NAVIGATING A RUGGED COASTLINE - Bonn International Center for ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
F O R S C H U N G S PA P I ER 1 Netzwerk für Extremismusforschung in Nordrhein-Westfalen CoRE – Connecting Research on Extremism in North Rhine-Westphalia NAVIGATING A RUGGED COASTLINE Ethics in Empirical (De-)Radicalization Research Kerstin Eppert I Lena Frischlich I Nicole Bögelein Nadine Jukschat I Melanie Reddig I Anja Schmidt-Kleinert
NAVIGATING A RUGGED COASTLINE / EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A.
SUMMARY ZUSAMMENFA S SUNG CONTENT S
The current paper documents the results of a 1 Research ethics in (de-)radicalization
multi-disciplinary international workshop on research 3
ethics in (de-)radicalization research, organized by
the authors at the Bochum Center for Advanced 2 Research context and project set-up 4
Internet Studies (CAIS) in March 2019.* In light of 2.1 Empirical (de-)radicalization research in context 4
the specificity of each research context, this paper 2.2 Awareness and transparency regarding public
aims to develop a baseline for ethical standards and private obligations 5
across projects, disciplines and contexts. The 2.3 Mapping the field and determining
aim is to provide the instruments that allow for operational requirements 5
a project-specific setup that enables researchers 2.4 Setting up accountability mechanisms 6
to develop their own concepts and solutions in
the context of their empirical field. The following 3 Data collection 7
sections elaborate on the above issues, pointing to 3.1 Identifying, collecting and creating data 7
practices and possible proceedings so as to ensure 3.1.1 (Inter-)active approaches to data collection 7
ethical standards in (de)radicalization research. 3.1.2 Passive-observant approaches of data collection 9
3.2 Delimiting and outlining responsibility and
Das vorliegende Forschungspapier dokumentiert accountability 10
die Ergebnisse eines multidisziplinären interna- 3.3 Reproducing the field 10
tionalen Workshops zur Ethik in der (De-)Radikalis-
ierungsforschung, den die Autorinnen im März 4 Data management 10
2019 am Bochumer Center for Advanced Internet 4.1 Organizing data 11
Studies (CAIS) organisiert haben.* Im Hinblick auf 4.2 Data processing and data sharing 12
die Besonderheiten der einzelnen Forschungskon- 4.3 Archiving data 13
texte soll dieses Papier eine projekt-, disziplin- und
kontextübergreifende Basis für ethische Standards 5 Data analysis and data interpretation 14
entwickeln. Ziel ist es, Instrumente bereitzustellen, 5.1 The risk of analytical bias and perpetuating
die eine projektspezifische Aufstellung ermögli- stereotypes 14
chen, die Forscherinnen und Forscher befähigt, 5.2. Analysing empirical data towards outputs
eigene Konzepte und Lösungen im Kontext ihres for policy and practice 15
empirischen Feldes zu entwickeln. Die folgenden 5.3 Applying general quality criteria of social
Kapitel gehen auf die oben genannten Fragen ein science to radicalization research 15
und weisen auf Praktiken und mögliche Vorge-
hensweisen hin, um ethische Standards in der (De-) 6 Publication of findings 17
Radikalisierungsforschung zu gewährleisten.
7 Concluding remarks and practical steps 18
References 20
* The work on “Ethics in (De)Radicalization Research”
goes back to the work of an off-spring working group to
the network Connecting Research on Extremism (CoRE)
in North Rhine-Westphalia, CoRE Digital, that was con-
stituted in 2017 and in which the authors are active. We
are grateful for the synergetic effects of the network.
Our deep gratitude further goes to the Bochum Center
for Advanced Internet Studies (CAIS) who provided
financial and organizational support to an interdisci-
plinary peer workshop that we consider the kick-off
for what we hope to become a continuous dialogue in
online and offline research.
2EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A. / NAVIGATING A RUGGED COASTLINE
Navigating a Rugged Coastline—
Ethics in Empirical (De-)Radicalization Research
“Make ethics an integral part of research from the beginning on.”
(Dechesne 2019)
1 Research ethics in (de-)radicalization challenges of (de-)radicalization research. More specific
research contributions (Mahlouly 2019, Hutchinson, Martin &
Sinpeng 2017, Marwick, Blackwell & Lo 2016, van Gorp
Ethical considerations play a crucial role in all empirical re- 2013) are rare and not often shared across disciplinary
search, with particular challenges in every discipline: From boundaries (Hutchinson, Martin & Sinpeng 2017, Marwick,
theoretical considerations, over the planning and con- Blackwell & Lo 2016, Winkler 2018, van Gorp 2013, van
ducting of empirical studies, up to data analysis and the Gorp & Feedes 2013). Yet, research in the field of (de-)
publication of results, researchers are always confronted radicalization is ‘interdisciplinary’, raising the need for
with various situations in which they need to navigate the common ground on which research teams from different
rugged coastline between ethically right versus wrong disciplines can build on.
behaviors, carefully preserving the equilibrium between,
for instance, the protection of research participants and Researchers need to act responsibly to protect vulner-
the value of academic insights gained. In empirical (de-) able persons who participate in or are affected by their
radicalization research, particular challenges arise from research; they need to be mindful of both, and their
the implications that the research process itself, as well social environment, while also being mindful of their own
as its findings, may have for the safety and well-being of exposure. Researchers in (de)radicalization research are in
individuals, social groups, and the society at large. Fur- contact with multiple actors and stakeholders, including
thermore, empirical (de-)radicalization research is in the youth and young adults, individuals at risk, their families,
spotlight of public attention and may come into the focus educational staff and counselors, but also police services,
of security sector agencies, for example, in the scope of judges and courts or policy-level institutions. This creates
risk assessments (Hoffmann et al. 2017). At a fast pace, a context where different personal concerns and insti-
politicized topics and research fields become salient and tutional mandates may converge and potentially lead to
evolve—often carrying high hopes as to their short-term tensions and conflicts of interest that need to be resolved.
policy and practical output. Additional challenges arise At the same time, researchers conducting interviews
from broader societal developments such as technological in the field, for example, may take on an exposed role
advances and the access to big data, and debates about in a contested field and therefore encounter threats by
open science, developments which are not limited to but individuals or groups that are connected to a radical scene
also concern (de-)radicalization research. and who may take a negative interest in both, the research
and the researcher.
Applying standards of good scientific practice is generally
rather straightforward where rules and regulations are In the quest to standardize and professionalize ethical
spelled out, and where technological procedures of data practices of academic research, a number of scholarly
management are already established. Ethical guidelines books and practitioners’ guidelines are dedicated to
and practice support are the more required where our fundamental questions of moral responsibility (Forge
work reaches into yet unregulated realms and makes 2008, Briggle & Mitcham 2012) and autonomy (Kämper
use of new technological features, for example, in data 2016) reflected in scientific standards and guidelines of
collection. At present, guidelines on research ethics in ethical practices in data collection and analysis (Van Gorp
general (Deutsche Forschungsgemeinschaft, DFG, and & Feedes 2013, Roth 2005, Deutsche Forschungsgemein-
Leopoldina National Academy of Science, LNW 2014) schaft [DFG] and Leopoldina National Academy of Science
or online research in particular (Eynon, Fry & Schröder [LNW] 2014). Others address questions of normativity
2008, Tscherwinka 2014) do not account for these specific (Unger 2014), the specifics of ethics and data protection
3NAVIGATING A RUGGED COASTLINE / EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A.
in Internet research (Eynon, Fry & Schröder 2008, Golla, ly, the ‘deviant’ from the ‘norm’ that eventually poses a
Hofmann & Bäcker 2018, Tscherwinka 2014) and (big) data threat to society (Schiffauer 2015). Implicitly, the concep-
management (Boellstorff et al. 2012, Ess 2002 Markham & tualization of radicalism and extremism as fringe phenom-
Buchanan 2012, Rieder & Simon 2017, Ebel & Meyermann ena denies extreme attitudes in the mainstream (Oppen-
2015, Meyermann & Porzelt 2014, Zimmer & Kinder-Kur- häuser 2011, Falter 2011). This perspective is compounded
landa 2017), whereas clinical research (Wendler 2017, by an almost exclusive use of the term radicalization in the
Wendler & Ried 2017) has provided transferable knowl- context of Islamist radicalization (Toprak & Akkus 2019).
edge on ethics in human-related research. Yet, in the case This not only results in tendencies to overlook or trivial-
of empirical (de-)radicalization research, transferring ize right-wing extremism (Schmidt-Kleinert 2018), it also
standard practices requires further adaptation, as we affects the public discourse on Islam and Islamophobia
argue in the following sections. in Western societies (Amir-Moazami 2018). Researchers
need to acknowledge the profound ethical implications of
Drawing on prior work and the above-mentioned interdis- their research.
ciplinary workshop with 26 experts from social sciences,
law and the humanities, this paper aims at establishing The epistemic foundations of a research project require
common ground and constitutes a first step to what we careful thought and critical review. Decisions about the
hope will become a continuing discussion on the subject. theoretical terms, conceptions and research questions
In the following sections, the argument is developed of a project can already have a severe impact on the life
along the empirical steps of a generic research process, of the studied individuals (de Koning, Bartels & Koning
covering (1) Research context and project set-up; (2) Data 2012). Research projects are part of a process of knowl-
collection, (3) Data management; (4) Data analysis and edge production that, in and by itself, constructs the
data interpretation; and the (5) Publication of results. In social world. Research projects operating with concepts
each section, the specific challenges of the respective step of “radicalism”, therefore, inevitably take part in a process
are highlighted and discussed as to possible responses. in which social categories are not only described but also
Finally, the conclusion summarizes best practices that may co-constructed (Hummel et al. 2016). The mere definition
serve as a starting point for disciplinary refinement within of the term “radical” at the outset of a research project
research projects in this field. already shapes expectations and may impact the public
perception of who is, and who is not, a radical. This could,
subsequently, lead to discrimination and social exclusion
(Hummel et al. 2016). Related to that, research partici-
pants may be concerned about observation and prosecu-
tion by state agencies if they participate in the research. If
2 Research context and project set-up researchers are not transparent about their professional
role, this might cause fear and suspicion among research
Radicalization is a problematic concept that is increasingly participants or vulnerable communities (Amir-Moazami
criticized within the (de-) radicalization research com- 2018, Toprak & Akkus 2019). In both, interaction with re-
munity (Logvinov 2018, Schmidt-Kleinert 2018, Jukschat search participants and public communication, it is hence
& Leimbach 2019). Though widely used, the term lacks important to be transparent about project funding and
clarity and, therefore, analytical potential. Although its partner institutions.
meaning is ambiguous and definitions vary, the concept
of radicalization tends to individualize the problem and
implicitly conceptualizes the process of radicalization as 2.1 Empirical (de-)radicalization research in context
linear and determined This is already reflected in the titles
of radicalization models, for example, Borum’s (2011) Two contextual factors in (de)radicalization research need
“Four-Stage Model”, Moghaddam’s (2005) “Staircase” or to be considered for the project setup and the specifi-
McCauley and Moskalenko’s (2008) “pyramid”. The most cation of research questions. The first contextual factor
widely shared understanding of “radicalization” is that of concerns a strong political interest in the findings, the
a process through which people become violent “extrem- second, resulting therefrom concerns the emphasis on
ists”, whereby, similarly to that of radicalization, the terms and expectations of the practical use and benefits of (de-)
extremists and “extremism” are subject to critique: they radicalization research for policymaking. As a result, over
are both imprecise, judgmental, and normatively connot- the past years, policy-driven funding for (de-)radicaliza-
ed and, therefore, orientate scientific, political and public tion research has substantially increased, particularly in
discourses. When used in political contexts, radicalization European countries (Wehrheim 2018, Teune & Ullrich
becomes a categorical term of security policy that distin- 2018). Nowadays, a substantial part of policy-related secu-
guishes “non-radicals” from “radicals” or, more specifical- rity sector and radicalization research receives funding
4EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A. / NAVIGATING A RUGGED COASTLINE
from state agencies—in some cases, research consortia or Both, the securitization of (de-)radicalization research
thematic networks even include operational branches or and the increasing dependence on third-party funding
internal research departments of executive institutions, create dynamics by which researchers risk losing sight of
or even intelligence agencies (de Koning, Bartels & Koning their professional roles. Dechesne (2019) emphasizes that
2012). researchers may fulfil different roles in a given research
context and should be mindful of clarifying in what capac-
These context factors set challenges for academic ity they present themselves: the role of the public or pri-
research that researchers and state agents alike need to vate person, that of an employee of a research institution
be aware of. First, the framing of research funding within and that of a local community member, may not always
policy programs may lead to revised criteria for and de- be separable. This goes, in particular, in situations, where
cisions on financial support to research: (Implicit) criteria the failure to comply with legal obligations constitutes a
defining a project’s legitimacy and relevance for policy criminal offence. For most EU-countries, this applies, for
development might replace the assessment of theoretical example, to denial of assistance2, the obligation to report
and empirical soundness and a project’s epistemological planned crimes3 or any threat to a child’s welfare. To avoid
scope. Such a paradigm shift would put at risk basic prin- this happening and provide appropriate guidance to pro-
ciples of academic research. Second, concerning research ject staff, it is highly recommended to discuss and develop
cooperation between state agencies and academic insti- standardized procedures for cases where researchers
tutions, the involved stakeholders need to keep in mind receive data on, or witness illegal or dangerous situations,
that the functional logics of the sectors are juxtaposed to and establish distinctive referral procedures to psycholo-
one another, as concerns their mandate, accountability gists or the police in advance.4 Ambiguous behavior may
structures and their internal organization. impact and, in the worst case, delegitimize research. It is
therefore important to regularly reflect on the boundaries
For the organization of the research process, from pro- of our respective public and private speaker positions.
curing the equipment and sharing empirical data, to the
organization of meetings and the publication of results, We suggest that researchers can check their main roles by
this constitutes a challenge that requires continued asking themselves the following questions:
attention and exchange. For all project partners, written
cooperation agreements that cover all substantial areas of (1) Can I ascertain that I produce rigorous, structured, and
good scientific practice are most helpful since they detail confirmed knowledge that may be communicated and
responsibilities and expectations, and secure the buy-in proofed by others and that, therefore, aims for intersub-
from institutional hierarchies.1 jectively comprehensible knowledge of objective validity
(the scientific role)? Or, (2) do I try to provide new informa-
tion about current events of general interest? (This would
2.2 Awareness and transparency regarding be one of the main functions of journalism) (Weischen-
public and private obligations berg, Malik, & Scholl 2006). Or, rather, (3) am I trying to
change (societal or political) conditions, for instance, be-
In the field of (de-)radicalization research, scientific knowl- cause a group is treated unfairly? (This would be a goal of
edge production is often intertwined with the production collective action) (van Zomeren, Postmes & Spears 2008).
of security knowledge. Using the concept of “(radical) Or, finally, (4) does my project strive to implement legal
ization” can quickly lead to research questions that deal conditions consistent with the general law? (This would be
primarily with security issues. Researchers should be the mandate of the executive institutions in a state).
aware that projects examining, for instance, the radicali-
zation potential of certain living conditions or groups pro-
duce knowledge that can—and most likely will—be used 2.3 Mapping the field and determining
by security agencies. In this respect, the projects contrib- operational requirements
ute inevitably to a securitization of social life in Western
societies (Singelnstein & Stolle 2012). Increasingly, (de-) (De-)radicalization research confronts researchers with
radicalization research is further used to serve prevention specific challenges—amongst others the regular en-
purposes and, thereby, becomes an overarching mode of gagement of researchers and participants with security
future-management (Bröckling 2008).
2 In Germany, for example, this is §323c StGB (Strafgesetzbuch,
German Criminal Code).
1 This similarly applies to research cooperation with the private sec- 3 For Germany, § 138 StGB.
tor. The monetization of research results touches upon another 4 See Marwick, Blackwell & Lo (2016) for further guidance on risk
dimension. assessment and institutional awareness in “risky research”.
5NAVIGATING A RUGGED COASTLINE / EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A.
agencies and research conducted in securitized contexts. external colleagues (knowledgeable peers), who convene
Further to reflecting on the different roles that may affect regularly (e.g., once or twice a year), or on an ad hoc basis
the research process, Dechesne (2019) suggests conduct- in case of need, to review and validate ethical standards in
ing a mapping of the field to identify relational interde- operational proceedings and the research design, access
pendencies of actors, institutions and agents involved in to the field or any other issues the team of researchers
the research and their—implicit and explicit—expecta- feels need to be discussed. While codes of ethics are part
tions towards the project. Such a mapping exercise helps of most professional associations’ constitutional frame-
to understand contingencies and allows to address ethical work (Deutsche Forschungsgemeinschaft 2013, Ameri-
concerns. These interdependencies may be visualized as can Psychological Association 2016) and internal review
suggested below (Figure 2.1) along relational dependen- boards become more and more common in fields such as
cies and institutional hierarchies. Such a systematic over- psychology and communication, a systematic implemen-
view helps to depict challenges and risks emerging on all tation of ethical standards at the project level or internal
relational dimensions at an early project stage and offers ethical review boards are not yet standard practice in
the possibility to include perspectives of a diverse set of social sciences.
partners (e.g., via a stakeholder exchange).
In the expert’s workshop as well as within the wider
network of participating researchers, many colleagues
reported on various project-related arrangements they
Relaonship dimensions on which
potenal risk should be analysed developed to cover the need for ethical guidance and
counselling. In addition to formal advisory boards, the
?
described practices included peer-to-peer as well as pro-
Authories fessional supervision to cope with psychological pressure
Researcher(s) or the establishment of peer support networks to share
experiences. In some instances, universities make psy-
chological counselling available to their staff and provide
structures within their occupational health management.
To date, professional supervision of research teams work-
ing in sensitive contexts and lacking access to adequate
university-bound support structures is funded only insuf-
Parcipant(s)
Intermediators
ficiently, if funded at all (Mahlouly 2019). To raise aware-
ness of the issue, researchers may want to include the cost
for the use of case networks and the development of team
Figure 2.1. ‘Mapping relational dimensions and risks’
support structures (e.g., supervision/retreats) in their
(Based on the ‘Safety and security mapping’ of the H2020 project
budget.
DARE ‘Dialogue about Radicalisation and Equality)
The litmus test of universality, fairness and transparency
As a “fast check” of one’s own research practices against
Such mapping also allows updating the current interde-
ethical benchmarks of universality, publicity and fairness,
pendencies in the course of the project and, thus, account-
Hünler (2019) suggests to pose the following questions:
ing for changing requirements due to an evolving field of
Could my research practice be universal? That is, (1) would
research changes, for example, after policy changes, as
I recommend this course of action to anyone else in similar
a response to new technologies or a changing security
circumstances? Moreover, (2) would I condone my behav-
situation.
ior in others? Could my research practice be made public?
In other words, (3) would I tell other researchers what I
intend to do? (4) Would I be willing to have my actions and
2.4 Setting up accountability mechanisms
the rationale for them published on the front page of the
local newspaper or reported on the evening news? Is my
In light of the complexity of planning and implementing
research practice fair? That is, (5) would I treat another
research projects in social sciences, projects that operate
participant, independent of his or her status, in the same
in sensitive contexts and aim to ensure high standards of
situation differently?”
internal and external accountability are well advised to
‘discharge’ part of their responsibility by setting up inter-
nal review mechanisms. Such a mechanism may, for exam-
ple, consist of a small board of three to five internal and
6EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A. / NAVIGATING A RUGGED COASTLINE
3 Data collection 3.1 Identifying, collecting and creating data
“Freedom of research also means To determine whether or not it is ethical to collect data
freedom to not conduct certain research.” in a specific field of research, it is necessary to assess the
(Hünler 2019) method (that determines which data is collected in what
manner) against the context parameters and the consti-
tution of the field. This equally applies to quantitative and
Once the research context is defined, and the operation- qualitative methods, as for both approaches, personality
al setup of the project is underway, the formulation of rights, self-determination and personal safety consider-
research questions further determines the empirical field ations concerning research participants remain relevant
of research and narrows down methods of data collec- throughout the process of data collection. We propose to
tion. Both, the ontology of the field and methodological discuss ethical concerns regarding the identification, col-
questions (e.g., what kind of data sources are available, lection and production of data by differentiating (inter-)ac-
what kind of data is to be collected and how?) are usually tive approaches on the one hand, and passive-observant
sharpened and refined in an iterative process as the work approaches, on the other. While (inter-)active approaches
evolves. Beyond specific requirements of each scientific involve, for example, any methods that require personal,
discipline, ethical considerations in the process of data verbal or written exchanges among researchers and re-
collection mostly concern the principle of ‘doing no harm’ search participants; possibly participants’ instructions on
(Anderson 1992). to others by research interventions, and data collection instruments; experimental settings or any
by “striving for beneficence and nonmaleficence” (Amer- kinds of interviews, to name but a few, passive-observant
ican Psychological Association 2016, p. 3). In most cases, approaches include the establishment of document or
researchers do not remain at the periphery of their field media corpora, protocols of non-participant observation,
but leave imprints in it, for example, by interactions with secondary or meta-data analyses, etc.
counterparts and participants, or by shaping the overall
interpretation of the topic. They thus take an active role in The distinction of data collection processes along these
the construction of the field and this possibly even beyond two categories, so our argument goes, allows carving
the mere period of active data collection. out the differences in regard to provisions of privacy and
data protection. More specifically, it allows researchers
By consequence, ethical concerns in the process of to self-assess to what extent the data they are about to
data collection closely relate to preserving individual collect and process:
self-determination and autonomy as well as safeguarding
personality rights and ensuring the personal safety of all 1. requires informed consent of participants or necessi-
involved persons (for a more general discussion of current tates ex-post communication on aims and implementa-
privacy and data protection regulations see section 4.1). tion of the study;
2. relates to the public or private sphere and carries spe-
Against this background, the following questions are cen- cific authorship or ownership rights;
tral to ensuring a critical screening process: 3. makes provisions for opting-out of the study/research
possible; or
1. What are the legal, organizational and normative deter- 4. necessitates financial or material compensation of
minants, constraints or vulnerabilities that govern the participants.
research field? How do they determine where (access)
and how (method) data is collected and produced?
2. How do participants agree to take part in the study? Is 3.1.1 (Inter-)active approaches to data collection
an explicit consent to participation and the processing
of data possible? Data from (inter-)active approaches can be created
3. What is the impact of data collection on actors, agents through qualitative (e.g., interviews or focus groups etc.)
and dynamics in the field? To what extent are things like as well as quantitative research (e.g., surveys or experi-
post-intervention support or risk mitigation needed? ments), both online and offline. Depending on the field of
research, cultures and contexts (Markham & Buchanan
The following sections give details of the proceedings for 2012), ethical standards of data collection vary, for exam-
ethical compliance at this stage of the research process ple, with respect to the required anonymity of research
and provide further guidance to self-assessments within participants. We would like to draw on current practice
project contexts. in psychological research, where ethical standards on
data provided by human subjects require the researcher
to adhere to the criteria of (a) informed consent, (b) the
7NAVIGATING A RUGGED COASTLINE / EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A.
avoidance of unintended side-effects of data collection, of pseudo-anonymized data8, where the right of deletion
and (c) incentivization. cannot be executed meaningfully (see section 4.3).
Informed consent. ‘A priori consent’ implies that re- Avoiding unintended side-effects. (Inter-)active data
search participants are fully aware of (a) the aims of the collection allows researchers to execute certain levels of
research project and its financial sources, (b) the answers control over the questions, stimuli, and conditions they
or actions required during data collection, as well as (c) the confront their participants with. The ethical principle of
treatment of their data after data collection (see also sec- striving for beneficence and nonmaleficence is the guiding
tion 4.3). Ethical guidelines for social sciences research5, principle here. Yet, implementing this principle in the con-
as well as the EU General Data Protection Regulation (§13, text of empirical (de-)radicalization research is sometimes
GDPR) considers informed consent a necessary precon- more tricky than anticipated. For instance, seemingly
dition for creating data (Schaar 2017). Getting informed “easy” decisions, such as avoiding the exposure to scenes
consent for (de-)radicalization research projects often of graphic violence (Rieger, Frischlich & Bente 2013) or
needs time for building “networks of trust” (Dechesne removing final scenes of beheadings from extremist
2019), particularly, where vulnerable groups (e.g. minors propaganda videos (Cottee & Cunliffe 2018, Klonk 2019)
or minority groups) are part of the target group. In line used in quasi-experimental research, have been discussed
with the German Science Foundation DFG’s affirmation for critically as stimulating “curiosity” (Winkler 2018). As
honesty in “oneself and towards others” as a most crucial another example, innocuous questions about reasons that
value within the scientific process (Deutsche Forschungs- induced interviewees to join a radical movement or ter-
gemeinschaft [DFG] 2013, p. 67), transparency about one’s rorist organization could stimulate rose-tinted memories
research aims and decisions is a moral obligation both about the person’s past and therewith impair intergroup
towards the scientific community and one’s participants. relations.9
Article §13 of GDPR (§13) requires that participants can
contact the researcher directing the study.6 “In an exem- Incentivization. Where possible and appropriate, the
plary manner, the ‘Dialogue about equality and radicali- time invested by an interview partner or participant
zation project’ (DARE) uses an information sheet outlining should be recognized and compensated. This can be done
the purpose and procedures of the conducted research, in-kind (e.g. by offering access to the research results,
which provides full disclosure and has a reference to the invitation to a transfer event, etc.) or monetarily. In the
consortium website and a contact address” (Dechesne context of radicalized groups and individuals, however,
2019). 7 one might argue that monetary compensation could be
used to support ideological groups and their activities.
However, not all research allows for a priori transparen- Although it is hard to avoid the misuse of funds irrespec-
cy, in particular, when such information is likely to bias tive of the research context, the risk of “financing” illegal
participants’ answers. If it is necessary to conceal research activities can be a particular concern in the context of
aims, a combination of a priori information for obtaining empirical (de-)radicalization research. It is necessary to
consent and a post hoc debriefing that enables research remember here that compensations for participants are
participants to retract their data is considered acceptable usually rather low. However, if individual compensation
(BPD & DGPS 2016). Researchers can include a similar is a problem in this field of research, at project level, one
“opt-in” option in (pseudo-)anonymized questionnaires or strategy for compensation could be to donate funds to a
verbally in an interview situation. An accordant debrief- charitable or other civil society organization that works on
ing can be implemented in qualitative and quantitative democratic grounds and supports moral values promoted
studies and should allow for the immediate deletion of by the interview partner(s). Similar considerations apply
collected data consistent with the “right to deletion” (§17, for providing funds to research participants in powerless
GDPR)—although there are exceptions for scientific archi- situations such as minors, detainees, etc.
val purposes when deletion “is likely to render impossible
or seriously impair the achievement of the objectives of
that processing”. This is particularly likely in the context
8 On the definition of pseudo-anonymization, see §3, GDPR and
5 For Psychology, see, for example, American Psychological https://gdpr.report/news/2017/09/28/data-masking-anonymi-
Association (2016); BPD & DGPS (2016); for Sociology, Deutsche zation-pseudonymization/ (last accessed, November 18, 2019).
Gesellschaft für Soziologie (DGS) (2017). https://gdpr.report/news/2017/09/28/data-masking-anonymiza-
6 This level of transparency is also compatible with recent open tion-pseudonymization/
science movements, such as the call for pre-registration.https:// 9 For a study about the relationship between such “collective nos-
osf.io/prereg talgia” and hostile intergroup attitudes, see Smeekes, Verkuyten &
7 http://www.dare-h2020.org/ Martinovic (2015).
8EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A. / NAVIGATING A RUGGED COASTLINE
3.1.2 Passive-observant approaches of the easy access to data has led to a shift in the focus on
data collection data collection for many research projects (Kaiser 2018).
Regardless of the origin of data and the effort with which
Data collected by passive-observant approaches can be it can be collected, however, ethical standards require
obtained offline or online and analyzed qualitatively and/ transparency and adequate documentation about how
or quantitatively. For data generated by methods such as and where data was collected. The ‘free’ availability of
ethnographic or corpus analysis, ethical standards gen- data does not relieve researchers of the obligation to ac-
erally focus on questions of (a) privacy and (b) acknowl- knowledge sources via adequate citation and attribution
edgement of data sources, along with more specific needs of authorship (Markham & Buchanan 2012). In the case
for ensuring (c) the safety and security of researchers of data from propaganda documents or media outlets of
themselves (particularly, when it comes to research with terrorist organizations, the format by which sources are
potentially violent groups). acknowledged needs to be chosen with care and accom-
panied by a disclaimer as to contents, or else, researchers
Privacy. In contrast to (inter-)active approaches, in pas- risk fueling violent competition for attention by extremist
sive-observant approaches, it is not possible to secure a movements or terrorist groups (Kruglanski et al. 2018,
priori consent. To determine whether or not researchers Koloma Beck & Werron 2013) [see also section 3.3 and
can assume that any author or data owner have given Chapter 6].
their consent, it is helpful to distinguish whether data
relates to the public or the private sphere (Markham & Security. Relating to the issues of confidentiality, integri-
Buchanan 2012). Public mass communication—articles ty and civic obligations, researchers might need to decide
published in a newspaper or a propaganda-magazine—is whether to reveal their identity or conduct research
usually considered unproblematic for analyses, where- incognito. Researchers studying risky topics such as (de-)
as interpersonal conversation (for example, in a home radicalization need to watch their own security and the
or closed family chat) relates to the private sphere and security of their team members (Marwick, Blackwell & Lo
is given special legal protection. While this distinction 2016). In consequence, some (de-)radicalization research-
seems to be quite obvious offline, the limits of mass com- ers choose to stay anonymous when monitoring violent
munication and interpersonal communication become extremists’ online communication channels (Bloom, Tiflati
blurred in social media (Walther & Valkenburg 2017). & Horgan 2017) or offline networks (Ebner 2017). From a
Here, individuals might consider their (public) tweets as technical point of view, the use of fake identities is rather
private (for much-needed insights into user perspectives simple. From an ethical perspective and regarding re-
in this context, see Williams, Burnap & Sloan, 2017), and search methods, however, it can be very problematic as it
classify social media platforms as semi-private spaces violates the already mentioned principle of honesty. Sim-
(Neubaum 2016). The resulting challenge has already ilarly to other decisions that concern a derogation from
stimulated some debate in the context of empirical (de-) ethical practices, the decision to use fake identities for
radicalization research (Buchanan 2017). The increasing safety reasons needs to be carefully justified and weight-
relevance of “dark social” platforms and encrypted mes- ed against the epistemic value of the research (see section
sengers for the general public (Newman et al. 2019) and 3.1).10 Here, the principle of proportionality as well as the
extremists suggests that semi-private spaces will remain consideration of explicit or implicit consent, as discussed
a crucial field for future (de-)radicalization research in the next section, are helpful.
(Bloom, Tiflati & Horgan 2017, Neumann et al. 2018). As
an attempt to solve the inherent ethical challenge, Golla
(2019) argues that research may be considered ethical
when the social value of the results outweights the in-
trusion. Taking medical research as an example, a further
justification might need to be given when not only third
parties but also the researched individual (data owner)
him- or herself are expected to benefit from the research
(Wendler 2017).
10 Methodically, the use of fake accounts is also concerning in
regard to the metrics of accounts, chat groups and websites as it
Acknowledgement of data sources. Data access is a is no longer possible to estimate whether a group of 300 persons
challenge in all empirical research and, when given the actually contains third party observers or original members and to
ascertain the origin of a communication. As to the criminal-legal
option, researchers may prefer to choose to collect data in perspective, the infiltration of online groups can become an issue
an easy-to-access field rather than in a closed-up one. Par- where consequences of communication can no longer be con-
ticularly in the case of publicly available data from online tained, for example, where chat partners might feel encouraged
to pursue violent acts due to ambiguous communicative behavior
sources such as Twitter, VKontakte, Facebook and others,
on behalf of the rest of the group.
9NAVIGATING A RUGGED COASTLINE / EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A.
3.2 Delimiting and outlining responsibility 3.3 Reproducing the field
and accountability
Even when collected data refer to propaganda material
In light of their social responsibility (§1 para. 3 DGPUK collected online only (i.e. there is no direct interac-
2017), researchers are accountable for their actions and tion with human participants), viewing, downloading,
have moral obligations to act responsibly, particularly or bookmarking accordant content on platforms with
towards their research subjects, to adhere to the law interaction-based algorithmic recommendation systems
(American Psychological Association 2016), to promote (i.e. social media sites) inevitably fuels the “success” of
fairness (DGPUK 2017) and—more generally—human accordant materials at least in the “eyes” of the recom-
rights (Markham &Buchanan 2012). These obligations mender algorithms. Restricting the number of clicks on
also need to be considered when establishing research extremist material is thus well advised, although it might
instruments—including the consideration of unintended require downloading and sharing accordant material
side effects—and inform strategies for dealing with critical within research groups. Since the (public) dissemination of
incidents (see also section 2.3). Accordingly, decisions can extremist propaganda is against the law in many countries
be easily combined with other best practices such as the including Germany, storing and sharing of accordant mate-
discussed risk analysis, routines of pre-registration, includ- rial needs to be done with care and according to high data
ing the justification of meaningful effects sizes (Anvari protection standards. Besides, particularly in the context
& Lakens 2019) and the planning of one’s sample size in of (de-)radicalization research, material collected and/
qualitative research (Haven & van Grootel 2019). or provided for research purposes may violate national
laws. Another strategy to avoid the inflation of interaction
As a minimum requirement, an ethical research design is to cooperate with intermediaries outside of research,
needs to ascertain that (1) research participants may quit e.g., platforms (see, for instance, the Facebook SoSci One
whenever they feel uncomfortable (particularly in the or Crowdtangle grants, Twitter’s academic partnership,
case of detainees or participants that are governed by or alphabet’s “moonshot countering violent extremism”
other dependency relationships) without having to fear branch), security agencies (e.g., for file access, to reach
negative consequences (including losing one’s incentives); out to incarcerated interview partners), NGOs, religious
(2) participants can contact the research director and/ organizations, or business partners. Although such coop-
or his or her focal point.11 Besides, (3) empirical (de-) erations are extremely valuable, and some of the authors
radicalization research may require additional safety nets have substantially benefitted from them, we are also
for further advice, information or referral, such as, for aware that they need to be constantly reflected to ensure
instance, including contact information to (de-)radicaliza- independency of research, particularly in light of the secu-
tion experts, psychological care, or an extensive debrief- ritization phenomenon discussed in Chapter 2.
ing, allowing the participant to voice potential negative
sentiments before incentives are provided. From our own
work, we found that sharing experiences with participant
groups and collecting background material is extremely
helpful for the development of safety nets and awareness
of potential issues. To give an example: Teachers, security 4 Data management
officials or teenagers tend to judge media content quite
differentially from each other, and what seems innocuous “Data protection is a resource of trust.”
in one cultural context might have diametric effects in an- (Golla 2019)
other. Talking to people outside of one’s narrow “research
bubble” or the literal “ivory tower” in advance can help Data management refers to the organization, processing,
to become aware of such pitfalls (e.g., through regular ac- sharing, storage and archiving of ‘created’ and ‘observed’
counts to and exchanges with the advisory board suggest- research data. While data management and data protec-
ed in section 2.4). tion are oriented toward fulfilling ethical and professional
standards regardless of the discipline, specific legal and
technical requirements may apply to online and offline
(de)radicalization research. Building on key questions we
have encountered in our own research, the next section
offers some guidance and possible response to these
specific challenges. The sections are developed along the
aforementioned elements of data management—from
11 These criteria are also standards set by the EU-GDPR and ethical
guidelines of professional associations, for example, American organization to archiving—and focus on legal and techni-
Psychological Association (2016). cal aspects respectively in each section.
10EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A. / NAVIGATING A RUGGED COASTLINE
Challenges One of the best ways of dealing with data organization and
planning ahead is to prepare a project-specific data pro-
For European researchers, in particular, it is sometimes tection policy (“Datenschutzkonzept” in German) at the
hard to keep up with the evolving EU and national legal outset of a research project. This allows thinking through
regulations as well as with ethical guidelines of profession- the technical set-up, legal regulations and the ethical
al associations and research institutions. Also, evolving code applicable to the project and, thereby, establishing
technological innovations allow for new approaches to basic rules for data collection and management. In many
conducting, sharing, processing and analysing research research environments, such a data protection policy note
data, but may also lead to yet unregulated realms of data is already mandatory. A suggested structure for such a
management. Particularly in a sensitive field such as (de-) concept is presented in section 4.2 below. The data pro-
radicalization research, there is an inherent tension be- tection policy should integrate the legal requirements as
tween the values of open science and privacy protection. well as technical aspects of data management.
Related to that, researchers might find themselves in a dou- In projects that are currently conducted in the EU, the
ble- bind situation as regards their obligation of confiden- General Data Protection Regulation15 enacted on May 25,
tiality towards respondents and—depending on national 2018, provides most likely the most important normative
regulations—their legal obligations to disclose information framework to privacy rights and data protection.16 National
on severe criminal offences or to testify before a court.12 legislation and policies that apply specifically to scientific
research also need to be checked. In Germany, for instance,
Concerning decision-making processes, some professions relevant institutions, working on data protection and ethics
are bound to institutionalized codes of conduct and legal are the National Office of the Ombudsperson for Data
guidelines, as are, for example, psychotherapist13, which Protection17, the sub-national Ombudspersons (Landes-
help them navigate accordant questions. In contrast, dateschutzbeauftragte), the Commission for Data Ethics
other professions, such as some in social sciences, operate at the Ministry of the Interior18, and the institutional focal
in a much more fluid and contingent field, which increas- points for data protection of public and private research
es exposure and ambiguity. For researchers in Germany institutions. In some cases, the legal departments of the
and the United Kingdom, for example, personal liability research institutions also provide guidance and advice as
regarding data management and/or the disclosure of do university-based focal points for research ethics or insti-
information is regulated by different laws.14 tutional ethics committees. Also, international and national
professional associations and advisory committees provide
useful guidance notes on the application and transfer of
4.1 Organizing data legal frameworks to scientific research. In Germany, the Rat
für Sozial- und WirtschaftsDaten (RatSWD) constitutes a
In many research contexts, research institutions, research standing advisory committee to the government on issues
partners, beneficiaries and participants demand that the of data management in social sciences and also publishes
researcher provide a written commitment to data protec- position papers, for example, on data protection.19 The
tion and privacy regulations. Research institutions provide legal framework of data protection has an impact on the
different infrastructures to support researchers in writing technical and procedural organization of data.
such a concept. Usually, researchers can use the sup-
port of the institution’s legal department, research data Once the empirical field of research is determined and
management services—often attached to the library—or evolves as researchers immerse themselves into data
decentralised support units in departments or faculties. It collection, the more intuitive parts of data management
is also worth checking, whether one’s university provides generally relate to the
appropriate training for doctoral students or postdocs.
We strongly advise involving local contacts for data pro-
tection at an early research stage.
15 GDPR at https://eugdpr.org/.
12 See, for example, University of Sheffield, Policy Note No. 12, Re- 16 Beyond the EU’s geographical scope, the OECD Privacy Framework
search involving illegal activities, at https://www.sheffield.ac.uk/ (2013), under revision as of December 2019, constitutes another
polopoly_fs/1.112762!/file/Research-Ethics-Policy-Note-12.pdf; relevant policy framework.
Lowman & Palys (2001). 17 Bundesdatenschutzbeauftragte/r at https://www.bfdi.bund.de/
13 In Germany as per §§203 and 138 StGB respectively. DE/Home/home_node.html.
14 For Germany, these are §§1,2 VerpflG, §5 BDSG, and §§203-205, 18 Datenethikkommission at https://www.bmi.bund.de/DE/themen/
353b StGB. For the United Kingdom, the Terrorism Act (2000) it-und-digitalpolitik/datenethikkommission/datenethikkommis-
specifies a “Disclosure of information: duty” in section 19 (1), see sion-node.html.
http://www.legislation.gov.uk/ukpga/2000/11/section/19). 19 https://www.ratswd.de/themen/datenschutz.
11NAVIGATING A RUGGED COASTLINE / EPPERT, K. & FRISCHLICH, L., BÖGELEIN N., JUKSCHAT, N., REDDIG, M., SCHMIDT-KLEINERT, A.
❙ data format (e.g., audio files, transcripts, video se- the storage of data containing private information usually
quences, photos, digital pictures, rich text, document is prohibited.
scans, experimental and survey data, etc.);
❙ software needed to structure, organize and eventually In line with EU legislation, researchers are obliged to
analyze the primary or secondary data (e.g., statistical establish procedures to control and document access to
programs, quantitative or qualitative analysis tools like data that contains private information. To this end, data
Atlas.ti, MaxQDA, F4; media player, SQL/postgreSQL security tools, such as firewalls on internal networks, or
databases); commercial, respectively freeware, encryption tools may
❙ hardware needed to store data to and access data from be used. To control physical access, many institutions
for working sessions (see ‘Data Processing and Data working with sensitive data have safe rooms or special ar-
Sharing’ below); and chives accessible to authorized personnel only. Here, mo-
❙ arrangements for data archiving or secure deletion. bile storage tools (like SSDs, hard drives, USB sticks, etc.)
or hard copy documents and other data can be stored, for
To sum up, some of the major challenges facing the organ- example, in steel cabinets. From our own experience, it
isation of data are: may further be practical to establish stand-alone work-
stations without Internet connection or connection to in-
❙ An increasing obligation for researchers to observe ternal networks to secure authorised access and maintain
legal regulations and ethical guidelines; high data security standards throughout processing and
❙ At the same time, often a practical lack of knowledge analysis. These workstations could also be used to store
and training about these obligations or recent changes; data.
❙ Lacking juridical support or advice to the researcher in
the research process; Since sensitive data requires special technical equipment,
❙ Legal protection on privacy and data protection may it is necessary to include the cost for additional technical
allow for the processing of data, regulations may, equipment into project budgets. In some of our projects,
however, still conflict with the researcher’s professional we worked with a substantial number of court files, parts
ethics” of which we scanned for the analysis. Since most network
scanners store all documents on an internal memory that
cannot be accessed and erased, one institute chose to
4.2 Data processing and data sharing buy a separate copy/scan machine that was available at a
local workstation; another institute worked with a mobile
From a legal perspective, any import of research data into scanner that had no internal memory. To minimize the
a software is already considered as data processing and risk of losing files and mobile storage tools in cases where
thus falls under regulations of privacy and data protec- cooperating institutions are physically remote, it is advisa-
tion.20 Researchers need to be aware that this can even ble to limit physical transportation and prioritize personal
include copying data, e.g. from social media, via the copy- handover that can be documented and signed-off. Some
and-paste option into any other software to store it. Also, institutions, mostly of the private sector, have access to
any sharing of data is data processing. highly securitized clouds or may recur to well-protect-
ed internal servers that may serve the purpose of data
Thus, the first step in protecting data from unauthorised transfer.
access is to secure privacy rights. As to storage, it is im-
portant to know that EU privacy legislation only accounts We followed and valued the standard procedures of
for data stored on servers in the EU. Yet, in practice, a processing and storing of sensitive data in the above-men-
number of cloud services are provided by firms registered tioned research projects that dealt with court files, i.e. the
in the United States, where the legal situation is different “data life cycle” that is described in the following:
and standards for data protection are much lower. One
notable difference is that government agencies may get
access to data stored on US servers without the knowl-
edge of the data owner. Using EU-based solutions might
not always be possible: cloud solutions like ILIAS or Power
Folder usually provide a description of the kind of data
allowed for storage on these platforms and indicate that
20 See § 4, (2) and (6) GDPR and Recital 26, GDPR at https://gdpr-info.
eu/recitals/no-26/ , last accessed, November 19, 2019.
12You can also read
