD and led by a previous chair, had found that the harms of LSD did not justify such a high classification (Nutt, King, & Phillips, 2010).But the recommendation was made by pharmacological analogy to an existing classified substance (i.e.LSD) -a case of 'guilt by molecular association' (Stevens & Measham, 2014).
At a later 'away day' of the Council, we had a more general discussion in small groups about the process of classification.I argued with two natural scientists that we should consider the effects of classification when making our recommendations.They argued back that it was not up to us to consider what the policy decision should be, only to provide advice on the substance's harms.Their argument is consistent with a positivist, value-free vision of what science can do and with the separation that some in the field would like to see maintained between scientific advice and policy decisions (Humphreys & Piot, 2012).However, it runs contrary to the fact that decisions were often taken to recommend classification in the absence of evidence of actual harm, as with the novel tryptamines referred to above.It also disregards the statutory mandate of the ACMD, which is to give ministers advice on measures which ought to be taken 'in the opinion of the Council' (Misuse of Drugs Act, 1971, Section 1).
Such use of 'opinion' is anathema to many scientists, as it opens the door to 'unscientific' use of values to inform recommendations.We had frequent discussions on how to minimise the influence of mere opinion on policy advice, culminating in the publication of a Standing Operating Procedure for using evidence in ACMD reports (ACMD, 2020).My contribution to these discussions was to bring in criminological frameworks for categorising social harms of drug use (e.g.Greenfield & Paoli, 2013) and to repeat what I had learned about the Bradford-Hill criteria for causal attribution from David McDonald and others (McDonald & Strang, 2016).These criteria allow for the use of nonexperimental evidence in the evaluation of potentially causal effects, and so can be used to include a wider range of research than the usual focus of 'evidence-based' research syntheses on systematic reviews of experiments and randomised trials.

The scientific 'argumentation game'
As I internalised the rules of the 'rhetorical argumentation game' (Greenhalgh & Russell, 2006) of scientific advice, I developed my ability to deploy academic authority in policy discussions.Several meetings provided me with opportunities to use my specialist knowledge about the lack of evidence there is that criminalisation of drug possession reduces drug-related harms (Stevens, Hughes, Hulme, & Cassidy, 2019).On one enjoyable occasion, I took the opportunity to ask a question of the minister responsible for a new crime reduction strategy.On the basis of existing criminological research, I knew this strategy used the same premises and measures as the previous -evidently failed -strategy.My question was why the minister expected the new strategy to produce different effects.It was answered with a passionate soliloquy on how seriously the government now took the problem, not with evidence on why the new strategy would work.
There were some differences in my performance of academic credibility, as enacted in the committee largely made up of academics and the one made up of politicians.With the ACMD, I used explicit citations to published research and conference papers to back the points I made.With the Health and Social Care Committee, I felt less need to do this.As I was the only academic in the room, it seemed more likely that my contributions would be taken as the accepted 'scientific' view.Performances of 'epistemic authority' (Geuss, 2001) can differ across different venues and audiences.
These initially satisfying displays of my 'niche' knowledge were not particularly effective in influencing policy.I managed to insert concerns about criminalisation and law enforcement into only two ACMD reports, one of which remains unpublished, and one letter to the Home Secretary.The letter (ACMD, 2015b) expressed the Council's reservations about the Psychoactive Substances Bill, which was passed into law in 2016.The published report (ACMD, 2016a) was our feedback on the draft version of the 2017 drug strategy, which did not deviate from the criminalising status quo ante.
I raised similar concerns in the ACMD's (2018) discussion of the appropriate schedule in which to put cannabis as a medicine.I argued that it should be in Schedule 4(ii) of the regulations in order to avoid criminalising people who use cannabis, and not Schedule 2, which retains criminal sanctions for possession (Stevens, 2018).I was in an extremely small minority.These discussions were hurried and done mainly by online indication of binary agreement or disagreement with a draft document.When I later had the chance to ask why Schedule 2 had been chosen for CBMPs, I was told that this could be reconsidered as time went on.But CBMPs are still in Schedule 2, alongside opiates and other drugs that are -unlike cannabis -deadly in overdose.
In contrast, my hope that the Health and Social Care Committee would recommend decriminalisation of drug possession was eventually fulfilled.Committee members were already interested in decriminalisation of drug possession in Portugal, of which I had some specialist knowledge (Hughes & Stevens, 2010, 2012).For their field visits to Lisbon and Frankfurt, I recommended some contacts who are knowledgeable and supportive of decriminalisation and harm reduction.I also circulated to the committee a redacted advance copy of some work I carried out for the Irish government on alternatives to criminalisation (Hughes, Stevens, Hulme, & Cassidy, 2019).Despite this, the draft of the committee's report did not directly recommend decriminalisation of drug possession.Rather, it used a more ambiguous phrasing for reducing the use of criminal penalties for possession.I pondered how to work on this.I concluded that it would be better if it were a MP on the committee who proposed a change to the wording.During the final committee meeting on the report, I mentioned that the 'technical term' for what they were proposing was 'decriminalisation'.One of the more liberal members of the committee then proposed that this term should be used in the report, and so it was (Health and Social Care Committee, 2019).

The evidence selection process
In these experiences in the field, I confirmed in practice what many analyses of evidence and policymaking have suggested in print.Information is more likely to become evidence by being accepted into the making of advice and policy if its content and purveyors fit with pre-existing bureaucratic and cognitive structures and interests (Ailsa et al., 2011;Cairney, 2016;Décieux, 2020;Kelly, 2018;Lancaster, Seear, Treloar, & Ritter, 2017;MacGregor, 2017;Masood et al., 2020;Monaghan, 2011;Monaghan et al., 2018;Nutley, Boaz, Davies, & Fraser, 2019;Oliver & de Vocht, 2015;Ritter & Bammer, 2010;Roberts, Petticrew, Liabo, & Macintyre, 2012;Smith & Joyce, 2012;Stevens, 2007Stevens, , 2011c)).This is not some random, natural or wholly unconscious process.While policy actors may be imperfectly aware of the presuppositions we bring to our work, we all presumably act in ways that are, like our theories, 'for someone, and for some purpose' (Cox 1981, p. 128, italics in original).The result of this collective process in policy advisory panels is the making of particular forms of evidence.

Exclusion and contestation of other knowledges
While some people and some forms of knowledge are selected, others are excluded.In addition to denying recognition to potentially dissenting individuals, other knowledges can be kept out of policy panel discussions by agenda-setting and self-censorship.

Agenda-setting
The setting of agendas includes deciding what gets researched and discussed.Parliamentary select committees make their own decisions on what topics to examine.The ACMD can also set up its own inquiries, as well as responding to commissions from the Home Secretary.I played a role in setting these agendas by pushing for the ACMD to report on opiate-related deaths and custody-community transitions.The resulting reports are a form of evidence, even if they have not been translated into policy (ACMD, 2016b(ACMD, , 2019b)).
Some forms of knowledge do not get made into evidence, as exemplified by the ACMD's (2019a) response to the request from the Home Secretary to: … set out how the ACMD will assess the various impacts of rescheduling cannabis-based products for medicinal use to Schedule 2 under the 2001 Regulations.
The potential impacts of this rescheduling are indeed 'various'.They include how it affects the estimated 1.4 million people who currently break the law while using cannabis for their own medical purposes (Couch, 2020).The effect on these people was not a primary consideration in ACMD discussions of the topic.The potential impacts included in the 'logic model' in the published assessment framework are but three: benefits to patients who are prescribed licensed CBMPs (which is currently in the tens, not thousands or millions); harms of diversion of these products; and the creation of a UK CBMP industry.The working group which produced this framework did not include anybody with avowed experience of using cannabis medically (or otherwise).Had it done so, and had the scientific discourse around CBMPs included the harms of criminalising cannabis possession, the assessment framework might have included different impacts.As it is, future iterations of the ACMD's perennial consideration of cannabis policy will not be able to draw on its assessment of the impact of scheduling on the criminalisation of people who use cannabis.Such an assessment will not exist.
There are some actors who can impose 'strain' (Chambliss, 1976) on the Home Office and the ACMD when recommendations go against their interests.They have some power to influence agendas.An example is the British pharmaceutical industry.In 2016, following the advice of the ACMD, all 'thirdgeneration' synthetic cannabinoid receptor agonists were controlled under Class B of the Misuse of Drugs Act 1971.This effectively and accidentally criminalised the work that pharmaceutical companies were doing on substances that fell under the ACMD's definition.The industry's representatives were able to secure meetings with members of the ACMD, which then created a new, more tightly focused definition (ACMD, 2017).Eventually, ministers passed this revised definition through parliament to enable continued commercialisation of these molecules (Home Office, 2019).People who have less financial resource and political access are less able to influence the agenda of policy advisory panels.Dissenting evidence can get through advisory panels into policy and law if it has powerful backers.

Self-censorship
I have already referred to the argument I had with natural scientists about the limits of what we, as the ACMD, could recommend.I suggest now that their reluctance to fulfil the Council's statutory remit by giving their informed opinion of how best to deal with a substance is a form of selfcensorship.I observed several other instances where ACMD members argued that we should not make a particular recommendation because it had no chance of being accepted by ministers.In such cases, I tended to revert to the imagined separation between matters of evidence and politics by arguing that it was not up to us to decide what the policy should be, but rather to provide evidenceinformed recommendations on what the most likely consequences of different options were.On more than one occasion, I was told that such advocacy for the evidence was a form of 'campaign', and 'the ACMD is not a campaigning organisation'.I got the impression that being seen to emphasise too strongly any particular research finding -and especially those with which ministers disagreedwould become a threat to the credibility of the Council, and of its individual members.
Taking this lesson on board, I engaged in some mild self-censorship of my own in drafting the ACMD's (2016b) report on drug-related deaths (DRDs).I was aware of research from other countries on medically supervised drug consumption rooms (DCRs).This suggested that such facilities are effective in reducing harms related to drug use (Potier, Laprévote, Dubois-Arber, Cottencin, & Rolland, 2014).But I also knew that to baldly recommend their establishment in the UK would be controversial, and that there were limitations to this evidence and its transferability.There are, for example, no randomised trials of DCRs, so they would not count as 'scientifically' justified by the standards of the type of systematic review that is usually used to test the strength of evidence for medical interventions (Pardo, Caulkins, & Kilmer, 2018).So, instead of including a direct recommendation that DCRs be set up in the UK, I inserted into the draft report the suggestion that: Consideration is given -by the governments of each UK country and by local commissioners of drug treatment services -to the potential to reduce DRDs and other harms through the provision of medically-supervised drug consumption clinics in localities with a high concentration of injecting drug use.(ACMD, 2016b, p. 40) My deliberate softening of the tone of this recommendation was successful in enabling it to pass through the full Council into the final version of the report.
There is an interesting parallel here with a recent report on the proceedings of scientific advice panels in the run-up to the UK's coronavirus epidemic in early 2020.When asked why his panel had not initially considered recommending the severe social distancing that was eventually implemented in late March, Professor John Edmunds reportedly replied that this was because 'no one thought it would be acceptable politically "to shut the country down … We didn't model it because it didn't seem to be on the agenda"' (Grey & MacAskill, 2020).This example of how 'policymaking tends to favour the politically feasible over the technically possible' (Monaghan et al., 2018, p. 436) shows that agenda-setting and self-censorship can reinforce each other in ensuring that some scientifically plausible policy options are not even considered.This is likely to compound the effect of excluding potentially dissenting voices in narrowing the range of evidence which enters the policy process.

Reflections on expert evidence-making
Is the evidence made by advisory panels only that which is provided by government-approved experts?Does it respond solely to the agendas that ministers and other powerful actors set?Does it only use a positivistic hierarchy of evidence, with systematic reviews and randomised trials at its apex?The answers, I think, are a little more complex.The ACMD's independence is still nominally protected by the 'working protocol', although this has been challenged by political vetting of applicants to join the Council.It retains the ability to set up its own inquiries, even though its ability to work on these is limited by lack of resources and the volume of work that is passed to it by the Home Secretary.The ACMD's new Standard Operating Procedures (2020), and the processes of evidence gathering which it and parliamentary select committees conduct, are open to non-experimental evidence.The system of policy advice is not completely closed to other people and other forms of knowledge.I did, however, observe processes which resulted in the giving of more credence and influence to some people and some types of evidence than others.
A reflexively sociological analysis also requires reflection on my own motivations and actions.So what of them?Was I acting as a neutral conduit of dispassionate advice from value-free scientific inquiry, based solely on specialist, technical knowledge?That possibility is ruled out both by the assumptions of critical realism and by the practices I engaged in and observed.Nobody in the field was completely disinterested, and there is no value-free science (Bhaskar, 1975).So was I a politically motivated 'campaigner', seeking only to promote pre-existing partisan positions?Again, I think the best account would be rather more complex than that.
My answers to these questions are inevitably shaped by my disciplinary training in sociology and criminology, as well as by my acknowledged ethico-political preferences.If my only intention in carrying out the work I have described had been to shift drug policy towards a left libertarian position, it would probably be judged a failure (at least in the short term).But, as a critical realist, I think it is consistent with my theoretical commitments that I tried to present the best, most normatively and empirically justifiable account that I could of the world as it is intelligible to us, both while providing policy advice and in creating this analysis of that work.In line with the 'judgemental rationality' of critical realism (Archer et al., 2017), my arguments for particular policy positions (as in my work on decriminalisation for the Health and Social Care Committee) were based on my commitments to valuing scientific standards for judging knowledge claims -including validity, reliability and authenticity -as well as respect for the human rights of people who use drugs.

Conclusion
I have presented a micro-level analysis of the making of evidence for policy in advisory panels.I observed that some types of people and some forms of evidence are selected over others.People who are outside existing relational networks of esteem are less likely to be recognised as an 'expert', especially if they are seen as potentially challenging to current arrangements.There are explicit and implicit rules by which recognised experts include particular forms of evidence in policy advice.Agenda-setting and self-censorship serve to exclude some ways of thinking, although this can be challenged within policy panels and overcome by powerful external actors.Together, these social interactions serve to narrow the range of evidence that enters the policy process through such panels.This matters because a less varied 'diet' of evidence is less likely to lead to policy innovation (Oliver & Faul, 2018).So these interpersonal processes contribute to the reproduction of structural inequalities, as would be anticipated by readers of Bourdieu's sociology or of Archer's critical realism (Archer, 2000;Decoteau, 2016).
The valuable role that advisory panels play in organising knowledge for use in policy is not separate from the field of politics, but part of it.If criminologists and other academics are to operate effectively as 'democratic under-labourers' (Loader & Sparks, 2011), we can -and should -use mechanisms that reduce the introduction of arbitrary bias into the knowledge that informs policymaking.These can include standardised, transparent, open systems for evidence collation, reporting and assessment.Creating and adhering to protocols of independence could protect the academic and political freedom of selected and prospective advisers.
I argue, on the basis of these observations and a critical realist perspective (Stevens, 2020b), that we should do more to open up the process of evidence-making to a wider range of research methods (including observational and qualitative research) and participants.This should include potential dissenters and people with a range of experiences of drug use.Even when experts-by-experience are directly involved, policymaking can still be 'animated by a strong privileging of particular types of data ' (valentine et al., 2020, p. 13).We should not expect the organisation of knowledge to operate independently of asymmetric power relations.The idea of there being 'nothing about us without us' is promising but problematic (Ahmed, Windle, & Lynch, forthcoming), with risks of tokenism, cooption, and ongoing marginalisation of some groups.But including a more diverse range of voices in advisory panels would not only enable more balanced 'negotiation' of policy advice (Jasanoff, 1990), it would also enable policy advice to be based on more accurate, detailed representations of actual events and real causal processes in drug and other policy areas.
I hope this article makes visible some of the affective, ethical, professional, scientific and strategic challenges involved in working as an 'expert' with policy advice panels.Above all, I would like readers to understand that the process of creating evidence in these panels is itself deeply normative and political.

Implications
All readers of these auto-ethnographic reflections may wish to bear them in mind the next time they read a report from an expert panel, or hear a minister boast of having consulted the experts on a specific policy problem in criminology or other fields.This article prompts consideration of the processes of selection, exclusion, agenda-setting and self-censorship that have shaped the evidence given to ministers.Qualitative criminologists may wish to consider the potential for going beyond interviews with policymakers and documentary analysis to expand the role of ethnographic methods in understanding how evidence and policy are made.Quantitative criminologists may wish to design studies that track the process of selection of people and evidence from funding to publication, dissemination and citation in policy advice reports, to test the hypothesis that there is systematic bias in the types of people, research methods and findings that get into the policy process.Policymakers seeking to improve the advice they receive may wish to strengthen their efforts to ensure that they receive a more diverse range of advice.This has the potential to improve resulting policies, on illicit drugs and many other issues.