1Designing an ethical research protocol to collect data from the Internet can be a confusing endeavor. Indeed, one is confronted by a booming and dynamic field of research with widely diverging practices. These diverging practices in turn create compelling debates in the academic community, and spur rich epistemological and ethical reflections. However, far from these debates, researchers also have to comply with another form of ethics. It is the normative and institutional type of ethics, as illustrated by the need to receive approval by one’s Ethical Review Board (ERB). Moreover, in the European Union, the General Data Protection Regulation (GDPR) is another norm by which digital researchers must abide.
2To help current and future Internet researchers, this article will present the ethical dilemmas and eventual choices that were made in the case of an online study of antifeminist communities. Since ethics is necessarily a situated practice, it does not aim to give one-size-fits-all rules, but rather to illustrate a concrete ethical decision-making process and thus outline this process for others. The first section briefly introduces the study and the ethical issues it raised. Next is a review and analysis of available resources for researchers designing an ethics protocol for Internet-based research. Then I present a case study on two aspects of the protocol: informed consent and data protection. Afterwards, some recommendations to make the process easier are presented. Lastly, an appended diagram recapitulating the article is provided.
3This article is based on the ethical issues that arose while designing a protocol for my PhD research study currently being conducted at the universities of Lille and Kent.
- 1 For definitions and taxonomy of manosphere groups, see Manoel Horta Ribeiro et al., “The Evolution (...)
4The study is a qualitative analysis of an antifeminist Internet discourse corpus. The corpus encompasses five online communities, often collectively referred to as “the manosphere”1. It comprises YouTube videos, blog posts, forum threads, e-books, and discussions from social network Reddit, spanning the last two decades. Most of the corpus was selected for its relevance, while a portion of it was randomly sampled on publicly accessible online message boards for representativity.
5This discourse corpus is critically analyzed using various qualitative methods from the social sciences. These include examining word choices, word associations, sources of knowledge, knowledge circulation, deconstructing arguments and ideas, etc. None of these analyses are automated.
- 2 The documents selected in the corpus contain occurrences of antisemitic, hateful, misogynistic, vio (...)
- 3 Such as Jake Davison who shot five people in Plymouth, UK on August 12, 2021, or Alek Minassian who (...)
6Given the wealth of data represented, and the sensitive nature of the content2, it was clear from the outset that an ethics protocol would be necessary. In fact, one of the groups under study is the “incel” community, in which several mass killers have been involved in the past few years3. The protocol thus needed to consider how to protect not only the data and anonymity of research subjects, but also the researcher’s safety.
7While most of the existing literature on Internet research ethics is rather theoretical, there were some very practical questions at hand: should all manosphere Internet users whose discourse is analyzed be informed? Should their consent be sought? Could they be quoted verbatim in the dissertation? Should their online usernames and pseudonyms be protected?
8To see how the decision-making rationale to answer those questions emerged, the following section will review the resources under consideration when designing an Internet research protocol.
9In the European Union, the General Data Protection Regulation (GDPR) came into force in 2018. It drastically limited the right of private and public actors to process other people’s personal data without consent. Indeed, its very wide definitions of “personal data” and “processing” cover most types of action necessary to study Internet communities4. For example, covertly collecting content on social media, or storing forum messages on one’s device is forbidden. So is dissemination of such content.
10While this piece of legislation can at first appear stifling to research endeavors, it contains an exception for researchers. Indeed, Article 89 states that exceptions can be made when “personal data are processed for scientific or historical research purposes or statistical purposes5”. Thus, if all is carefully justified and submitted to proper authorities (such as the university’s Data Protection Officer), the GDPR does not preclude any type of data processing for research purposes, even without informed consent.
- 6 These principles are 1) a) lawfulness, fairness and transparency, (b) purpose limitation, (c) data (...)
11However, knowledge of the GDPR and its principles remains crucial to avoid any unintentional breach of the law. Moreover, this landmark piece of legislation is rooted in values similar to those of research ethics, since it aims at protecting individuals from any harm caused by the processing of their data. Thus, even though there can be derogations, the seven GDPR principles embedded in Article 5 remain a guiding set of values for personal data protection6.
12In addition to legislation, universities have their own ethics requirements and guidelines. In my case, these were the University of Kent’s Ethics Code and Code of Ethical Practice for Research, while the University of Lille recommended two European Commission reports: Ethics in Social Science and Humanities, and Ethics and Data Protection. Moreover, there remained one last formal requirement: the research protocol needed to be approved by the Ethical Review Boards (ERBs) at both universities.
13Yet, while it is one’s duty as a citizen, strict abidance of the law and other requirements is not all there is to ethics, nor to methodology. A truly ethical decision-making process implies going beyond and engaging with the academic literature on the matter.
- 7 Noortje Marres, Digital Sociology: The Reinvention of Social Research, Cambridge, Polity Press, 201 (...)
14There are as many research guidelines as there are learned societies. Yet, when it comes to Internet research, traditional guidelines are simply disconnected from the new possibilities entailed by the digital. Digital sociologist Noortje Marres illustrates how existing academic guidelines can be completely out-of-touch with the reality of the field: “It is certainly not easy to imagine how the ethical guidelines contained in the BSA [British Sociological Association] ethics statement of 2002 could be implemented in projects of social media research […]. To highlight one seeming absurdity: should we really try to message all accounts in a given Twitter data set notifying them of our data analysis project and giving them the opportunity to opt out of the data set7?”.
15Overall, compared to traditional social science guidelines, or to the more stringent biomedical ethical precautions, the field of Internet research ethics is a terra incognita of sorts. There are a lot of grey areas, and practices and outlooks in the field are immensely varied. This is caused by two factors: the swift pace of technological innovation which thwarts most attempts at establishing unified and stable frameworks, and the various disciplinary backgrounds of Internet researchers. Indeed, a Big Data scientist, a virtual ethnographer, and a literary scholar can all study the same online community, but they will do so with such diverging paradigms and epistemological premises that this is sure to entail widespread divergence on ethical issues and choices.
- 8 Not only is it widely cited in academic literature, but it is also cited by institutions such as th (...)
- 9 Aline Shakti Frantzke et al. and the Association of Internet Researchers, “Internet Research: Ethic (...)
16However, one society stands out, because it is consistently referred to in the field of Internet research ethics, and because it is specialized on the matter: the Association of Internet Researchers (AoIR)8. This association has been issuing guidelines for Internet research and reflecting on these complex questions for more than two decades. Its most recent ethics report is entitled Ethical Guidelines 3.0 and was issued in 20209. It represents a go-to for all Internet researchers during the design phase of the research.
17The approach of the AoIR is self-described as pluralistic, reflective, and dialogical. It is defined as a “process approach”, or “one that aims to develop guidelines from the bottom up in a case-by-case-based approach without making a priori judgements whether some research per se is unethical10”.
18Strict rules would not make sense in most cases for the risks, regulations, and contexts are too variable, and ethical questions are by nature complex: “[…] Such pluralistic approaches again foreground the role of judgment and the possibility of multiple, ethically legitimate judgment calls–in contrast, that is, with more rule-bound, ‘one size fits all’ ethical and legal requirements. Taken together, all of this means that the best we can do is develop ‘guidelines, not recipes’11.”
19In fact, the Ethical Guidelines argue that “the issues raised by Internet research are ethical problems precisely because they evoke more than one ethically defensible response to a specific dilemma or problem. Ambiguity, uncertainty, and disagreement are inevitable12.”
20To sum up the spirit of the AoIR guidelines: there is no absolute truth in the domain. Thus, researchers should make appropriate ethical decisions on a case-by-case basis. To help researchers make those judgment calls, the Association provides series of questions for risk self-assessment, processes for ethical decision-making, and examples from the literature.
21What better way to decide how to solve a problem than to see how peers decided to solve that same problem? If some guidelines are clearly outdated, and if others refuse to provide “recipes” to the Internet researcher, then looking at the practices in the field is crucial. For each methodological decision inspired by ethical dilemmas, reviewing the choices made by other researchers can prove invaluable. It does not, however, imply mimicking their research designs and decisions.
22In trying to determine whether to collect informed consent from manosphere Internet users, reviewing the choices of other researchers made the decision much easier. For each study I reviewed, the following questions were asked: How did the researchers handle the issue of informed consent? Did they justify their choice? How so? What are the underlying values and principles behind their justification? What criticism could be formulated against their justification? The advantages of such a review are illustrated in the following section.
23In this section, the ethical choices made in the manosphere study protocol are presented for two issues: informed consent and data protection. The rationale behind those choices was to reconcile legal and institutional injunctions, AoIR guidelines, and examples from the literature to create cogent ethical decision-making.
- 13 European Commission, Ethics in Social Science and Humanities, 2018, p. 13.
24Informed consent is the cornerstone of traditional social science and biomedical research ethics: “Most social science research endeavours are such that human participation requires evidence of the voluntary, free, and informed consent of those who contribute their time, insights, effort and data for the use of researchers13.”
25However, there is a clear consensus that informed consent is not always applicable to Internet research designs, most notably because it is completely impossible to apply to “Big Data” research. Official guidelines now recognize this, although they do not indicate any alternatives, as illustrated by the European Commission’s Ethics in Social Science and Humanities report: “In principle, living individuals should not be the subject of a research project without being informed, even in the relatively rare cases where research methods, conditions or objectives dictate that they are not made fully aware of the nature of the study until its completion. However, the advent of the internet and the widespread use of social media platforms and other ICTs have dramatically expanded opportunities for researching human behaviour without the express consent of the subjects. In turn, this has created a range of ethical dilemmas and challenges for the research community14.”
- 15 Aline Shakti Frantzke et al., op. cit., p.10.
26The Association of Internet Researchers (AoIR) documents why obtaining informed consent is not always possible and shows that in such cases, the attention of researchers is focused on “mitigating risk against research subjects”15 rather than a strict application of the informed consent principle.
27To the best of my knowledge, no study of online manosphere groups has ever obtained informed consent from Internet users to process their forum and social media content. Researchers justify this decision in many ways.
- 16 Adeline Branthonne and Elena Waldispuehl, “La netnographie pour étudier une communauté masculiniste (...)
- 17 Angela Nagle, “An Investigation into Contemporary Online Anti-Feminist Movements”, Phd Thesis, Dubl (...)
28Firstly, some argue that the research subjects would refuse, making any attempt at studying them impossible: “The specificity of our online research field is not that we hate our research object, but that our research object hates us and denies our capacity to produce rational analysis. As such, getting informed consent from the website’s users is by definition impossible for they express deep distrust towards female social scientists16.” Indeed, as explained by another scholar researching the manosphere: “These spaces are extremely hostile to any attempts to de-anonymise their subcultures […]17.”
29Moreover, this scholar adds that it might represent a cybersecurity or safety hazard to the researcher: “[…] a polite academic request would have been met with disdain and would have made the researcher vulnerable to being hacked by contributors to the site18.”
- 19 Katrine Rummelhoff, “Incels and Misogyny; What’s so Appealing about Hatred?”, Master's Thesis, Univ (...)
30In fact, a Norwegian Master’s student studying the manosphere found herself in a situation very similar to mine: “According to the American Sociological Association’s (ASA) Code of Ethics, in dilemmas of informed consent “sociologists may seek waivers from an authoritative body with expertise in the ethics of social science research” (2018). I therefore contacted the Norwegian Center for Research Data (NSD) to ensure a proper research procedure. In conversation with NSD they understood that it would be difficult and unsafe for me to try to visibly insert myself into their online environment19.”
- 20 Scott Wright, Verity Trott, and Callum Jones, “‘The Pussy Ain’t Worth It, Bro’: Assessing the Disco (...)
31Finally, seeking informed consent might not even be feasible given the volume of data under study, as explained by a research team: “It was not practical to get informed consent from all users given the large volumes involved20.”
32These justifications not to obtain informed consent all apply to my own research design. Moreover, since a sizeable part of the corpus is composed of archives from now defunct websites, obtaining consent from people who used to post on these forums is impossible. Reviewing peer choices on this specific issue made the decision quite clear, while also providing potent arguments for requesting a GDPR derogation. The goal is not to mimic these choices, but to see if their justifications apply to one’s own research, as was the case here.
33Therefore, it was decided not to seek informed consent from the people whose online content will be analyzed. This breach of GDPR was approved by the university of Lille’s Data Protection Officer (DPO), and a derogation was issued under the provisions of Article 89. However, following the recommendation of the DPO, an information notice was posted on my research blog. This message is addressed to manosphere members who wish to know if they are part of the study and to be removed from the dataset21. It was validated by the DPO and will remain online for the duration of the study and the dissemination phase.
34Yet, analyzing online material without consent does not mean that privacy should not be protected. In fact, it makes privacy and data protection concerns even more salient.
35Once it had been established that informed consent to access personal data would not be asked, other issues remained concerning the analysis, storage, and dissemination of the research data from the corpus: could it be directly quoted? Could the Internet pages under analysis be linked towards in the PhD dissertation? Should usernames and pseudonyms be anonymized?
- 22 Elizabeth H. Bassett and Kate O’Riordan, “Ethics of Internet Research: Contesting the Human Subject (...)
- 23 Ibid., p. 245.
36Theoretical outlooks on this point often diverge. In a 2002 article, Bassett and Riordan call for paradigm change22. Indeed, they argue that Internet research is still being governed by the 20th-century Human Subject Research Model, a model that fails to consider that the Internet is not made of people, but of texts. Thus, they advocate for dealing with Internet content as one would with texts, and not with subjects: “There are issues and rights at stake in these debates other than those of privacy and safety. The Internet user is also entitled to a degree of representation and publication in the public domain. If an individual or group has chosen to use Internet media to publish their opinions then the researcher needs to consider their decision to the same degree that they would with a similar publication in traditional print media23.”
- 24 Cited in Luciano Paccagnella, “Getting the Seats of Your Pants Dirty: Strategies for Ethnographic R (...)
37Along the same lines, an approach that is often cited and used in the literature–no doubt because it is conveniently permissive–is that of Sheizaf Rafaeli: “We view public discourse on CMC [computer-mediated communication] as just that: public. Analysis of such content, where individuals’, institutions’ and lists’ identities are shielded, is not subject to ‘Human Subject’ restraints. Such study is more akin to the study of tombstone epitaphs, graffiti, or letters to the editor. Personal?–yes. Private?–no24.”
- 25 David Wilkinson and Mike Thelwall, “Researching Personal Information on the Public Web: Methods and (...)
38In short, some researchers hold that Internet content is just like a published text: voluntarily released and publicly accessible, and therefore “fair game” for analysis or criticism. On the other hand, some scholars contend that no direct quotations of Internet content can be made, even if the author is pseudonymous, and even if the content is public: “[…] This may serve to draw attention to the people involved and hence could be seen as breaching privacy. In consequence, it seems necessary to avoid including identifying information in research publications and to avoid quotes or anonymize them by paraphrasing or altering words so that they are not searchable25.”
39All the material in my manosphere corpus is publicly accessible, without any password protection or the like. Moreover, none of the sites under study require creating an account to view content. However, people posting on those sites can still be expected to have expectations of privacy. In fact, European Commission guidelines give a lot of weight to user expectations:
- 26 European Commission, Ethics and Data Protection, 2018, p. 13.
“The fact that some data are publicly available does not mean there are no limits to their use. If your research project uses data from social media networks and you do not intend to seek the data subjects’ explicit consent to the use of their data, you must assess whether those persons actually intended to make their information public […] It is not enough that the data be accessible; they must have been made public to the extent that the data subjects do not have any reasonable expectations of privacy26.”
40This outlook is shared by the AoIR guidelines, that distinguish between subjects (who do not intend for their public online activity to be widely circulating) and authors (who intend just the contrary): “If participants are best understood as subjects in the first sense […], then greater obligations to protect autonomy, privacy, confidentiality, etc., are likely to follow.
- 27 Association of Internet Researchers, Ethical Decision-making and Internet Research, 2002, p. 7.
41If, by contrast, subjects may be understood as authors intending for their work to be public […], then fewer obligations to protect autonomy, privacy, confidentiality, etc., will likely follow27.”
42Most of the content in my corpus comes from notorious manosphere activists. Their identities and political commitments are public, and their goal is to reach and disseminate their political ideas. They can safely be considered as authors of texts. Therefore, their content will not be anonymized, and no privacy measures will be enforced. Since they willingly give interviews, film themselves or get photographed to promote their ideas, no restrictions will be applied to using pictures of them or screenshots from their videos, for example.
43On the other hand, pseudonymous Internet users writing on forums or writing comments on articles, posts or videos probably have an expectation of privacy. Therefore, in keeping with the above AoIR and European Commission recommendations, their messages, comments, and posts will be treated as personal data. Consequently, even though it is publicly accessible online, this data from pseudonymous Internet users will be processed as follows:
-
It will be fully anonymized in writing.
-
No pictures of these pseudonymous Internet users shall be reproduced.
-
No superfluous data shall be retrieved or stored (data minimization principle).
-
The data will not be shared with third parties and will be stored on a secure external hard drive.
-
No hyperlinks towards the original webpages shall be provided.
44The goal of these measures is to prevent re-identification of “natural persons” (this is GDPR terminology) based on the information shared. Although content will be cited directly, identifying the person behind it would entail finding the website where it was posted, then finding the message on that website to identify the pseudonym (all users on manosphere platforms are pseudonymous). Then, the pseudonym would somehow need to be linked to the natural person. Moreover, most of the websites and platforms in the corpus are currently offline and can only be accessed using Internet archive services, making this task even more difficult. Thus, these measures respect the spirit of the GDPR, protecting research subjects from potential re-identification and harm.
45While the ethics protocol for this study also dwelt on other topics–most notably researcher safety–, these examples shed light on the decision-making process. Being cognizant of the law, attuned to AoIR guidelines, and mindful of peer practices is a potent and convincing combination. In fact, the protocol was quickly granted approval by both ERBs and by the DPO.
46After reviewing some of the steps implied in designing an ethical Internet-based research protocol, here is some advice to make the process more fruitful, at the personal and institutional levels.
47The institutional ethical review process takes time. While waiting for the review board to approve one’s protocol, any form of personal data processing is suspended. This can stall the research process and generate unnecessary frustration. This delay should therefore be fully acknowledged and accounted for from the beginning of research design and scheduling. Tasks that do not require any form of data processing, such as literature review, can thus be scheduled during the waiting period for institutional ethical approval, preventing any frustrating scenarios.
48The GDPR can be very intimidating in the early stages of research design, potentially hampering all types of data analyses. More seminars and programs should be offered, in conjunction with resident Data Protection Officers, to better explain the values and principles of this law and to help researchers integrate them in their work, while also recognizing the possibility of obtaining derogations for scientific research.
49Quite often, methodology and ethics are seen as two separate dimensions of a research project, and their relationship is sequential or even oppositional. First, one designs an ideal methodology; then, to comply with regulations or other institutional requirements, some parts of the methodology are shaved off. Thus, ethics is unfortunately too often seen as a procedural artifact that can impede the good development of a project.
50Yet, there is no fundamental difference between methodological and ethical choices during the design part of a project. Both involve pondering the consequences of the research design and making decisions based on a set of values. The difference is in the set of values involved for methodology (accuracy, transparency, representativity) and ethics (protecting privacy, physical integrity, mental wellbeing). Therefore, the goal is to establish a more symbiotic relationship between the two: they should be considered jointly in the early stages of research. Thus, ethical reflection would stop being a negative but mandatory presence in the research design process, to become a rich and fertile step that could inform methodological choices and vice versa.
51In fact, this process gave me more confidence in my methodological decisions by establishing solid justifications for all the items in the research protocol. Ultimately, ethical choices are very personal: while respecting regulations, recommendations, and peer practice, they should also feel right. Personal accountability is crucial, as expressed by this rule-of-thumb from Kent’s Ethics Code, “consider how you would explain this decision or action if you had to justify it to close friends or family, or if it was on the front page of a newspaper.28”
52Confidence in one’s justification can only be strengthened by conducting ethical reflection early in the research design phase. Hopefully this article can shed light on ways to make this process more efficient and thorough.
Appendix: resources and obligations for ethical Internet research design
Flux RSS
