The ethics of researching ‘terrorism’ and political violence: a sociological approach

ABSTRACT In this article, we propose a sociological model for the assessment of ethics in research on conflict and terrorism. We move beyond the rather narrow, procedural approaches that currently dominate contemporary discussion, seeking to broaden ethical considerations to include questions of social power, academic freedom, and the politics of knowledge production, as well as a consideration of the public function of the university. We argue that social scientists have both a professional responsibility to protect the integrity of scientific knowledge, and public responsibilities to the wider societies of which they are part. Navigating ethical questions, we suggest, therefore requires a reflexive engagement with the social conditions of knowledge production; a careful consideration of the social impact of research; and a dialogue with a variety of ‘publics’, not merely policy actors. The main body of the paper reviews the range of writing on the ethics of ‘terrorism studies’, engages with the question of institutional oversight and then examines the ethics of the current ‘impact agenda’ in UK universities. We conclude by drawing on our empirical findings and applying them to our proposed model to argue for: a significant revision to ethical policies and guidelines (and better means of enforcement) so as to better protect vulnerable research subjects; offer greater protections to researchers from (especially) powerful interests which attempt to smear, constrain or undermine independent research; make unethical research (which we argue is widespread) more visible, with the intent that it be managed down.

The current ethical discourse, largely inherited from the medical model, is mostly focused on protecting vulnerable research subjects from potential harm arising directly from participation in research (Hammersley & Traianou, 2012). Although these issues should be taken very seriously in the context of researching 'terrorism', there are also distinct and broader ethical questions when it comes to potential harms to research subjects and indeed to other actors, including researchers themselves. As we have argued elsewhere in this issue, harm to research subjects may arise not only from the research process itself, but more broadly and more significantly from powerful actorsoften parties to the conflicts under investigationseeking to access, make use of, or influence/manipulate research findings or how research is received, understood and used or not. The former brings particular challenges when it comes to confidentiality and security of data, whilst the latter requires that the societal impact of research be carefully considered. In either case, an understanding of power, and the power of the state in particular, is crucial.
Though they have not received much attention from 'orthodox' terrorism studies scholars, that is those 'scholars work in collaborative relationships with government and counterterrorist agencies' (Breen-Smyth, 2009, p. 209), (see also Jackson, 2007;Jackson, Breen Smyth, & Gunning, 2009;Miller & Mills, 2009), these sorts of concerns have been highlighted by a number of controversial cases of covert and overt involvement in academic research by state security apparatuses, as noted by Horgan (2012Horgan ( , 2013, and discussed more extensively by 'heterodox' scholars of political violence, especially those associated with 'critical terrorism studies' (Breen-Smyth, 2009;Sluka, 2012). Our aim here is to synthesise some of these issues and concerns with those raised by 'orthodox' terrorism studies scholars, and to develop a more sociological approach to the politics and ethics of research in this area; an approach which considers not only the potential harm caused to research participants, as is presently the focus of research institutions, but also potential harm to researchers, as well as the broader societal impact of research.
Michael Burawoy's (2011) 'Public University' model, based upon his influential work on public sociology (Burawoy, , 2005a(Burawoy, , 2005bClawson et al., 2007), serves as a useful starting point for developing such an approach since it allows us to better situate academic research in relation to broader social forces and interests. Burawoy (2011) outlines four functions of the university, or four types of knowledge, each with distinct professional practices on the one hand, and audiences on the other. Professional knowledge in Burawoy's typology is that produced in research programmes and evaluated by other academics. It requires academic research design and methodological skills. Policy knowledge, meanwhile, is geared towards providing solutions to specific problems, and is produced at the service of a particular goal set by 'clients'. Both these types of knowledge are defined by Burawoy (2011) as 'instrumental', meaning they fall within a puzzle-or problem-solving approach, as opposed to being 'reflexive', meaning involving some dialogue about ends and an interrogation of the 'value premises of society' as well as academia. Burawoy (2011) includes within 'reflexive' knowledge both 'critical' and 'public' approaches. The former, like professional knowledge, is orientated towards fellow scholars, but through its reflexivity serves as the 'conscience' of academic work. The latter, like policy work, is orientated towards extra-academic audiences, but is part of a dialogue between social scientists and 'wider publics' about the direction of society. Burawoy's (2011) typology allows for a multifaceted conception of social scientific research that emphasises the great importance of both professional standards and academic freedom, whilst offering a conceptual framework able to socially situate academic practices in relation to wider interests in society, whether in terms of engaging with policy makers, or other powerful actors, or undertaking what would more readily be regarded as politically engaged academic work (embedded research within subaltern social movements, for example).
Within this framework, professional knowledge is recognised as central to the function of the public university, but is also situated in relation to wider interests in society. It is important to emphasise at this stage that each of the four types or knowledge are dependent to some extent on the others and, concomitantly, that issues affecting predominantly one approachsay the (conflicting) orientation towards power of 'policy' or 'public' research, or the commitment to 'professional' or 'critical' researchwill commonly affect other approaches. This point underlies our comments later in this article on the need for a common agenda to defend research viability and integrity.
In what follows, we use this framework to develop arguments about ethics in research on conflict and 'terrorism'. First, we examine how ethics is discussed in 'orthodox' terrorism research, which we assess to be largely undertaken to, in Burawoy's terms, develop policy knowledge. In the second section we examine some issues raised by the question of 'impact', and how this might be understood in relation to the contrasting approaches of policy and public orientated research. We examine two 'impact' cases; one in which ethical issues were raised, and one in which they were not. We conclude that the impact of academic research should be considered as part of the currently very circumscribed ethical considerations and procedures. Such assessments should be made through processes of critical reflection born from academic engagements with a wide range of 'publics'. Conceived in this way we are able to avoid some of the problems associated with the dominant conception of 'impact' that creates institutional incentives for the greater instrumentalisation of social science. Resisting such pressures in our view is important in order to both protect the integrity of social science as well enable the wider potential for social science to benefit society as a whole.

Ethics in (policy orientated) terrorism research
Despite the enormous amount of literature on 'terrorism', for some time relatively little was written on the ethics of terrorism research. The Routledge Handbook of Terrorism Research (Schmid, 2011) contains no discussion of ethics, and neither does Routledge's Terrorism Studies: A Reader (Horgan & Braddock, 2012). Silke's edited collection Research into Terrorism (2004) includes a few passing references to ethics from Horgan (2004), who notes the danger a researcher might face if obtaining information of 'use to the security forces' (2004, p. 34), but contains no substantive discussion.
Research ethics, however, is given much more attention in a more recent collection, Conducting Terrorism Field Research: A Guide (Dolnik, 2013). The focus there is on how researchers can (or whether they should) effectively conduct fieldwork whilst satisfying the standard ethical requirements demanded by ethics committees or institutional review boards (the US name for the body known elsewhere as ethics committees or similar). Overall, the concern is with how to circumvent ethical principles in research, or to redefine behaviour regarded as unethical as ethical.
In his chapter, Kenney (2013, pp. 35-36) considers research ethics mainly in terms of the privacy of potential interviewees, and Knights (2013, p. 120) notes the risk of the identity of participants being 'leaked'. Dolnik (2013, pp. 229-230), the editor of the collection, meanwhile, notes the tension between the need to protect participants' identities and the requirement for informed consent, noting that written consent runs the risk of participants being targeted. The question of consent is given particular attention in the collection, with several contributors arguing that the standard ethical requirement for informed, and especially written, consent is ill-suited to terrorism research. In addition to potentially exposing participants to harm, Taarnby (2013, p. 213), argues that 'consent becomes irrelevant in anarchic settings characterized by a high degree of illiteracy', whilst Rasmussen (2013, pp. 70-71) considers the requirement for written consent to be 'unworkable' given 'the opportunistic nature of terrorism research'. Like Dolnik (2013, p. 226), Rasmussen describes the idea of seeking written consent from 'former terrorists' as 'ludicrous' given 'the worldwide opposition to many of the US policies in the war on terror and the surge in anti-Americanism these have generated' (Rasmussen, 2013, p. 70). This political context, and the challenges it presents for satisfying standard ethical requirements, particularly informed consent, is more directly addressed in that collection by Horgan (2013, p. 197), who considers how a requirement for consent can be satisfied in circumstances in which research participants will likely be hostile to the research sponsors.
This dilemma is the focus of a RAND working paper cited by Dolnick entitled Ethical Principles in Social-Behavioral Research on Terrorism (Bikson, Bluthenthal, Eden, & Gunn, 2007). That paper documents the proceedings of a daylong workshop hosted by the RAND Corporation and attended by terrorism researchers, ethicists and other 'stakeholders'. Much of the discussion at that workshop centred on whether it was ethical for terrorism researchers to deceive participants about US government sponsorship of research in the Middle East. Here the point of comparison was with psychological experiments in which some degree of deception is used for the sake of obtaining accurate results. It was noted that terrorism research projects often propose to use partial deception or to conceal certain information from the prospective participants, especially during the consent process. The piece of information that terrorism researchers particularly want to conceal is sponsorship by the U.S. government or its agencies and military services. Researchers may wish to conceal sponsorship simply because of the antipathy that prospective subjects may have toward the culture, religion, or politics of the sponsor. (Bikson et al., 2007, p. 73) Two contributors to the discussion appear to have been supportive the use of deception in terrorism research, apparently on the basis that hostility to the United States is misplaced or ill-founded. Christine Fair from the Center for Conflict Analysis and Prevention, United States Institute of Peace remarked that since subjects will often erroneously associate any US government agency with the CIA, then a requirement for full disclosure would introduce 'a scope for misinterpretation of the study' (cited in Bikson et al., 2007, p. 62), whilst Cunningham (a Professor of Psychology and the Vice Chair of the relevant Institutional Review Board at the University of Louisville) argued that: if a subject from an adversarial culture incorrectly believes that any research conducted by any U.S. organization contributes to the Great Satan, then it is reasonable to discuss whether such beliefs must be respected by disclosing a fact that activates a prejudice or can be respected by avoiding mention of, or concealing, the inflammatory fact. (quoted in Bikson et al., 2007, p. 60) A related discussion around informed consent at the RAND workshop concerned the potential harm not only to research participants, but to the broader populations of which they are part. Brian Jackson, associate director of the think tank's Homeland Security Program, remarked: In counterterrorism research, one of the goals is to affect negatively the well-being of the terrorist groups that you're looking at. The downsides can go beyond incarceration to actual physical harm, not unlike the physical harms that regulation of medical research is intended to prevent. (quoted in Bikson et al., 2007, p. 121) RAND sociologist and Institutional Review Board member Ricky Bluthenthal similarly noted that Within the context of terrorism research, [the requirement for beneficence] can be very challenging, as the ultimate beneficiaries are likely to be in the U.S. while likely harms fall almost exclusively upon potential research participants and others who might support terrorist activities. (quoted in Bikson et al., 2007, p. 84) This issue is starkly expressed by Dolnick in Conducting Terrorism Field Research: given the nature of terrorism research there is a frequent conflict between the interests of the research funders and the people who are being studied, and some research results are likely to assist in the formulation of governmental counterterrorism efforts, which in essence seek to deliberately undermine the 'wellbeing' of 'participants'. (Dolnik, 2013, p. 224) Significantly for our purposes, this situation is treated by Dolnik (2013) as if it calls into question the appropriateness of standard ethical guidelines and institutional safeguards, rather than raising ethical questions about participating in 'governmental counterterrorism efforts'. Similarly, in the RAND workshop, there was no substantive discussion of the ethics of working for the US Government, despite a clear awareness at times of the ethical issues at stake. This is perhaps surprising given the historical and contemporary controversies surrounding the relationship between academics and the US national security state, though perhaps less so given the history and orientation of the Rand Corporation itself (Abella, 2009;Amadae, 2003;Burnett & Whyte, 2003;Collins, 2002;Jardini, 2013;Kaplan, 1991;Smith, 1966). This apparent contradiction illustrates one of our central contentions about the need to see ethics not as an abstract and worthy, if irritating, distraction from, or barrier to, research, but as an arena that can open up broader debates on research practice and partnerships. It is, in our view, crucial that such debates integrate questions of social power, academic freedom and the societal responsibilities of scientists and academics.
In Conducting Terrorism Field Research, the lack of attention to this question is perhaps most striking in Knights' discussion of the ethics of interviewing detainees. Knights notes that 'human research in Iraq [initially] took a relaxed view to informed consent but this tightened up considerably after the Abu Ghraib prison abuse scandal'. 'US militaryaffiliated researchers', he continues, 'successfully received approval to interview prisoners in US military custody in 2007 and 2008 for human research subject review boards established by the Pentagon'. To satisfy ethical requirements, Knights notes, the detainees were informed that they were under no obligation to speak to the researchers and were offered no assurances that they would receive any assistance in securing their release (Knights, 2013, p. 120). Knights then cautions against researchers conducting interviews in post-occupation Iraq given 'well founded allegations of torture and extrajudicial detention' in the country (Knights, 2013, p. 120). This remarkable disregard of human rights abuses committed by Western powersso often justified with reference to 'counter-terrorism'is underscored by Ranstorp's rather casual citation of a conversation with MI6 officer Mark Allen in his account of conducting fieldwork in the Middle East. (Ranstorp, 2013, p. 48) Mark Allen was at that stage under investigation for his alleged involvement in the kidnapping and torture of two Libyan dissidents, Abdel Hakim Belhaj and Sami al-Saadi, as well as the kidnapping and imprisoning their families, including the former's pregnant wife, Fatima Boudchar. That criminal investigation established that Allen was aware of these abuses and that he had reported them to colleagues in the UK, although the Crown Prosecution Service declined to bring charges, citing insufficient evidence on the particular allegations against Allen. In May 2018, following an out-of-court settlement of a legal claim made by Belhaj and Boudchar against the UK Government, the British Prime Minister Theresa May apologised to the couple for their 'appalling treatment' stating that '[t]he UK government believes your accounts. Neither of you should have been treated in this way. The UK government's actions contributed to your detention, rendition and suffering'. (quoted in Cobain, Bowcott, Crerar, & Shaheen, 2018, May 10).
In the same collection, Horgan (2013, pp. 193-194) at least acknowledges that ethical issues are raised by a close association with the state counter-terrorism apparatus, referring to the 'dilemma of how social scientists can usefully contribute to government policy initiatives without violating their professional ethics', and noting the controversy which surrounded official involvement in research in conflict situations including Project Camelot in the 1960s (the covert US Army counterinsurgency study focussing on Latin America (Horowitz, 1974)), the post 9/11 Minerva Project (which funded social science after 9/11 (Gusterson, 2009;Kelly, Jauregui, Mitchell, & Walton, 2010)) and the Human Terrain Team system (which recruited anthropologists to enhance military cultural competence in Afghanistan (González, 2009, this volume)). Yet he does not go on to address this dilemma, focusing instead on how 'terrorism' research can be effectively conducted whilst satisfying standard ethical requirements. This apparent lacuna in the ethical issues raised by scholars working closely with 'national security' interests is typical of 'orthodox' terrorism research, where scholars, in Burawoy's (2011) terms, operate very much within the 'policy' paradigm.
One notable exception to this is a 1991 paper published in a special issue of the terrorism studies journal Terrorism and Political Violence entitled 'Terrorism Research and Public Policy'. That special issue came out of a 1987 seminar attended by ten terrorism researchersincluding prominent orthodox experts Clark McCauley, Martha Crenshaw, Ariel Merari, Jerrold Post and Ehud Sprinzakas well as officials from the US Departments of State and Defense, and the FBI. A year before that seminar, the International Court of Justice had ruled in favour of Nicaragua in its case against the United States over its support for the so called 'Contras'. The judgment, among other things, held the US responsible for the publication and dissemination of a 1983 manual on 'Psychological Operations in Guerrilla Warfare' (colloquially known as the Tayacán) which included instructions on the use of 'implicit and explicit terror' and 'armed propaganda'. (Case, 1986;Tayacán, 1984) Nicaragua v the United States examined one part of a much broader pattern of US support for (state and non-state) terrorism in Latin America and elsewhere (Blakeley, 2009;Chomsky, 2002Chomsky, , 2015Menjívar & Rodriguez, 2009;Raphael, 2009). Yet despite this context, which was made more prominent by the Iran-Contra scandal, only one contributor to the aforementioned seminar and the subsequent special issue, raised any ethical concerns about working with US state counter-terrorism operations: the Peruvian journalist Gustavo Gorriti. He warned that close collaboration between researchers and 'state security' can threaten the safety of researchers and risks undermining intellectual freedom and the quality of research (Gorriti, 1991, pp. 105-107). He concluded his paper by calling for 'a clear distance between independent (academic, journalistic) research on terrorism and government', as opposed to the 'slightly incestuous echo-talk with trusty in-house academics'. (Gorriti, 1991, p. 115) These examples illustrate the lack of attention given to, or the unease with, ethical standards in the 'orthodox' terrorism literature. But they also say something about the ideological conformity in the field that is evident in the inability of researchers to think in ethical terms about the agencies with which they closely work, or the communities that they research.

Reassessing institutional oversight
In recent years, terrorism researchers have focused especially on the strictures imposed by Institutional Review Boards or Research Ethics Committees (IRBs/RECs). Rasmussen (2013, p. 71) argues there is a need for terrorism researchers to 'educate' such bodies 'on the nature of fieldwork in their field', whilst Atran (2007) has criticised the 'dictatorial powers' they wield, describing the sorts of requirements they place on fieldworkers as 'nuts'. Radicalisation scholars Neumann and Kleinmann (2013, p. 378) have similarly complained that 'research boards frequently impose unreasonable and unrealistic conditions that prevent fieldwork from being carried out'. IRBs/RECs have been subject to much wider criticism for hindering significant research and restricting the academic freedom of researchers (Dingwall, 2008;Hammersley, 2009, p. 218;Schrag, 2010). We agree with some of the concerns raised about such bodies, and believe that ethical guidelines and procedures need to be reassessed. Any reassessment, though, needs to preserve, and in some cases broaden, ethical protections and considerations. Ethical oversight is in principle extremely important, not only for protecting research subjects, but also for defending the autonomy of social science and the integrity of social scientific knowledge. In some cases the strongly risk averse approaches complained of may be inappropriate, but the real problem, in our view, lies in the narrow way ethical concerns are currently addressed by these bodies.
IRBs/RECs, influenced especially by medicine and psychology, have been primarily concerned with protecting research participants from harm, and academic discussions of research ethics have generally operated under the same assumption that the 'researched' need to be protected from researchers, with the former assumed to be in a subordinate position vis-à-vis the latter. There is, moreover, a considerable body of literature on methodology in the social sciences that draws attention to the power differentials typical in qualitative research in particular. In some research, though, power may lie more with the 'participant' than the researcher. This is a point made in the context of terrorism research by Dolnik (2013, p. 228), who writes: 'The notion that an academic sitting with gunmen in their stronghold is the more powerful figure in the relationship, and that the gunmen constitute a "vulnerable population" is somewhat bizarre'. Dolnick suggests that in terrorism research researchers will likely be under more immediate threat from a 'terrorist' than vice versa. Whilst this may be true in the case of some fieldwork, 'terrorists' on the other hand are almost by definition less powerful than the political and military adversaries in which they are engaged in 'asymmetrical' conflict, and with which many terrorism researchers have some association. Indeed, Dolnick somewhat overplays his fieldwork scenario. Knights' account (2013) discussed above is more typical in terrorism research, with the researcher embedded within the state machinery, and subjects vulnerable for that reason. We should remember also that identifying 'terrorists' amongst a research population of civilians is not necessarily an exact science and the temptation to define research subjects as a 'suspect' community brings with it its own ethical issues. This is especially important to bear in mind in terms of the trajectory of the debates over research ethics. Consider, for example, Atran's suggestion that certain kinds of IRB approvals should be taken at the national rather than the university level … The advantage of a national board is that their sponsors could be government agencies whose interests focus more on their mission (such as national security) than on protection of undergraduate students. (Atran, 2007) This illustrates the problem very well. Having noted the onerous restrictions placed on researchers, Atran proposes a solution that would weaken ethical protections in line with how states define 'national security'. This will only further weaken the autonomy of social scientists in decisions about research. 1 As with Dolnik (2013), IRBs/RECs are treated by Atran largely as unwarranted obstacles to social scientific research. But Atran not only fails to recognise that ethical issues are raised by scholars working for the 'national security' apparatuses, he even advocates the greater involvement of such interests in social science as a solution to the perceived problem of overly cautious IRBs/RECs.
In principle, we agree that standard ethical prescriptions are ill-suited to research in which the researcher is vulnerable. This may be because the research is being conducted in a dangerous environment. Or it may be because the research subjects are powerful enough to threaten the researcher even within their own institutional setting. Indeed, one of the problems with the way ethical questions are discussed and prescribed is a failure to consider the different sets of ethical questions that might emerge when researchers are engaged in 'studying up' (Miller, Mills, & Massoumi, 2018;Nader, 1972;Williams, 1989). There is a clear case for the defence of ethical prescriptions in terms of the ability of researchers (public, critical, professional and policy) to undertake valid and reflexive research. The encroachment of the state in such oversight would harm all types of research. Ethical precepts, far from being brought more under the influence of the state, should rather be strengthened to protect researchers and research institutions from state incursions, and should include much more robust protections from institutions and organisations that have the potential to harm researchers and undermine research.

The ethics of impact
Impactin Burawoy's terms the utilisation of academic knowledge by extra-academic audienceshas become a greater priority for universities and funding councils in recent years, and is the subject of much commentary and debate. However, it has not been considered enough in ethical terms. A sociological model, drawing on Burawoy's typology, enables us to move beyond a dominant conception of 'impact', which is currently understood too often in terms of deploying the instrumental knowledge of professional social science towards problems defined by policy actors. This narrow policy-driven conception of impact normalises and depoliticises policy-orientated social science and mitigates against any normative evaluation of impact (Selby, 2018).
The dominant narratives around impact assume that policy or practice should be changed as a result of research. But as some have argued, it may be as or more important to ensure that proposed changes are not introduced. Demonstrating the impact on the maintenance the status quo, though, may be more difficult than demonstrating impact on change (George, 2012). The evaluation of impact leading to change or stasis, also depends on other evaluations of the desirability of the specific change; perhaps better expressed as the extent to which it will likely cause or ameliorate harm. To further explore these ethical considerations in relation to terrorism research, we consider an impact case study by the leading terrorism research centre in the UK, the International Centre for the Study of Radicalisation (ICSR) at Kings College, London (KCL).
The ICSR submitted an 'impact case study' for the 2014 Research Excellence Framework (REF), claiming that it had 'informed some specific changes to UK government policy' (Kings College London, 2014). The case study (Kings College London, 2014) goes on to claim that ICSR's research was prominent in the review and revision of the UK official strategy for Preventing Radicalisation that was undertaken by the government in 2011. In June 2011, both ICSR's reports on online radicalisation and on prison radicalisation were cited in the revised official strategy for Preventing Violent Radicalisation, alongside an earlier report.
Though the document notes that the UK Government cited the ICSR's work on radicalisation, there is no claim that the ICSR originated either the concept of radicalisation, or its emergence as a new area of research. This is because the current use of the concept did not originate with social scientists, but with the state itself. Ranstorp, who was noted above for his reference to Mark Allen, specifically recalls how in 2003 he 'happened to share a unique speaking platform with Sir David Omand, Tony Blair's Cabinet Intelligence and Security Coordinator, at the British Defence Academy in Shrivenham' (Ranstorp, 2010, p. 1). Omand that day unveiled the UK's new counter-terrorism strategy, which would later influence policy on 'radicalisation'. 'For Sir David, myself and others present at the Defence Academy that day', Ranstorp writes, 'it was clear that preventing violent radicalization had to be an overarching priority to complement the tactical intelligence, law enforcement and military firefighting efforts occurring across different theatres around the world'. (Ranstorp, 2010, p. 1) The concept of 'radicalisation', as Ranstorp's remarks illustrate, originated with the ideas and practices of state personnel and institutions, only then to be developed by researchers, many of whom closely identify with the interests and perspectives of those institutions (Mills & Miller, 2017). Radicalisation theory, therefore, is an example of what Burnett and Whyte (2003) term 'embedded expertise', in action. Alex Schmid, a leading orthodox terrorism scholar, notes, it 'has become a political shibboleth despite its lack of precision'. (Schmid, 2013, p. iv) Its emergence and growth, has been driven by the desire of European and North American states to develop knowledge about Muslims that could inform counter-terrorism practices. Schmid writes: 'we have to admit that in the final analysis, "radicalisation" is not just a socio-psychological scientific concept but also a political construct, introduced into the public and academic debate mainly by national security establishments' (Schmid, 2013, p. 13).
In addition to noting that the government cited its research, the ICSR case study claims a specific contribution to policy: In publishing several reports on aspects of radicalisation and counter-radicalisation, ICSR made an important contribution to anchoring this new area of policy practice in a body of scholarship and giving it empirical and conceptual grounding.
Whether the ICSR has in fact contributed to a sound body of scholarship on 'radicalisation' is, though, questionable. ICSR director Peter Neumann has noted the field's indeterminate conceptual ambiguities and the lack of consensus even around fundamental causal questions (Neumann, 2013;Neumann & Kleinmann, 2013;Vike-Freiberga & Neumann, 2015). Presumably he would consider the ICSR's work an exception in this regard, but even if we accept that this is the case, there is little evidence that any sound radicalisation scholarship (from ICSR or any other source) has informed UK Government policy (which operationalises this ill-defined concept with negative impacts on civil liberties and human rights (Kundnani, 2009;Mohammed & Siddiqui, 2013;Singh, 2016)). Without such evidence, the above 'impact' statement is more suggestive of the legitimation of state policy by an independent research institution, than its anchoring in social science, let alone policy change rooted in evidence-based research. In this respect it is revealing, we think, that Neumann has responded to 'critical' scholarship on 'radicalisation' by calling on researchers to stop questioning its conceptual validity, and instead to 'work harder to understand and embrace a concept whichthough ambiguousis likely to dominate public discourse, research and policy agendas for years to come'. (Neumann, 2013) The obvious risk with this approach is that researchers will adopt not only research agendas, but also pseudo-scientific concepts, from policy makers, with the result that the evidence base is undermined and the state is meanwhile provided a veneer of social scientific respectability for harmful and ill-informed policies.
This brings us to a related problem with the impact agenda. The concept of radicalisation is, as many scholars, have noted, including those in ICSR itself, intellectually slight. Nevertheless, policies based on it have had significant negative impacts on civil liberties and human rights. (Kundnani, 2009;Mohammed & Siddiqui, 2013;Singh, 2016).
This illustrates that the question of the potential or actual harms caused by the impact of social research requires more attention, since, in principle it might be the case that resources flow to those that engage in research that either contributes to policy changes that may incur harm to citizens or to the political process more widely.
The close relationship with the UK government claimed by the ICSR, also raises another issue with the impact agenda. There is, as a recent study found, an 'increased risk of conflicts of interest emerging' as academics 'work more closely with beneficiaries who co-fund or support their work' (Chubb & Reed, 2018). We can see this in the example of the ICSR. Neumann himself has written (in an article funded by a grant from the US Department of Defense) that it is 'difficult to imagine any sustained scholarly effort in the areas of terrorism and radicalization research without [governmental] funding', and in the same article he concedes that this funding has contributed, if not to potential conflicts of interests, then at least to 'bad research' (Neumann & Kleinmann, 2013). A further example, however, suggests that the risks are not only to the integrity of the social scientific evidence base, but are potentially broader.
The ESRC funded project on dissident Irish Republicans is discussed in length by Mark Hayes in this issue. The project proposal, titled A Theoretical and Empirical Assessment of the Membership, Strategies and Tactics of Dissident Irish Republican Organisations, had a problematic 'pathways to impact' statement. It stated that the research would be disseminated via 'special briefing notes' to a wide range of policy and political figures including 'senior officers within the Police Service of Northern Ireland'. The proposal also stated that the researchers had already briefed Chief Constable of Northern Ireland, the General Officer Commanding (GOC) of the British Army in Northern Ireland, the Garda Siochana (the police force in the Republic of Ireland) and MI5, the British domestic intelligence agency. The researchers involved in the project (and who were conducting the interviews with Irish Republican dissidents) were unaware of this statement in the original proposal until it was made available online. The implications of this are detailed in Mark Hayes' article (this volume), so we will not repeat them here. However, it does underline some of the broader problems with how impact is currently conceived, specifically in relation to the potential harms that such issues may cause for researchers, the researched, and for the evidence base.

Conclusion: protecting professional social science
Whilst there is, ethically speaking, nothing inherently objectionable in policy oriented research, the interests and intentions of the 'client' and the researchers' relationship with policy actors should properly be the subject of ethical reflections, transparency and perhaps restraints. Yet, the way ethics has been institutionalised within the academy means that funding relationships and impact related activities are rarely scrutinised in ethical terms at any stage in the research process. If relations with extra-academic actors prompt such reflection, it tends to be on relations with social movements and other non-elite actors involved in research or public engagement. From the perspective of professional social science, this is ironic since such actors, being less powerful, pose much less of a threat to the integrity of social scientific knowledge and practices than powerful policy actors, who are much more readily able to shape research and broader institutional priorities and practices. Indeed, the dominant conception of 'impact' plainly creates institutional incentives for the greater instrumentalisation of social science, which in our view must be resisted. Burawoy's (2011) typologywhich sociologically situates knowledge production practices as well as relations with extra-academic actorsallows us to think about 'impact' ethically and in policy and political terms; less in terms of narrowly defined 'beneficiaries' and more in terms of the ethical repercussions (potential harms) of social scientific research. It also allows us to consider the potential 'audience' for academic knowledge as including not just policy makers, but also social movements, 'civil society' actors, as well as broader 'publics'. Thus the sorts of 'reflexivity' Burawoy (2011) associates with 'critical' and 'public' academic work is therefore essential, not only in epistemological terms, but also ethically in terms of the use to which the 'instrumental knowledge' will likely be put. This would obviously include scholarly collusion in military and intelligence operations and in torture, deception or other forms of coercive power (Miller & Mills, 2010).
Much research on conflict and 'terrorism' has operated within the policy paradigm, producing instrumental knowledge for state actors, a tendency which has been readily problematised by critical terrorism studies (Jackson, 2007;Jackson et al., 2009). As we have made clear, we certainly do not consider policy relevant research as inherently problematic. It is plainly in the public interest that scientific expertise be readily utilised in policy making. But we do believe that relations with state actors should be the subject of both ethical, political reflection and institutional safeguards so the state actors and/or 'policy' paradigm is not allowed to undermine professional social science. Problems to be solved can legitimately arise from policy priorities and processes, but the research questions, design, and methodologies used to understand and solve problems must be free from external interference, to the greatest extent possible. If the professional integrity of social science is to be maintained, state incursions into the research world must to be resisted. All aspects of the research process can be affected from research funding, access to data, contractual issues, editorial influence, the extent to which data and research findings can be published and/or made public, to attacks by powerful interests or the media on research which challenges vested interests. These should never allow policy actorsor other intereststo unduly influence research programmes. Transparency and academic freedom should be sacrosanct. Note 1. Cunningham, an 'IRB professional' who participated in the aforementioned RAND seminar, seemed at least to some extent amenable to such a shift, remarking that: 'National security interest is not a rationale that trumps all other ethical considerations, but it may contribute some additional merit to justify some increased risk to the subjects'. (Bikson et al., 2007, p. 52)

Disclosure statement
No potential conflict of interest was reported by the authors.