Research Articles

Do Not Harm in Private Chat Apps: Ethical Issues for Research on and with WhatsApp

Authors: Sérgio Barbosa (University of Coimbra) , Stefania Milan (University of Amsterdam)

  • Do Not Harm in Private Chat Apps: Ethical Issues for Research on and with WhatsApp

    Research Articles

    Do Not Harm in Private Chat Apps: Ethical Issues for Research on and with WhatsApp

    Authors: ,


WhatsApp has remained under the radar for it is scarcely accessible to overt scholarly scrutiny. Encrypted chat apps allow for a certain degree of perceived secrecy. Yet the high frequency of civic engagement makes ethnographic research a time-consuming exercise. This article investigates how digital ethnography inside WhatsApp groups requires up-to-date, innovative ethical guidelines. We suggest a two-pronged approach. On the one hand, we should rethink and update ‘known’ ways of doing ethics, undertaking at least three conceptual operations: going back to the basics, positing as central the notion of ‘do not harm’, which allows to re-centre the user within the research process; avoid reducing research ethics to a one-stop checklist, to privilege instead a recursive, iterative and dialogic process able to engage research subjects; moving past the consent form as the sole and merely regulatory moment of the researcher-research subject relationship. On the other hand, while thinking through innovative ways of considering ethics in chat app research, we ought to take infrastructure seriously, both the site of research and the research ecosystem; embrace transparency and avoid by all means covert bypasses; and guarantee full anonymisation to our research subjects.

Keywords: engaged research, research ethics, digital ethnography, digital activism, Telegram, Signal, WhatsApp, chat apps

How to Cite:

Barbosa, S. & Milan, S., (2019) “Do Not Harm in Private Chat Apps: Ethical Issues for Research on and with WhatsApp”, Westminster Papers in Communication and Culture 14(1), 49–65. doi:




Published on
14 Aug 2019
Peer Reviewed

Introduction: Into the goldmine

October 2017, Brazil: A report by journalist Bruno Abbud of the cultural magazine Piauí exposed private conversations in a WhatsApp chat of a right-wing group called Movimento Brasil Livre (see Messenberg, 2017). Among other things, the group discussed the behind-the-scenes meetings of their members with national politicians and leaders, now accessible to public purview in a total of over 685 pages of transcripts dealing with health, public policy, and education, among others (Abbud, 2017).1 This is just one of the many instances in which trust between participants in chat apps of a political nature is misused, no matter what the reasons. More and more scholars are devoting their attention to political activism facilitated by real-time chat applications like Whatsapp and Telegram (see, among others, Barbosa, 2017; Alimardani and Milan, 2018; Potnis, Demissie, Trimmer and Cleek, 2018; Agur and Frisch, 2019). We can safely argue that messaging apps are today one of the primary sites for media activism to emerge and unfold (see, e.g. Pickard and Yang, 2017). However, and despite the fact they represent a real goldmine for scholars, messaging apps have long remained under the radar of scholars of social movements and digital activism, largely because of the limited permeability of chat groups to overt and covert observation alike. The many ‘closed’ groups mobilizing on chat apps are typically difficult to penetrate; their existence is often unknown to scholars. In addition, the high frequency of exchanges makes digital ethnography a 24/7 exercise. Nonetheless, the popularity of chat apps for political debate and the coordination of collective action makes them a fruitful venue for research – and more and more scholars are taking advantage of this amazing opportunity. But, as the opening example shows, things might be more complicated than they appear at first sight. The recent introduction of end-to-end encryption to chat apps, then, risks giving activists a false sense of security that encourages individuals and groups to use these channels also for sensitive exchanges, thereby exposing participants to potential risks that researchers might indirectly contribute to amplifying. It is precisely this centrality of chat apps in contemporary activism and the perceived secrecy surrounding the space that studying interactions in what we can imagine as a ‘(private) public sphere’ (Myers West, 2017) requires carefully-negotiated ethical commitments.

This article emerges out of a common interest of the authors, sparked by a conversation that followed the presentation of Barbosa’s study of #UnidosContraOGolpe (UCG, United Against the Coup). This mobilisation, which emerged in Brazil in the aftermath of the impeachment of then-President Dilma Rousseff (2016), was facilitated by a private group on WhatsApp. As we will explain, Barbosa embarked on a digital ethnography of group interaction. Milan, a long-term student of the relation between movements and technology (2013), was intrigued by this new genre of media activism, which expands the repertoire of digital activism and encapsulates the political potential of chats apps as a civic engagement tool. In particular, she wondered about the potential ethical ramifications of doing research on chat apps, and private groups in general – and the two researchers started to exchange ideas. What follows is the result of this ongoing exchange and intends to stimulate a much-needed collective reflection on the subject matter. What are the limits of (c)overt participant observation in this emerging ‘(private) public sphere’? How can we safeguard the ethics of research, protecting user privacy and respecting the intimacy of their exchanges, while simultaneously taking advantage of such a rich data source? As we shall see, the nature of the medium in question, and of the interactions between group participants lend themselves particularly well to covert observation, in ways that ‘traditional’ observation in ‘real’ life typically does not. Even when one discloses their researcher identity, the fluidity of belonging and participation in a chat app creates a situation that it is not really possible to inform every group participant about the ongoing observation, let alone remind each of it; ensuring ‘informed’ consent is probably just wishful thinking in very large groups. Deceiving participants is thus a concrete risk and a tempting possibility. What’s more, the informality of the space and the confidentiality of conversations, mediated by an intimate technology like the smartphone, exposes ordinary users to more risks that they might themselves imagine. It is a slippery slope that researchers can no longer ignore.

The article investigates the methodological and ethical challenges of conducting research of an ethnographic nature on WhatsApp. Considering chat apps as simultaneously a social phenomenon and a tool with specific affordances for political participation as well as for research, we reflect on the ethics of qualitative data collection when the field site is an encrypted chat app such as WhatsApp, Telegram and Signal. In particular, we focus on the possibilities of digital ethnography in private groups and offer a viable approach to thinking through the ethics of doing research in this digital space. The article builds on and brings into dialogue findings from two research projects. As mentioned above, Barbosa has investigated interactions on the #UnidosContraOGolpe private group on Whatsapp (2017), while Milan has been collecting data on the evolution of political participation and social media usage amongst grassroots activists in various countries since 2012, as part of an ongoing project on ‘cloud protesting’ (see, inter alia, S. Milan, 2015, 2018 and on ‘data activism’ Milan, 2017).

The paper is composed of four parts. Firstly, we present what is at stake for researchers in encrypted chat apps, focusing on the diffusion of WhatsApp and its siblings worldwide and their role in digital and media activism today, and showing, data in hand, why this represents a great research opportunity. Secondly, we explore the case study that triggered the conversation and inspired our joint interrogation into the ethics of research on chat apps. Thirdly, we reflect on digital ethnography as an adaptive method to study interactions in chat apps, reflecting in particular on the risks that come with it. We then offer our considerations for thinking and practising ethics in chat app research, building also on existing literature on research with politically active subjects.

The blast of WhatsApp and its siblings

Why should we care about Whatsapp and other similar platforms in the study of media activism? WhatsApp is a cross-platform messaging app allowing users to exchange messages over a phone’s data traffic without paying extra for short-text messaging. Launched in 2009, it turns ten year old in 2019 and is currently the world’s fastest growing platform. It became a global phenomenon in 2012; Facebook bought it for USD 19 billion in 2014. Over the years, its functionalities were updated several times. Today a WhatsApp user can share pictures, send files, and call other users, including video and/or voice and group calls. She can also engage in the very popular ‘private group chats’ that are analysed in this article.

By the end of 2017, WhatsApp counted 1.5 billion active monthly users (Garimella and Tyson, 2018). It is the preferred app in more than 100 countries around the world (Constine, 2018), with high penetration rates in countries as diverse as India, Indonesia, Malaysia, Mexico, Brazil and South Africa (Sevitt, 2017). According to a research by the company We Are Social, Whatsapp is the top social messenger app in Central and Latin America, Canada, most of Africa and Europe, and Russia (Iqbal, 2019). Whatsapp is pursued by Facebook Messenger with 1.3 million active users, the Chinese multi-purpose app WeChat and Tencent QQ (respectively 1 million and 803,000 users), followed by Skype, Snapchat, Viber, LINE and Telegram, with a mere 200,000 users (Statista, 2019) – in other words, the large majority of smartphone holders have installed one or more of these applications.

In recent years, several of these services have adopted end-to-end encryption, in response to public concerns about state snooping as uncovered by the Snowden revelations (for an overview, see Greenwald, 2014). End-to-end encryption (the end indicates the ‘endpoint’, meaning the client device) prevents the hosting server or any intercepting third party from accessing the content of the message (Ermoshina, Musiani, and Halpin, 2016).2 Following the example of, among others, Signal – an open-source, encrypted messaging service spearheaded in 2014 by Open Whisper Systems (now the non-profit Signal Foundation) – WhatsApp has implemented end-to-end encryption in April 2016 – although not without trouble, as reported by Santos and Faure (2018). Technically, it means that messages remain private. Yet, WhatsApp’s end-to-end encryption does not necessarily imply that Facebook is not in a position to collect WhatsApp data (Zanon, 2018), in what has been called ‘a metadata imbroglio’ (Brewster, 2017), whereby some data – such as a user’s phone number, contact list, and usage data – are shared with the ‘mother’ company Facebook.3 Table 1 summarises the main features of the most popular messaging apps.

Table 1

Differences between three chat apps frequently used by activists and ordinary users.

WhatsApp (
  • Chat app created by Jan Koum and Brian Acton (ex Yahoo! employees) in 2009.

  • Self-defined as ‘simple, personal, real-time messaging app’.

  • Since 2014 integrated with Facebook (mother company).

  • One of its key features is the group chat in groups, which can be ‘private’ (by invitation, run by administrators) or ‘public’ (joined via a link). Users can send messages to groups with up to 256 people.

  • Implements end-to-end encryption since 2016.

  • Its desktop version works in conjunction with the mobile app.

  • High penetration in the Global South.

Telegram (
  • Chat app created by the Russian Durov brothers (owners of the popular social network VK). Given its functionalities, it is the main direct competitor to WhatsApp on a global scale.

  • It supports group chats to up to 100,000 people.

  • Implements end-to-end encryption. It is popular amongst politically active citizens, and in countries like, e.g. Iran.

  • Telegram has a limit of 1.5 GB for the size of file you can share on the service (better than WhatsApp and Signal).

  • The desktop/web versions are independent of the mobile app.

Signal (
  • Open-source project, it is the favourite amongst digital rights activists. It, too, supports one-to-one voice and video calls.

  • Since its inception, it runs end-to-end encryption and supports mechanisms by which users can independently verify the integrity of communications channels and of the identity of their interlocutors.

  • Developed/maintained by a non-profit entity (now Signal Foundation), it has received funding from, e.g., the Open Technology Fund.

  • The desktop version works in conjunction with the phone app.

Chat apps play a key role in contemporary activism, including media activism (see, e.g. Custódio, 2017; F. L. F. Lee and Chan, 2016; Caetano et al., 2018; Treré, 2018; Akbari and Gabdulhakov, 2019). Despite it being frequently considered a flywheel of fake news, especially in countries of the so-called Global South (Burgos, 2019; cf. Marda and Milan, 2018), WhatsApp has established itself as a powerful political tool for spreading political information (Montag et al., 2015). In Brazil, for example, it has been central in the organisation of the 2018 truck drivers’ strike, considered the largest in the history of the country (Fox, 2018). But it has also acted as a ‘weaponized’ chat app for the dissemination of misinformation during the 2018 Brazilian Presidential elections (Pereira and Bojczuk, 2018; see also Nemer, 2018; Córdova and Barbosa, 2019). Despite these potentially contradictory manifestations, it is increasingly popular among ordinary users in Brazil (Spyer, 2017) as well as politically-active individuals. Little research however has focused on Whatsapp as such.4 Next, we provide an example of how ethical considerations played out in studying interactions on Whatsapp in the absence of codified ethical guidelines.

#UnidosContraOGolpe: The challenge of studying social interactions in a private chat

Brazil is one of the forerunners as far as WhatsApp penetration is concerned, with 120 million active users (Digital Information World, 2019). As reported by the WhatsApp Nation study (2016), 76 per cent of messaging app users regularly use ‘Zap Zap’, as Brazilians call Whatsapp (Vieira, 2018). According to the 2017 Digital News Report, 46 per cent of Brazilian citizens use WhatsApp to find news (Reuters Institute, 2017). With such a prominent role in a country that is rightly considered one of the cradles of media activism in Latin America, Barbosa set out to observe how the app enabled activism during a difficult moment in the recent history of the country: the controversial impeachment of its president.

#UnidosContraOGolpe (UCG) was a group of leftist activists and concerned citizens created on WhatsApp to oppose the impeachment of Rousseff. In fact, the impeachment was perceived by many Brazilians as an unwarranted soft ‘coup’ (Souza, 2016; Weber, Bastone and Barbosa, 2018). Participants advocated the return of Rousseff to the Presidency of the Republic. The individuals enrolled in UCG, a ‘private’ group with ‘by invitation only’ membership, expressed indignation with the political situation and discussed potential actions and alternative scenarios. The confrontation between diverse identities in the group allowed civic engagement to flourish in this space. Most users were also active protesting in the streets, but within WhatsApp they created a new style of political participation able to connect the square and the digital sphere. The diverse opinions among users grew alongside with interaction on the chat, resulting in a novel, shared collective voice – as observed in earlier forms of media activism (Hackett and Carroll, 2006; Meikle, 2002; S. Milan, 2013, 2017).

How did UCG come about? The group was created on 30 March 2016 by a concerned citizen from Florianópolis, in the South of the country, around the time in which the Parliament Lower Chamber was called to vote on the request of impeachment against Rousseff.5 Barbosa became a member of the group and performed an extensive digital ethnography, logging information daily in order to investigate member interaction. At the time of joining, the group counted 256 registered users (the maximum allowed by the service), with the exact number varying from day to day; interestingly, the vast majority of members were listed also as ‘administrators’, which allowed them to add new members or exclude existing ones, signalling the absence of identified leaders.6 However, guidance with respect to the ethical implications of a research approach of this kind was not readily available locally to our researcher. To begin with, the hosting institution did not require an ethical clearance – as it is often the case in countries outside of the Anglo-Saxon context. Available ethical guides and indications (e.g., Association of Internet Researchers, 2002, 2016; Chesters, 2012; S. Milan, 2014) did not yet provide ready-to-use insights on the specific app, as obviously they do not have the capacity to keep up with the pace of all tech innovations and user adoption. In the lack of codified codes of conducts, Barbosa embarked in a process of moral decision-making.

The first challenge was of methodological nature: how to develop a creative approach to digital ethnography that did not harm or interfere with the interactions among chat members? Upon asking to join the private group in April 2016 and being immediately admitted by the administrators (those controlling the ‘gates’ of the channel), Barbosa posted a message to inform group members of the research and to announce that some of them might be invited to an interview at a later stage. Registering no opposition, he considered this a ‘green light’. Thanks to his active participation in group discussions he began building trust, which, like admittance, was facilitated by prior acquaintance with some of the UCG group members. The researcher-participant was thus quickly integrated into the group, and soon considered a ‘regular’ member. It is worth noting, however, that he was well aware that with chat apps, consensus cannot be assumed as a once-off ocurrence, as one cannot guarantee that every user (including future members!) has consented to the research, on account of the fluid aspect of interactions on WhatsApp (e.g. engagement can be intermittent, and not all users read all messages; new users can be added, as well leave a group). He did take care to send occasional reminders about his research, paying attention not to disrupt the conversation. Interestingly, amongst the informal commitments he made to the group members upon joining and disclosing his identity of researcher, was the promise to be accountable with respect to public presentations of the research results. Over time, group members felt he acted as their ‘voice’ and ‘representative’ within academia, and welcomed with enthusiasm information about, for example, public presentations of the research.

Digital ethnography allowed the researcher to observe from up close the interpretations, practices and everyday experiences that foster dynamic realities inside chat apps when it is conceived as a connection point between the online and the offline sphere of action (Boyd, 2008), or in a permanent transit between these two dimensions (Máximo, 2007). He could observe identity formation at the group level (the meso level of mobilisation), and, with the help of semi-structured interviews, explore the micro level of meaning-making, too. Broadcasting indignation and solidarity, the bottom-up character of UCG political action inaugurated a unique way for individuals to share stories, interacting with each other, and connecting locally in this novel online-offline civic engagement space. Exploring whether the ‘WhatsAppers’ represent a new form of political participation able to generate social change appropriating the affordances of the chat app, the research exposed how the UCG group was united under a leftist project formed by several social actors looking for a collective experience. He could identify the dynamics of discussion and coordination, while making a point about the validity of investigating private chats, instead of merely focusing on public groups (Garimella and Tyson, 2018; Resende et al., 2018). Users with similar political views came together but also recognised the right of other WhatsAppers with distinct opinions to present their views on the political situation. The group thus hosted also dissent and discussion, but within a basis of mutual respect. It evolved contingently over time, in response to the political situation it addressed, expressing demands, frustrations and projects to unite against the rise of the extreme right-wing faction and the beginning of Temer’s (2016-18) government. Although similar types of ‘connective action’ (Bennett and Segerberg, 2012) are frequently associated with fragmentation and atomisation, the research showed how private group chats on WhatsApp did help to foster creative interactions through high-speed engagement. It also showed dynamics of this nature can facilitate the emergence of a collective identity based on lower-common-denominators yet nonetheless useful to the cause (S. Milan, 2015). These identity elements, seemingly individualised and fluid, can be facilitated by active group administrators in WhatsApp acting as ‘soft’ leaders or ‘choreographers’ (Gerbaudo, 2012).

Digital ethnography in chat apps: Methodological notes for an adaptive method

Berry and Fagerjord (2017) remind us that each digital interface calls for appropriate research methods. Adaptation of codified methods is a good point to start when approaching ‘new’ research sites such as chat apps. Digital (or virtual, cf. Hine, 2002) ethnography points to ‘iterative-inductive research’ that acknowledging ‘the researcher’s own role’ and ‘humans as part object/part subject’ allows researchers to ‘produc[e] a richly written account that respects the irreducibility of human experience’ (O’Reilly, 2005, 3). ‘As new technologies offer new ways of engaging with emergent research environments’ (Pink, Postill amd Horst, 2016, 8), we ought, however, to carefully interrogate how ethnographic practices shift as we approach chat apps as a source of data. Ethnography is particularly suitable to approach chat apps thanks to its adaptive nature (Hine, 2015, 192), able to respond to the distributed nature of field sites and/or analytical units.

Kozinets (2014) identified three types of (virtual) ethnography, conducted respectively on archived data, extracted data and field note data. If we adapt this classification to data extracted from chat groups in messaging apps, we have the following typology. The first type consists of copying directly from computer-mediated communications data from the page or observed group, importing text, photographs, and sound files – in other words, data for which the researchers are not involved in the creation. This one is very hard to do within chat apps, at least not in their entirety. The second approach refers to data that the researcher creates through interaction with group members, such as data collected through interviews by e-mail, chat, instant messages, etc. The third type refers to the field notes by the researcher, concerning for example the communication practices of chat members, their interactions, as well as the researcher’s own participation and sense of belonging.

Interactions on chat apps are, paraphrasing Hine, ‘embedded, embodied and everyday’ (2015, 19). They are ‘embedded’ because they tie online activities within other aspects of people’s lives. Their ‘embodied’ character is responsible for the composition of identity from the social relations created in virtual networks, whereas ‘everyday’ refers to the ordinary use of tools that have become part of the basic infrastructure of daily life. Chat app groups make an excellent case of the ‘everyday’ internet, as they have become part of the intimate sphere of individuals. From the methodological point of view, the real-time-ness of interactions in this space means that extracted data comes in large amounts. Field note data then become of paramount importance in the transformation of raw data, such as chat files extracted from the group scroll bar, chat backup, screenshots, and interview transcripts. Table 2 summarises the three types as they apply to chat apps.

Table 2

Three types of virtual ethnography adapted to chat groups. Source: Authors’ adaptation of Kozinets (2014).

Archived Data Full copy of the data used for the investigation, without direct involvement of the researcher.
Extracted Data Data extracted from chat groups, originated also through the interaction of the researcher with other participants.
Field Note Data Field notes on communication practices within chat groups.

What makes digital ethnography on chat apps simultaneously so special and challenging? Digital ethnography is particularly suitable for the study of contemporary forms of digital activism, and media activism in particular, as it unfolds on chat apps, for at least three reasons. Firstly, it allows the researcher to get access to mapping and describing the online-offline interaction presented in the backstage of these chats (Lee and Gregory, 2008), considered as the expression of the personal and everyday experience of users (Máximo, 2007). In other words, it empowers the researcher to develop what we may call a ‘thick’ account of group interactions.7 Secondly, it makes possible the shortening of the distance between time and space, due to the very same dynamics of the chat app’s architecture, in which multiple social groups and relationships are continuously (re-) arranged in high-speed interactions (see also Amar, 2011) and occasionally organised in a ‘closed’ structure (private groups). Thirdly, it allows the researcher to play with the online/offline dichotomy, in two ways. On the one hand, WhatsApp groups can be the origin of both political actions in the online space (chat), but also on the ground – such as demonstrations or petitions. On the other hand, the researcher can combine virtual interaction with group members with interviews, as observing user interactions from ‘up close’ facilitates the identification of interesting research subjects to talk to.

Towards an ethics checklist for research in (private) chat apps

Given the intimate nature of chat app interactions and the ‘false’ perception of secrecy that encryption brings about, ethical considerations are core prerequisites for engaging in digital ethnography in a way that is respectful of human subjects, especially when they are vulnerable individuals, as it is often the case with political activists.8 The field is inevitably slow at adjusting research ethics to upcoming ‘digital’ challenges. Yet such a research site and these methods are particularly prone to deception and misconduct. In this section we depart from the experience with UCG, to review existing reflections and provisions on research ethics and apply them to the field site of chat apps. In so doing, we aim at starting a much-needed conversation about the need and practice of research ethics in chat apps research, with a special concern for research on digital and/or media activism facilitated by chat app (and private/closed groups in particular). Our building blocks include documents prepared by professional organisations (e.g., Association of Internet Researchers, 2002, 2016), specialised groups (Ethics Feedback Panel for Networking and Security, 2014; Utrecht Data School, 2017), and existing contributions in the fields of media studies, humanities and the social sciences (e.g. Flacks, 2005; Pimienta, 2005; Chesters, 2012; Ess, 2013; S. Milan, 2014; Floridi and Taddeo, 2016; C. Milan and Milan, 2016; Zimmer and Kinder-Kurlanda, 2017; Ong, 2019). We also take into account the evolving legislative backdrop (e.g., the General Data Protection Regulation, which came into force in the European Union in May 2018 that has rapidly – even if questionably – become the new ‘gold standard’ of data protection worldwide) as well as the measures adopted by the industry following questionable data reuse practices such as those brought to light by the Cambridge Analytica scandal (see, e.g. Ward, 2018).

Generally speaking, researchers of all disciplines can prefer to avoid going through an ethics committee procedure, as these processes are tedious and complex, and often have the unwanted result of slowing down the research. What’s more, ethical considerations are only seldom made explicit in published manuscripts. To name just one, the field of social movement studies – one of the key disciplines devoted to the study of activism – has only seldom engaged openly with questions of ethics. This leaves a lot of colleagues thirsty for example and best practices.

In earlier works (2014), Milan identified four questions that pertain to the study of political activism and dissent, namely: relevance, which has to do with the nature of the researcher/research subject relationship, and asks what knowledge should be produced and for whom (see also Croteau, 2005; Flacks, 2005; Ryan, Salas-Wright, Anastario and Camara, 2010); risk, which interrogates the potential harms a research activity can entail for activists; power, which concerns the unbalanced relationship between research subjects and researchers (cf. C. Ryan and Jeffreys, 2008), and accountability towards research subjects, which includes also knowledge sharing. Milan and Milan later expanded the list into the STRAP framework, acronym for Sharing, Translation, Relevance, Accountability, and Power, designed to encourage researchers to engage communities on the ground as active agents of knowledge production (C. Milan and Milan, 2016). We now take these reflections one step forward to start addressing the lack of benchmarks and criteria for research on digital and/or media activism mediated by chat apps.

Rethinking (and updating) ‘known’ ways of doing ethics

Do not harm

The UCG case shows that persistent relation-building is fundamental to foster healthy research relationships. To encourage more ethical and responsible research practices and reduce the complexity and the abstrusity of current ethics processes, it might be useful to ‘go back to the basics’, revisiting the ‘old’ hacker imperatives of ‘do not harm’, ‘leave no damage’ and ‘leave things as you found them (or better)’ (Levy, 1984). These simple rules can help researchers to (re-)centre the research subjects and their environment and agendas, placing their well-being before academic goals of any kind and before any reflection on affordances of the technology. To be more than an evocative slogan, however, ‘do not harm’ calls for appropriate contextualisation in relation to specific social groups and research questions, as well as technological configurations (see also Cronin-Furman and Lake, 2018).

Beyond the checklist approach – towards ethics as a recursive process

As evinced in the UCG example, the ‘one-off’ approach is not sufficient in chat apps. Nor are the current mechanisms, whereby ethics takes the form of (lengthy, over-detailed, tedious) checklists, which are shelved away and usually forgotten the minute after a project receives the green light from the ethics board. These checklists reduce the process of reflecting on ethics to a one-off mechanism useful to ‘clear one’s conscience’ and results too static and unresponsive, unable to follow the quick pace of technological evolution. In other words, it risks being merely a ‘purr word’, ‘a word that sounds nice (like a cat purring) and conveys pleasant connotative thoughts, but a word that has virtually lost its substantive denotative meaning because of the many different conceptions that people have of it’ (Underwood and Frey, 2008, 371). Thinking about and implementing ethics in our research practices should instead be a recursive, iterative and dialogic process that engages the researchers and the research subjects in a ‘spiral’ conversation, as evoked also by ethnography more in general. In this respect, it is useful to take a look at the Data Ethics Decision Aid (DEDA) prepared by the Utrecht Data School at the University of Utrecht, the Netherlands, originally developed for public administrations and designed to foster a deliberative discussion within a team.9 The associated worksheet deliberately avoids the checklist format, to present decision-makers with a spiral. It is intended to be used recursively during the course of a given project. Similarly, its offspring DEDA for Research is an ‘online survey that asks the researcher a series of open questions. There are no necessarily right or wrong answers, the questions merely function to raise awareness of certain issues and help the researcher document their decision making process’ (Utrecht Data School, 2017). It includes questions about methods, data management, laws and codes of conducts but also visualisation and the responsibility towards the research community and society more in general. Although it does not concern chat apps, DEDA offers an excellent way to engage with ethics in a recursive manner.

Beyond the consent form

As the UCG case made apparent, it is hard in chat apps to capture (and get the attention of) all members in a chat group for the entire duration of the project. The codified way of obtaining informed consent does not work either, with research subjects expected to sign a consent form as the sine-qua-non condition for participating in the research. This approach is entirely outdated in the age of chat apps, and potentially dangerous. On the one hand, collecting consent forms from the over 200 participants in a private group on Whatsapp is plainly unfeasible. But, more importantly, an individual cannot be expected to ‘sign off’ her behaviour for the lengthy period of her engagement in a chat group. How can we then make consent ‘future proof’? More in general, what does consent mean in the age of chat apps? Consent should be envisioned as an ongoing negotiation and exchange where consent can be modulated according to the evolution of the contingent situation, rather than a one-off stop, and where the process of ‘obtaining consent’ is a conversation starter and the main goal is informing and educating the user of the protections accorded when participating in a research. The research subject, rather than the ethics board, should be put at the centre of the process.

Three non-negotiable propositions looking forward

Infrastructure matters (also in our practices)

As the kind of research described here takes place in an ecosystem in continuous evolution, both as far as the technological and legal set up and user preferences and awareness are concerned, investigators must be aware of the main functionalities and affordances of their research site. They must keep abreast of their evolution, as these may be modified by the service provider in the course of the research and take care of informing their research subjects of the conditions under which the research takes place. But infrastructure matters also when it comes to data storage, analysis and sharing. Today universities no longer run their own servers and services, preferring to rely on commercial alternatives such as Microsoft and Google, whose facilities typically come with non-negotiable and blurry terms of service. A researcher should prefer, whenever possible, self-run infrastructure able to protect user data.

A transparent research agenda (and methodology)

A certain degree of ‘secrecy’ typically surrounds a researcher’s plans. On the one hand, researchers have been trained to disclose their questions and hypotheses only partially (if at all), to avoid, e.g. socially desirable responding. On the other hand, ‘covert’ data collection has traditionally been presented as a valid alternative able to overcome the resistances of certain communities, or to facilitate the ‘blending’ of the researcher into said communities (see, e.g. Zelenkauskaite and Bucy, 2016). Although there is no ‘one size fits all’ when it comes to these matters, we believe that, whenever suitable and/or possible, a transparent research agenda is key to safeguard the trust in research in the age of social media, and chat apps in particular. It is also a key step in the implementation of an ‘engaged’ approach to research, whereby the needs and values of the research subjects become an integral part of the research agenda (Milan, 2010).

Full anonymisation for future-proofing research

Pseudo-anonymisation – or the de-identification procedure by which personally identifiable information fields are replaced by, e.g. pseudonyms – is usually presented as a safe approach to data protection. However, especially when dealing with vulnerable subjects such as politically active individuals, this measure is not always sufficient to protect the identity of users on, e.g. a chat group – let alone social media. We argue that users must be safeguarded by implementing full anonymisation, e.g. avoid direct quoting and preferring paraphrasis, even when it might seem redundant because of end-to-end encryption. Encryption, however, is not future-proof, as quantum computing might eventually make accessible what is today hidden.10


As the UCG case study of #UnidosContraOGolpe shows, activism today is increasingly facilitated by chat apps. New technological affordances, new practices of civic engagement, and new forms of surveillance and risks require novel approaches to research ethics. This article started from reflecting on the challenges of doing digital ethnography in a private chat group, to offer a number of pathways to help researchers to think though the limits of their research. To be sure, we do not want researchers to shy away from studying private chat groups. Rather, we intend to encourage academia to build strong practices of individual scholarship and community praxis in relation to research on chat apps, and digital ethnography in particular. We acknowledge ethical considerations by the research community to be a step forward from and a corrective measure against moral considerations by individuals. They are of paramount importance in fostering a healthy relationship between research subjects, especially when politically active, and researchers exploiting the affordances of digital technology to research ends. We believe that we cannot comfortably research media activism today without considering ethical challenges as well as the risk of surveillance and repression of activists on the ground.

We suggested a two-pronged approach. On the one hand, we should rethink and update ‘known’ ways of doing ethics, undertaking at least three conceptual operations: going back to the basics, positing as central the notion of ‘do not harm’, which allows to re-centre the research subject at the heart of the research process; avoid reducing research ethics to a one-stop checklist, to privilege instead a recursive, iterative and dialogic process able to engage research subjects; moving past the consent form as the sole and mere regulatory moment of the researcher-research subject relationship. On the other hand, while thinking through innovative ways of thinking ethics in chat app research, we ought to take infrastructure seriously, both the research site and the research ecosystem; embrace transparency and when the research question allows covert methods, avoid dishonest bypasses; and guarantee full anonymisation to research subjects to in order to future-proof findings and publications.

Far from providing a fully-fledged research agenda, this article hopes to serve as a way of kicking-off an urgent, collective conversation that can highlight ethical commitments to protect everyday citizens that appropriate chat apps for digital activism. Only in this way can academia implement best practices in exploring creative techniques to investigate media activism in chat apps today.


  1. The report failed to mention whether group participants were warned of the presence of an outsider spying on their conversations. We can reasonably assume the contrary, otherwise the author might have been prevented access.
  2. Resorting to encryption as a defence mechanism (Gürses, Kundnani and van Hoboken, 2016) is correlated with a negative view of the state as adversary of online privacy (Hellegren, 2017) – something that however, the UCG activists were not particularly inclined to at first. With the advent to power of the controversial President Jair Bolsonaro, users are however considering shifting to Telegram for fear that Bolsonaro supporters will continue to spread fake news on WhatsApp. But the (however, slow) migration to Telegram already started in 2015 on the occasion of a WhatsApp blackout ordered by the courts (Savov, 2015).
  3. The quasi-entirety of users involved in the case study, however, seem to ignore this ‘relational risk’ (Das and Teng, 2004) – or the probability that the service provider might break their trust and leak data from their conversations.
  4. For research on Whatsapp, although not in direct connection with activism, see Ling and Lai (2016), Waterloo et al. (2018), and Matassi et al. (2019).
  5. Not all users were from Florianópolis. Through interpersonal networks and in combination with protest on the ground, the mobilisation spread to a national level, including activists from the Northeast. Interestingly, some of these travelled to Florianópolis to meet in person.
  6. At the time of writing this article, membership had gone down to 174 participants. The group continued as a form of resistance to the advent to power of far right-wing president Jair Bolsonaro.
  7. About thick data see Blee and Taylor (2002).
  8. For a reflection on activists as vulnerable individuals see Kazansky and Milan (2019).
  9. See:
  10. For a layman’s account, see The Economist (2018).


Barbosa acknowledges all the UCG activists. Without them, this research would not have been possible, especially the fifteen users who agreed to be interviewed.

Competing Interests

This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No 639379 – DATACTIVE, awarded to Stefania Milan as Principal Investigator;

Author Information

Sérgio Barbosa is a PhD student in the program ‘Democracy in the Twenty-First Century’ linked to the Centre for Social Studies (CES) at the University of Coimbra, Portugal. In the academic year 2018-19, he has been an Erasmus+ exchange student at the University of Amsterdam, under the supervision of Dr Stefania Milan. His research explores emerging forms of political participation and resistance vis-à-vis the possibilities afforded by chat apps, with an emphasis on WhatsApp. Sérgio is a research associate of the DATACTIVE project and an active member of the Big Data from the South Initiative. He is also a member of the COST Action CA17135 ‘Constitution-Making and Deliberative Democracy’, and of Tecnopolitica, a ‘Latin’ research network connecting Brazil and Ecuador with Spain, Portugal and Italy. Contact:

Dr Stefania Milan (, is a scholar of social movements specialising in new forms of political participation facilitated by technological innovation. She is Associate Professor in New Media and Digital Culture at the University of Amsterdam, and the Principal Investigator of two projects financed by the European Research Council exploring data- and algorithmic-mediated forms of civic engagement (see and Passionate about building bridges between research, activism and policymaking, Stefania has been researching grassroots engagement with communication technologies since 2003. Her research has furthered the development of interdisciplinary theory in the realm of networked collective action, including on the ethics of research with vulnerable activists. She is the author of, among others, Social Movements and their Technologies: Wiring Social Change (Palgrave Macmillan, 2013/2016) and an editorial board member of the book series Critical Digital and Social Media Studies [University of Westminster Press]. Contact:


1 Abbud, B. (2017, 3 October). O grupo da mão invisível. Dois meses no WhatsApp do MBL. Retrieved from Piauí website:

2 Agur, C., & Frisch, N. (2019). Digital disobedience and the limits of persuasion: Social media activism in Hong Kong’s 2014 Umbrella Movement. Social Media + Society, 5(1). DOI:

3 Akbari, A., & Gabdulhakov, R. (2019). Platform surveillance and resistance in Iran and Russia: The case of Telegram. Surveillance and Society, 17(1/2): Platform Surveillance. DOI:

4 Alimardani, M., & Milan, S. (2018). The internet as a global/local site of contestation: The case of Iran. In Peeren, E., Celikates, R., de Kloet, J., & Poell, T. (Eds.), Global Cultures of Contestation. Mobility, Sustainability, Aesthetics and Connectivity (pp. 171–92). Basingstoke: Palgrave MacMillan. DOI:

5 Amar, G. (2011). Homo Mobilis: La Nueva Era de La Movilidad. Buenos Aires: La Crujía Ediciones.

6 Association of Internet Researchers. (2002). Ethical Decision-Making and Internet Research: Recommendations from the AoIR Ethics Working Committee (Version 2.0). Retrieved from:

7 Association of Internet Researchers. (2016). Appendix 1 to AOIR Guidelines: Ethical Decision Making and Internet Research Ethics. Retrieved from

8 Barbosa, S. (2017). WhatsApp e política: novas formas de ciberativismo em Florianópolis (Master’s Thesis, Federal University of Santa Catarina). Retrieved from

9 Bennett, L. W., & Segerberg, A. (2012). The logic of connective action: Digital media and the personalization of contentious politics. Information, Communication and Society, 15(5): 739–68. DOI:

10 Berry, D. M., & Fagerjord, A. (2017). Digital Humanities: Knowledge and Critique in a Digital Age. Cambridge: Polity.

11 Blee, K. M., & Taylor, V. (2002). Semi-structured Interviewing in Social Movement Research. In Klandermans, B., & Staggenborg, S. (Eds.), Methods of Social Movement Research (pp. 92–117). Minneapolis, MN: University of Minnesota Press.

12 Boyd, D. (2008). A response to Christine Hine. Retrieved from

13 Brewster, T. (2017). Forget about backdoors, this is the data WhatsApp actually hands to cops. Forbes.

14 Burgos, P. (2019, 27 June). What 100,000 WhatsApp messages reveal about misinformation in Brazil. First Draft. Retrieved from

15 Caetano, J. A., et al. (2018). Characterizing the public perception of WhatsApp through the lens of media. Presented at the Turin. Retrieved from

16 Chesters, G. (2012). Social movements and the ethics of knowledge production. Social Movement Studies, 11(2): 145–160. DOI:

17 Constine, J. (2018). WhatsApp hits 1.5 billion monthly users. $19B? Not so bad. Tech Crunch.

18 Córdova, Y., & Barbosa, S. (2019). WhatsApp and the Brazilian elections: Weaponized chat app or civic engagement tool? Retrieved from

19 Cronin-Furman, K., & Lake, M. (2018). Ethics abroad: Fieldwork in fragile and violent contexts. American Political Science Association, 51(3): 607–14. DOI:

20 Croteau, D. (2005). Which side are you on? The tension between movement scholarship and activism. In Croteau, D., Hoynes, W., & Ryan, C. (Eds.), Rhyming Hope and History: Activists, Academics, and Social Movement Scholarship (pp. 20–40). Minneapolis, MN: University of Minnesota Press.

21 Custódio, L. (2017). Favela Media Activism: Counterpublics for Human Rights in Brazil. Lanham, MA: Lexington Books.

22 Das, T. K., & Teng, B. S. (2004). The risk-based view of trust: A conceptual framework. Journal of Business and Psychology, 19(1): 85–116. DOI:

23 Digital Information World. (2019). WhatsApp usage, revenue, market share and other statistics (2019). Retrieved from

24 Ermoshina, K., Musiani, F., & Halpin, H. (2016). End-to-end encrypted messaging protocols: An overview. In Bagnoli, F., et al. (Eds.) Internet Science: Third International Conference, INSCI 2016, Florence, Italy, September 12–14, 2016, Proceedings (pp. 244–54). Cham: Springer. DOI:

25 Ess, C. (2013). Digital Media Ethics. London: Polity.

26 Ethics Feedback Panel for Networking and Security. (2014). Retrieved from

27 Flacks, R. (2005). The Questions of Relevance in Social Movement Studies. In Croteau, D. Hoynes, W., & Ryan, C. (Eds.), Rhyming Hope and History: Activists, Academics, and Social Movement Scholarship (pp. 3–19). Minneapolis: University of Minnesota Press.

28 Floridi, L., & Taddeo, M. (2016). What is data ethics? Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Science, 374. DOI:

29 Fox, M. (2018, 15 June). The Brazilian truckers’ strike: How WhatsApp is changing the rules of the game. Truthout: Economy and labour. Retrieved from

30 Garimella, K., & Tyson, G. (2018). WhatsApp, doc? A first look at WhatsApp public group data. ICWSM 2018 Social and Information Networks (Cs.SI). Retrieved from

31 Gerbaudo, P. (2012). Tweets and the Streets Social Media and Contemporary Activism. London: Pluto Press.

32 Greenwald, G. (2014). No Place to Hide: Edward Snowden, the NSA, and the US Surveillance State. London: Hamish Hamilton.

33 Gürses, S., Kundnani, A., & van Hoboken, J. (2016). Crypto and empire: The contradictions of counter-surveillance advocacy. Media, Culture and Society, 38(4), 576–90. DOI:

34 Hackett, R. A., & Carroll, W. K. (2006). Remaking Media. The Struggle to Democratize Public Communication. New York: Routledge. DOI:

35 Hellegren, Z. I. (2017). A history of crypto-discourse: Encryption as a site of struggles to define internet freedom. Internet Histories, 1(4), 285–311. DOI:

36 Hine, C. (2002). Virtual Ethnography. London: Sage.

37 Hine, C. (2015). Ethnography for the Internet: Embedded, Embodied and Everyday. London: Bloomsbury.

38 Iqbal, M. (2019, February 19). WhatsApp revenue and usage statistics. Business of Apps. Retrieved from

39 Kazansky, B., & Milan, S. (forthcoming 2019). Infrastructure and protocols for privacy-aware research. International Studies Perspectives Forum.

40 Kozinets, R. (2014). Netnografia: realizando pesquisa etnográfica online. Porto Alegre: Penso.

41 Lee, B. K., & Gregory, D. (2008). Not alone in the field: Distance collaboration via the internet in a focused ethnography. International Journal for Qualitative Methodology, 7(3): 30–46. DOI:

42 Lee, F. L. F., & Chan, J. M. (2016). Digital media activities and mode of participation in a protest campaign: A study of the Umbrella Movement. Information, Communication and Society, 9(1): 4–22. DOI:

43 Levy, S. (1984). Hackers: Heroes of the Computer Revolution. New York: Penguin Books.

44 Ling, R., & Lai, C.-H. (2016). Microcoordination 2.0: Social coordination in the age of smartphones and messaging apps. Journal of Communication, 66(5): 834–56. DOI:

45 Marda, V., & Milan, S. (2018). Wisdom of the Crowd: Multistakeholder Perspectives on the Fake News Debate. Annenberg School for Communication. Available at

46 Matassi, M., Boczkowski, P. J., & Mitchelstein, E. (2019). Domesticating WhatsApp: Family, friends, work, and study in everyday communication. New Media and Society. DOI:

47 Máximo, M. E. (2007). O eu encena, o eu em rede: um estudo etnográfico nos blogs. Civitas, 7(2): 25–47. DOI:

48 Meikle, G. (2002). Future Active: Media Activism and the Internet. New York: Routledge.

49 Messenberg, D. (2017). A Direita que saiu do armário: a cosmovisão dos formadores de opinião dos manifestantes de direita brasileiros. Revista Sociedade e Estado, 32(3): 621–47. DOI:

50 Milan, C., & Milan, S. (2016). Involving communities as skilled learners: The STRAP framework. In Wildermuth, N., & Ngomba, T. (Eds.), Methodological Reflections on Researching Communication and Social Change (pp. 9–28). Basingstoke: Palgrave MacMillan. DOI:

51 Milan, S. (2010). Towards an epistemology of engaged research. International Journal of Communication, 4(2010), 856–58.

52 Milan, S. (2013). Social Movements and Their Technologies: Wiring Social Change. Basingstoke: Palgrave Macmillan. DOI:

53 Milan, S. (2014). The ethics of social movement research. In della Porta, D. (Ed.), Methodological Practices in Social Movement Research (pp. 446–64). Oxford: Oxford University Press. DOI:

54 Milan, S. (2015). When algorithms shape collective action: Social media and the dynamics of cloud protesting. Social Media + Society. DOI:

55 Milan, S. (2017). Data Activism as the New frontier of Media Activism. In Pickard, V., & Yang, G. (Eds.), Media Activism in the Digital Age (pp. 151–63). New York: Routledge. DOI:

56 Milan, S. (2018). The Materiality of Clouds. Beyond a Platform-Specific Critique of Contemporary Activism. In Mortesen, M., Neumayer, C., & Poell, T. (Eds.), Social Media Materialities and Protest: Critical Reflections, (pp. 116–127). New York: Routledge.

57 Montag, C., et al. (2015). Smartphone usage in the 21st: Who is active on WhatsApp? BMC Research Notes, 8(33). DOI:

58 Myers West, S. (2017). Ambivalence in the (private) public sphere: How global digital activists navigate risk. Presented at the 7th Workshop on Free and Open Communications on the Internet 17, Vancouver, BC. Retrieved from

59 Nemer, D. (2018, 25 October). The three types of WhatsApp users getting Brazil’s Jair Bolsonaro elected. The Guardian. Available at:

60 Ong, J. C. (2019). Toward an ordinary ethics of mediated humanitarianism: An agenda for ethnography. International Journal of Cultural Studies. DOI:

61 O’Reilly, K. (2005). Ethnographic Methods. Abingdon: Routledge.

62 Pereira, G., & Bojczuk, I. (2018, 9 November). Zap zap, who’s there? WhatsApp and the spread of fake news during the 2018 elections in Brazil. Global Media Technologies and Culture Lab, MIT. Available at

63 Pickard, V., & Yang, G. (Eds.) (2017). Media Activism in the Digital Age. New York: Routledge. DOI:

64 Pimienta, D. (2005). At the boundaries of ethics and cultures: Virtual communities as an open ended process carrying the will for social change (the ‘MISTICA’ experience). Retrieved from

65 Pink, S., Postill, J., & Horst, H. (2016). Digital Ethnography: Principles and Practice. In Postill, J., et al. (Eds.). Thousand Oaks, CA: Sage.

66 Potnis, D., Demissie, D., Trimmer, J., & Cleek, J. (2018). WhatsApp for social activism in Ethiopia? Research in progress from US Diplomacy Lab. Presented at the Southern Association for Information Systems (SAIS) 2018 Proceedings 25. Retrieved from>s2018/25

67 Resende, G., et al. (2018). A system for monitoring public political groups in WhatsApp. WebMedia ’18 24th Brazilian Symposium on Multimedia and the Web (pp. 387–90). New York; ACM. DOI:

68 Reuters Institute. (2017). Retuers Institute Digital News Report 2017. Oxford: University of Oxford. Retrieved from

69 Ryan, C., & Jeffreys, K. (2008). The Practice of Collaborative Theorizing. Retrieved from

70 Ryan, C., Salas-Wright, V., Anastario, M., & Camara, G. (2010). Making research matter … matter to whom? International Journal of Communication, 4: 845–55.

71 Santos, M., & Faure, A. (2018). Affordance is power: Contradictions between communicational and technical dimensions of WhatsApp’s end-to-end encryption. Social Media + Society, 1–16. DOI:

72 Savov, V. (2015, 17 December). Brazil’s WhatsApp ban is driving million of users to Telegram. The Verge. Retreived from:

73 Sevitt, D. (2017, 27 February). The most popular messaging apps by country. The Market Intelligence Blog, Similar Web. Retrieved from:

74 Souza, J. (2016). Radiografia do golpe: entenda como e porquê você foi enganado. São Paulo: Leya.

75 Spyer, J. (2017). Social Media in Emergent Brazil: How the Internet Affects Social Mobility. London: University College London Press. DOI:

76 Statista. (2019). Number of monthly active WhatsApp users worldwide from April 2013 to December 2017 (in millions). Statista. Retrieved from

77 The Economist. (2018, 20 October). Quantum computers will break the encryption that protects the internet. Retrieved from

78 Treré, E. (2018). The sublime of digital activism: Hybrid media ecologies and the new grammar of protest. Journalism and Communication Monographs, 20(2): 137–48. DOI:

79 Underwood, E. D., & Frey, L. R. (2008). Communication and Community: Clarifying the Connection Across the Communication Community. In Beck, C. S. (Ed.), Communication Yearbook (vol. 31). New York: Lawrence Erlbaum. DOI:

80 Utrecht Data School. (2017). Data Ethics Decision Aid for Research. Retrieved from

81 Vieira, G. (2018, 3 July). WhatsApp e as possibilidades de fortalecimento da democracia. Le Monde Diplomatique Brasil. Retrieved from

82 Ward, K. (2018). Social networks, the 2016 US presidential election, and Kantian ethics: Applying the categorical imperative to Cambridge Analytica’s behavioral microtargeting. Journal of Media Ethics, 33(3): 33–148. DOI:

83 Waterloo, S. F., Baumgartner, S. E., Peter, J., & Valkenburg, P. M. (2018). Norms of online expressions of emotion: Comparing Facebook, Twitter, Instagram, and WhatsApp. New Media and Society, 20(5): 1813–1831. DOI:

84 Weber, I., Bastone, P., & Barbosa, S. (2018). Resistências ao golpe: O povo diz 2016 nunca mais. Astrolabio: Revista Internacional de Filosofia, 22: 151–66.

85 Zanon, G. (2018, 12 April). No, end-to-end encryption does not prevent Facebook from accessing WhatsApp chats. Retrieved from Medium website:

86 Zelenkauskaite, A., & Bucy, E. P. (2016). A scholarly divide: Social media, Big Data, and unattainable scholarship. First Monday, 21(5). Retrieved from DOI:

87 Zimmer, M., & Kinder-Kurlanda, K. (2017). Internet Research Ethics for the Social Age. New Challenges, Cases, and Contexts. New York: Peter Lang. Retrieved from DOI: