Cite this as: Davidson, K., Graham, S. and Huffer, D. 2024 A Protocol for When Social Media Goes Private: Studying archaeological or heritage discourses in closed Facebook groups, Internet Archaeology 67. https://doi.org/10.11141/ia.67.11
For archaeologists and historians (and indeed, the academy writ-large), social media data can be a compelling locus of study. Broadly speaking, the ethical issues related to studying public social media posts are understood and protocols have evolved to guide their study (with regard to archaeology for instance see e.g. Richardson 2018). Many of these issues deal revolve around respecting privacy and consent.
Over the last several years we have collected, studied and then disposed of a large amount of social media material concerning discourses around human remains made in public. We feel we have a good understanding of what the public face of the trade in human remains via social media looks like on major platforms and have recently recounted the synoptic perspective in a monograph length work (Huffer and Graham 2023). Now the question arises, how do people talk about human remains in private? Do inhibitions drop? Do they talk about the same things in the same ways but less afraid to name prices? Do they share more egregious materials (e.g. "murderabilia"; Denham 2020)? Over the last few years, the profile of human remains traders has been greatly increased by friendly profiles in newspapers (e.g. Beauregard 2022) and in magazines (e.g. Schwartz 2019) as well as by a number of cases where people buying and selling human remains, who obtain them under dubious circumstances, have come to light in the press or in the courts (e.g. Stoll and Kath 2023). We have noticed people taking their conversations into private chats or into other social media platforms that require some kind of barrier to entry when they reach a certain point, places where normally we cannot follow.
How can we ethically study private or closed social media where barriers to entry exist? In our case, we wish to understand how the desire to own, trade, control, and collect human remains as facilitated in closed social media platforms differs from what we see on public platforms. When people buy, sell, trade, display, collect, or discuss human remains in private, what are the discourses? Are these discourses different compared to when people participate in the trade in public? How can we even try to answer these questions ethically? In this paper, we detail the protocol that we have developed to explore these questions in an ethical manner, in cooperation with the Research Ethics Board at our institution, and in compliance with the guidelines and regulations set down by the Social Science and Humanities Research Council of Canada (part of the Government of Canada's three federal research bodies, along with Canadian Institutes of Health Research and the Natural Sciences and Engineering Research Council of Canada, referred to in this paper collectively as the "Tri-Council").
In our research to date, we have been looking at how people who collect, buy, sell or enjoy looking at human remains talk about them. Our research examines this question in the context of debates within cultural heritage and relationships around material culture, which are themselves contemporary concerns for shaping how archaeology and biological anthropology 'get along' with the public (Richardson and Almansa-Sánchez 2015, 195). We have been looking at the discourses of the trade, primarily relying on publicly accessible posts on social media platforms. In Canada, under the ethical guidelines set by the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans (TCPS), posts that are made with the public settings on "have no reasonable expectation of privacy" (Tri-Council 2022, 18). This is interpreted to mean that posts made available to the public (i.e. public-facing social media accounts, public posts on X or Facebook, or similarly blog posts and websites) do not require ethics review to study.
When we collect public media posts that are discoverable through the platform's own discoverability mechanisms, there is no obligation placed on us by the Research Council to reach out and get permission from each one of those users to study the things that they have said (a recent court case found that collecting such posts does not contravene the platform's terms of service either, since the person collecting, if not logged into the site, is not a user of the platform's service; Gennaro 2024; Meta Platforms Inc v. Bright Data Ltd 2024). Of course, that is not the same thing as saying that we therefore have every right to identify these people (see Huffer et al. 2019). Usually, we take an approach of anonymizing or de-identifying posts in order to protect the privacy of people involved in discourses around the human remains trade, who may or may not understand that the comments they have made are publicly available. Occasionally we directly quote examples of germane phrases or public social media posts to illustrate some point (see for instance, Graham et al. 2022). We cannot know a priori where in the world a person is located and hence which laws might apply to them, or which borders they may have had to cross when shipping human remains (and thus which laws related to cross-border trade that might be in play). Even if we did, the laws in many jurisdictions remain murky or are effectively silent with regards to collection, sale, and trade between private individuals within the jurisdiction (see e.g. Marsh 2015; Washington College of Law 2020; Huffer and Graham 2023, chapter 4). But since aspects of this trade are illegal and illicit (and, in our view, immoral), we do know that a person who discusses what they see could be put in a position of danger (in the words of the TCPS, "vulnerable"; Tri-Council 2022, 44-45). Thus, we do not identify individuals not just out of a respect for privacy, but also a care for individual safety (both our own, and the individuals who engage with this trade - including those who oppose it).
Research on social media groups, particularly in the context of illegal and illicit trafficking, is very important for understanding how subaltern, marginal or underground networks operate (e.g. Anagnostou and Doberstein 2024; Hughes et al. 2017; Latonero 2011; Paul 2018). Our protocol and example may be useful for other humanities or social science researchers interested in understanding the intersection of their particular discipline's research questions with the social media behind private groups. The result is a protocol for studying those posts but in a way that respects the privacy of the people who participate in these conversations. We will discuss the results of the study using our protocol in another paper (Graham et al. in prep); here, we go into the discussions we had and the protocol that has evolved which permits us to collect and study collector discourses (not - and this is important - the people making these posts themselves).
Our project has involved both close and distant reading of the posts we have collected where we 'read distantly' using various machine learning and neural network approaches, sometimes tens of thousands of posts at a time. These particular methods - aside from the collection of data - are not without their own ethical dilemmas considering some of the ways in which social media data can be used for societal good or ill (see boyd and Crawford 2012; D'Ignazio and Klein 2020; Olteanu et al. 2019). These platforms and methods are an inseparable part of our society now, and researchers need to understand how to do ethical research on/with/via these platforms. Because we are researchers affiliated with Carleton University in Ontario, Canada, we have a set of domestic regulations to follow in all research that we conduct where it involves humans. Currently, we are in the final year of a project studying the illicit trade of human remains online which has been ongoing since 2018, supported by Canada's Social Sciences and Humanities Research Council (SSHRC). SSHRC's guidelines for content that is in the public domain, including "cyber-material" that is freely available to the public (like public social media posts), indicate that no ethics review is required (Tri-Council 2022, 18). It also stipulates that we must not add additional identifiable information that could link data to a person in an identifiable manner as long as the research is taking place in a public forum using observation (and not intervention; Tri-Council 2022). As SSHRC-funded researchers, we are required to be familiar with these guidelines. SSHRC provides a training module (the TCPS 2: CORE-2022, where CORE stands for Course on Research Ethics; Government of Canada 2022) for all researchers to work through. This training is required in order to apply for ethics clearance, and ensures that we are familiar with the regulations around the ethics process before we begin.
Our institution's ethics board is composed of volunteers drawn from across the university's faculty. There are two main divisions in the ethics board, one dealing for departments and researchers coming primarily from the arts and humanities (CUREB-A) and one dealing primarily with researchers coming from sciences, engineering, psychology and cognitive science (CUREB-B). Proposals are made to the ethics board, which then meets to discuss the proposal and its adherence to guidance provided by SSHRC. Proposals may be returned for further clarification, approved, or denied. If further clarification is required, researchers may respond and amend their proposal. If approved, researchers are given a timeline and sunset date for reporting on the research results and the termination of the approval. Extensions may be granted. Our research fell under the auspices of CUREB-A.
The first step for us was to approach the ethics board informally and discuss the nature of our research, what it was we wanted to do and why we needed to do this, seeking guidance. We approached the arts and humanities research board (CUREB-A) and explained the nature of our research and our findings to date. This then prompted a serious discussion amongst the members of the board and the principal investigator about how social media works, the way that private groups work, the mechanisms that, for instance, Facebook puts into place to automate or semi-automate the kinds of necessary moderation for requests to join a private group, what the logistics of joining a group are, and what the risks were to group members should we conduct our study. The discussion revolved primarily around the concern for the privacy of the individuals within the groups we wished to study. After this initial meeting, the board retired to have their own private discussion of these issues and to conduct more research on their own. Once they had determined that our request was not a 'minimal risk' situation, they invited us to submit a formal proposal. The proposal forms are standardized to accommodate the needs of researchers from a wide variety of disciplines.
Our ethics board provides a template for crafting the proposal. We will detail the germane issues for the different sections. The initial questions were concerned with the goals of the larger study, who the project team members are and that they have completed the mandatory training from the Social Sciences and Humanities Research Council of Canada on ethics (Tri-Council 2022). We were also asked to indicate the purpose and benefits of the study. We stated that the research will benefit the growing community of researchers investigating heritage crime online, including art crime, as well as antiquities and human remains trafficking. (The block quotations hereafter are direct quotes from our proposal):
"The trade is thriving and readily makes use of the affordances of social media and similar platforms to find like-minded individuals. Buying and selling human remains is not universally illegal and whether or not collecting human remains is licit or illicit depends on the residence of the individual and can include complicating factors such as borders over which human remains are moved. We cannot tell where in the world a person is located (administrators of such groups sometimes ask for location information from individuals when they join a group. That information is merely a string of text and does not auto-populate from a user's account. Therefore it does not have to be true). But more importantly only the administrator for the group would be able to see that information, not the other group participants.
We are interested in understanding how people talk about or for want of a better word consume human remains to construct their own identity. We cannot determine where or from what groups human remains come from but historical research (e.g. Redman 2016) indicates that these are almost universally from racialized or subaltern communities. The research returns a measure of dignity to the dead. The research will also benefit society by providing tools through which to understand the antiquities trade as society increasingly becomes aware of this trade with its growing visibility on social media platforms."
The next section of the form asked how we will engage with people whom we are studying - and remember, we are not studying the people, rather, we are studying the discourses. We indicated that we are taking a 'critical inquiry approach' to observe people in online Facebook group settings who participate in collecting, buying and selling human remains as a microcosm of the wider global antiquities trade. That phrase, 'critical inquiry approach' is crucial to understanding how our approach to research ethics was ultimately approved, and will be discussed further below.
We then discussed whether or not the study constitutes a 'minimal risk'; this is a key phrase that triggers differing kinds of review. Studies that are deemed 'minimal risk' are expedited and tend to require much less detail, assuming that the context of the research takes place in a setting and with interlocutors who wouldn't encounter more risk participating in the research than they would in real life (Tri-Council 2022, 23). In contrast, non-minimal risk studies require engagement with all of the questions on the proposal template and a review by the full board at an in-depth meeting. The risks in our study are obvious. There is a risk to the researchers who investigate these topics publicly of harassment or other negative reactions by people who come across our findings. There is also a risk to the people who are participating in what may be an illegal activity depending on where they live. The form asks us to elaborate on the perceived risks by describing how we might interact with the study group, how they might be vulnerable, and what might mitigate those risks. We point out that we are not recruiting individuals, nor will we have any sort of interaction with the individuals; we will merely observe. We went on to say:
"The research group is seen as a vulnerable group, since the activities they are engaged in are sometimes in violation of state, national or international laws, particularly when this activity crosses [jurisdictional] borders (as in social media interactions). To protect the privacy of participants, we will not be including participants names in any publications, and we will use a double-blind coding system for anonymizing any names collected. We will also not be quoting any discussions that we observe, opting instead to paraphrase or summarize discussions that take place. Finally, we will not publish or show in academic presentations any screenshots from these groups."
Because direct quotes risk revealing the identities of people who made the posts (findable through key-word or -phrase searching), paraphrasing their content allows us to ensure no undue harm comes as a result of our research. In the later section on the specific kinds of risks (psychological, physical, social, economic), their nature, impact, probability and how we might mitigate these, we detailed under 'psychological' risk:
The remaining questions in this portion of the proposal concerned whether or not we had pre-existing relationships with the study 'participants', or whether conflicts of interests (financial or otherwise) might exist. There is also a section dealing with the duty to consult First Nations, any extant agreements, participant involvement in research findings, or direct benefits to participants, and data ownership, control, access, and possession in the case of research with First Nations. We were not directly dealing with First Nations individuals as study subjects so these sections were not relevant to our application.
The next section dealt with the methods of recruitment and was designed to guard against coercion, recruitment risks, and compensation. We wrote:
"There is some risk to people being observed, as they might be participating in an activity which is often seen as illicit. There is no way to know what jurisdiction may pertain, or when and with whom any sales of prohibited materials are conducted, and thus, whether or not any illegal activity is taking place. However, as we will be anonymizing any identifying information collected, there is no risk to people we observe greater than normal, based on their online interactions. This is why we are choosing to take a critical inquiry approach, so that no one participant is singled out to speak for or be a scapegoat for the trade."
On the recruitment risks to the researcher, we pointed out that there is a possibility that observed individuals might react negatively to our scrutiny of their discourses, and reiterated that this is why we would use an alias account, and why we would maintain a policy of no interactions. Finally, in response to a question about research incentives, we noted that there would be no compensation offered to the members of the groups we would study.
A principle of humanities and social science research is the idea of 'informed consent', that the people being studied should know the risks and potential benefits. 'Informed consent' is a hard-won principle that has emerged in the aftermath of dubious social and medical studies over the twentieth century that caused real harm (e.g. First Nations Information Governance Centre 2023). Obviously, obtaining informed consent in the context of the human remains trade as it manifests in private social media groups would negate the purpose of the study. Here, we turned to TCPS 3.6 which discusses and details the 'critical inquiry approach', which does not require permission to conduct research about a group. (This also renders moot the need to design withdrawal procedures, which are normally another element of the ethics application).
Critical inquiry, as defined in the TCPS Article 3.6, examines systemic or structural behaviors as they relate to groups or institutions within a society (Tri-Council 2022, 43). It allows us as researchers to examine activities that we ourselves are not engaged in to better understand the issues at play and the motivations of participants (Monahan 2017). A critical inquiry approach is understood to be necessary when the proposed research may not be accepted or approved of by the group being studied (Tri-Council 2022, 44). This assumed rejection may happen for a number of reasons, such as groups unwilling to discuss illicit activity, or critiques of institutions engaged in practices they wish to not come under scrutiny, amongst other things. It does not preclude working with groups that have additional requirements for ethical engagement, such as schools or First Nations (Tri-Council 2022, 44). Critical inquiry also assumes that the activity being studied is happening in a large forum, in the sense that the behaviours being studied are part of a wider trend or structure. In this sense, it is close to the field of institutional anthropology and institutional critique (e.g. Pereira and Moreschi 2021).
An analytical approach within critical inquiry entails drawing on observations to take stock of how power structures, hegemonic forces or enacted processes impact people in different arenas (Charmaz 2017, 35, 40; Eseonu and Okoye 2023, 3-4). These arenas, as field sites, include organizations, institutions, and communities at scales from the local to the global (Tri-Council 2022, 43). Field observations (as empirical evidence) are used to help draw out the mechanisms by which participants within a group navigate those power structures. This may present itself as resistance and revitalization, countering forces such as settler colonialism or racism, as observable through small or creative acts (e.g. Simpson 2011, 103; Eseonu and Okoye 2023). It may also present as normalizing illicit behaviour and supporting subaltern perspectives about a socially taboo practice (Davidson et al. 2021; Huffer and Graham 2023). In our research, we find that the content of social media posts within the bone trade signal both the mechanisms for attaining individual prestige (both economic benefit and social benefit within the oddities subculture), as well as the performance of an aesthetic that generates solidarity and notoriety for the group as a whole.
The next section of the research protocol dealt with the issues of deception and/or partial disclosure (and subsequent debriefing, if warranted by the study design). We wrote,
"In order to join these Facebook groups, sometimes moderators require a statement on the purpose of a request for joining, through a three-question questionnaire. These questions can be moderated by an administrator for the group. These questions can also be used with a Facebook tool called 'Admin Assist' that can automatically approve members who answer the questions: in such cases, the fact of answering is what matters, not the content. This automatic tool implies that such gatekeeping as does exist is perfunctory. Depending on the nature of the questions, if a truthful answer would 'out' us as researchers, we will not join that particular group."
Facebook's Admin Assist changes how people (including researchers or casual observers) interact with Facebook groups. While it allows for groups to better control who joins or does not join their private group (particularly in the interest of preventing bots, scrapers and spiders from accessing the information in a group), it also gives administrators the option to take a hands-off approach to group membership.
This is the portion of the proposal where we detailed the methods that underpinned the earlier discussions around risks, harms, and benefits. The questions were about how data is collected (surveys, research tools for online monitoring etc.), where participants are located, frequency and duration of interactions, photography or recordings.
"The data we are interested in are the text of posts and any associated images, and the interactions which take place in the comment sections of those posts. We will be observing interactions in private groups on Facebook (i.e. groups that require permission to join). This data will be summarized and double-blind anonymized."
"This research is taking place on the Facebook platform, which is owned and operated by the company Meta, out of California, USA."
Issues around data storage and analysis were dealt with in the final portion of the proposal. The chief concern here was the ability at a later date for someone to be 'outed' as a participant in the often-illicit human remains trade through this study. In order to ensure data privacy, we chose to take on an anonymizing protocol for the data. These anonymity protocols meet the requirements of TCPS Article 10.3 on virtual platforms where there is some limited expectation of privacy given the settings of private Facebook groups (Tri-Council 2022, 191). Because screenshots of the original posts could be used to identify people participating in this activity online, the only portions of the data that are retained are the encoded portions. Within the context of this research, we noted we would only keep summarized and paraphrased data for future research, following the encoding method below:
"Any identifying information will be double-blind anonymized using coded identifiers. (…) Only aggregate data will be published. None of the coded information will be published. "
"Any data collected will be summarized [i.e. The discussion that takes place through comments on posts in the group will be analyzed, discourses summarized through natural language processing - statistical distributions and associations of words - and the original statements discarded], any identifying information anonymized, and the results will be stored on a hard drive [in a safe location]. (…) All coded data, double-blind coded summaries and working documents will be archived for future reference."
"The results of this study may become a publication, or be presented in a public academic setting. No images will be shared in these settings. Study results will not be shared with the members of the observed group(s)."
"Because we are not collecting information about individuals, the risks to participants in the case of a data breach is low."
We also appended several documents to the proposal in addition to a short bibliography of relevant work. We shared with the board screenshots of the result of a search on Facebook for 'oddities' groups (under the auspices of a new account created for the purpose). Under each returned group, Facebook includes a snippet describing the group, whether it is private or public, the number of members, and the number of posts per day. We pointed to the fact that groups with tens or hundreds of thousands of members, even while marked 'private', were likely to use perfunctory group moderation tools since there were simply too many active members for moderation to be meaningful. We then provided screenshots of what the gatekeeping questions looked like for some of the groups. Some questions, for example, were about agreeing to abide by a group's 'code of conduct'; in that case, the code was largely about being polite, and using the group to promote sales. Answering the questions and agreeing 'yes' to the questions provided admittance to the group.
The board considered our proposal and deliberated. They returned it to us with several points requiring clarification before they could render their decision. The first point turned on the issue of deception, drawing attention to the fact that our alias account would have no history to it, and that sometimes a group might turn down an account with no history or 'friends'. They wondered to what extent we would or might need to deceive group administrators. We responded:
"The alias account will be a new pseudonymous account, which is not intended to have a network of friends or be used as a typical account. As stated, some groups also use the Admin Assist feature to accept all people joining a group. Although some private groups may not accept this alias account with its "new account" status, there are others that will."
These groups may be using the private Facebook group settings for a number of reasons, not just selectivity for their participants. These reasons include being omitted from Google searches, from being scraped by bots or, in some cases, because of genuine concern about the potentially illicit nature of these activities, though not always. There may also be an element of the 'joining' process providing group admins with a veneer of plausible deniability. The admins might set up the process so that a user has to agree to 'respect the law' (without the administrator having to get into the complexity of jurisdictions and legal interpretations) before being allowed to join - thus, administrators who provide/manage a forum for potentially illegal or illicit activity, shift the risk from administrators to the user. Whether or not participants actually have a good grasp of the respective laws is a question for another paper.
We reiterated that this research does not require us to interact directly with participants, nor does it contribute to any aspect of risk that participants do not already face through their online behaviour. Our purpose is to observe these private groups as online marketplaces for selling human remains. Because it would otherwise be impossible to do this research if we were proactively open and forthcoming about this purpose, we do not wish to reveal this intention to administrators if we do not have to. Due to there being so many of these groups online, we responded that we would limit our research to the first four groups which don't require full disclosure of our reasons for joining.
The next major question wanted further details about how we might respond to some of the gatekeeping questions. We had indicated in the proposal that if a question was posed that would require us to lie in response (e.g. 'This group is for Europeans. Please indicate which European country you live in'; the answer, 'Canada', would be a lie), we would not join or study that particular group. We responded that the questions asked by the different groups can vary greatly. Our intention was to limit the full disclosure of our research by opting to join groups where only partial disclosure is required. Some only require partial disclosure through arbitrary questions, while others require full disclosure and personalized questions. Because there are scores of these groups on Facebook, we argued that this does not limit the range of our study. We believe that full disclosure would make it impossible for this research to be carried out, as per TCPS Article 3.6, as we have seen such issues pushing vendors and groups further underground.
There was a question about how we might deal with posts that include links to external websites or other groups. We responded that we would not preserve or share those links. Instead we would include a note in our data that would describe the existence and thematic content of the link (a summary of it) but not the links themselves.
The final point that the board required clarification on was how data summarization and double-blind anonymization would be carried out. We elaborated by saying, because this study is focused on qualitative analysis, we are interested in the themes that emerge about online discourse around the sale of human remains in 'oddities' spaces. We are interested in the general content of the transactions (advertised values, specimen types, object histories or taphonomic features), as well as interactions between vendors and buyers:
"The information to be recorded will be the content of posts - the text of posts, images and any discussion threads. These will be collected by a first researcher or Research Assistant, who will encode identifying information.
The content of the posts will then be transcribed by a second researcher or Research Assistant into individual textual records of each post and any relevant activity, encoded with a second identifier. These will not include quotes, but rather paraphrases or summaries of the general topics of discussion, questions asked or data.
Our research will analyze these textual records for overall themes or trends. The resulting data will be a spreadsheet where each textual record has a unique identifier, a summary of the discussion, and any other relevant aspects of the transaction, such as advertised price or taphonomic features."
The identifiable information we refer to is the username attached to each individual's Facebook page, the date the post was made, and other optional information such as location, which we will remove through the encoding process. In any publications, we will use codes to refer to the platform/group (e.g. Group 1, Group 2).
To draw it all together, the protocol that has emerged requires a division of labour and knowledge amongst the researchers that could be operationalized like this:
In this way, we are able to study the discourses within these groups while respecting their privacy.
We returned our responses. The board then deliberated and, satisfied that we were following the appropriate methods in concert with the guidelines laid down by SSHRC, approved our study. Humanities and social science researchers may find the process of going through ethics clearance to be intimidating or confusing at times, but it is essential to understand that working with humans necessitates a concern for how our research will affect our participants. Especially considering how our discipline has not always treated interlocutors with respect (e.g. Hicks 2020), taking time to evaluate the research against important ethical issues ensures that the benefits of doing this research truly outweigh the potential for harm.
Social media is an ever-shifting landscape where no single approach for understanding or exploring it will fit every circumstance. The platforms themselves, their affordances and customs, have an effect on how people behave on them and the kinds of things that they are able to discuss. Without the supercharging effects of social media (Huffer and Graham 2023, 157), the trade in human remains would not be as popular or as visible as it is. The protocol that we have evolved is suitable for now but not for always. By sharing how the process works in our institution, and how our eventual protocol was formed in concert with guidance and critical questioning from our ethics board, we trust that this example will help other researchers in similar positions.
We are grateful to the comments and feedback from the anonymous reviewers. Errors remain our own.
This work was approved by the Carleton University Research Ethics Board A, Project #120176. The authors acknowledge that no conflict of interest exists.
This work is supported by a Canada Research Grant from the Social Sciences and Humanities Research Council.
Internet Archaeology is an open access journal based in the Department of Archaeology, University of York. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.
Terms and Conditions | Legal Statements | Privacy Policy | Cookies Policy | Citing Internet Archaeology
Internet Archaeology content is preserved for the long term with the Archaeology Data Service. Help sustain and support open access publication by donating to our Open Access Archaeology Fund.
Home