Tracing and archiving ‘constructed’ data on Facebook pages and groups: reflections on fieldwork among young activists in Zimbabwe and South Africa

Introduction and background of the study

This reflexive account is culled from the author’s ethnographic fieldwork in Zimbabwe and South Africa which sought to examine how, and why, politically engaged youths use Facebook for political purposes. Data were collected via qualitative research methods, including social media ethnography (online participant observation), qualitative content analysis and in-depth interviews. The use of a range of data collection techniques allowed me to produce a deeper understanding of the phenomena under investigation. As part of the fieldwork process, I spent over two years (1 August 2011 to 31 August 2013) observing Facebook profile pages, fan pages and groups of youth activists from six social movements in Zimbabwe and South Africa. In Zimbabwe, my respondents were drawn from the Crisis in Zimbabwe Coalition (CiZC), National Constitutional Assembly (NCA) and Youth Forum Zimbabwe (YFZ). In South Africa, respondents came from the Unemployed People’s Movement (UPM), PASSOP and Right to Know Campaign (R2K). The six case organizations were chosen because of their involvement in both offline and online political/social activism and their politically engaged youthful constituencies in their respective countries. As I discuss below, the fieldwork process relied extensively on what Postill and Pink (2012) call social media ethnography, ¹ as I immersed myself in the six Facebook pages and groups, spending most of the time, observing young activists’ posting behaviour and interviewing them about their online and offline political activities.

Traditional ethnography is mainly concerned with immersion, or long-term observations and interactions with the group being studied in order to experience and understand events similarly to how the research subjects experience and understand those same events (Marcus, 1995). Although often characterized as a method, ethnography encapsulates a range of approaches, all of which inscribe a certain relationship between the researcher and the researched. In the words of Malinowski (1922: 25), ethnography intends ‘to grasp the native’s point of view…to realise his/her vision of the world’. This is achieved when an ethnographer lives with the local community and compiles detailed accounts of life, traditions and cultural practices of the local people. Ethnography endeavours to investigate in particular the perspectives of participants, the nature and forms of their knowledge, their interactions, practices and discourses aiming to draw connections between practices, experiences and the context (Lüders, 2004: 225).

Virtual ethnography [or social media ethnography] is an adaptation of the traditional ethnographic studies in the sense that it focuses on the virtual space although the chasm between online and offline is very blurred. As Postill and Pink (2012) surmise, the everyday life of the social media ethnographer involves living part of one’s life on the internet [including Facebook], keeping up-to-date with and participating and collaborating in online discussions. This view is also corroborated by Hine (2005) who argues that virtual ethnography does not involve the ethnographer travelling physically to a field site but rather following and observing ‘internet events’. Internet events on social media platforms can be observed through monitoring profile pages, hashtags, group discussion threads and trending topics. As pointed out earlier, this particular approach enabled me to participate both overtly and covertly in youth activists’ daily lives on Facebook for an extended period of time, observing what is posted, making sense of online interactions, asking questions and collecting whatever data was available to throw light on the issues under investigation (Hammersley and Atkinson, 1995). Instead of simply dipping into and out of Facebook groups and profile pages, I also engaged in periodic interactions with the researched via Facebook chat and private messages.

Unlike traditional ethnography, which tends to reify speech as more authentic than writing as part of its romantic legacy, social media ethnography treats written texts associated with online cultures as valid accounts of the realities of those being studied (Postill and Pink, 2012). Writing on social movements and activism, Hanna (2013: 285) suggests that data from social media platforms may allow researchers to receive and observe activists’ self-reported activity from real-time data. It is important to highlight that this practice also presents new biases where activists with enough technical know-how and economic capital will be frequent users of social media for movement purposes (Hanna, 2013).

In keeping with the epistemological imperatives of the qualitative research tradition, I blended traditional and digital data collection techniques (Gatson, 2011; Hallett and Barber, 2014; Murthy, 2008), also known as ‘hybridization’ (Denzin, 2004). Hine (2007) describes this approach as ‘connective ethnography’ which moves between online and offline as users of computer-mediated communication do, looking at the construction of boundaries and the ways in which different forms of communication are used to contextualize one another. Instead of treating online and off-line practices as separate, the ethnographer works with the subject through the various contexts of their life to understand how online and off-line practices are constructed alongside one another (Hine, 2007). This hybrid approach ‘not only gives researchers a larger and more exciting array of methods, but also enables them to demarginalize the voice of respondents’ (Murthy, 2008: 837). Connective ethnography allows the researcher to move beyond a single site of interaction thereby tapping into several interrelated sites (Hine, 2007). Because political participation amongst young people is increasingly taking place both online and offline, it was crucial for my fieldwork to avoid the one sphere bias.

This particular study contributes to the methodological debate in at least three ways. First, it adds an important dimension on how to do a study that integrates both offline and online spheres. This important because existing studies have recommended the use of ‘connective ethnography’ (Hine, 2007) but little work has teased out the mechanics of how to make use of both online and offline data when studying activism in a hybrid media system. Second, this article presents an innovative approach when it comes to negotiating entry into a multi-sited fieldwork environment which encapsulates knocking on both physical and virtual gates. Third, the study also signposts some of the most important ethical dilemmas which researchers deploying ‘connective ethnography’ (Hine, 2007) have to negotiate in order to produce rigorous scientific research. This is of great importance given the centrality of both thick and big data in the digital age. The architecture and features of Facebook are discussed below in order to situate the context within which social media ethnography took place.

A brief overview of the Facebook as a ‘field’ site

Like most ICTs, Facebook constitutes a messy fieldwork environment that crosses online and offline worlds. This entails that the researcher has to follow an object across different field sites or settings. Fields are not only multi-sited but also non-continuous (Balsiger, 2014; Gatson, 2011). ² Similar to physical spaces, where an ethnographer enters a different social world with different culture, language and etiquette; Facebook also has its own linguistic capital, etiquette and emoticons. In both spaces, ethnographers are exhorted to spend as much time as possible in the field over an extended period in order to gain familiarity with the community they study.

Founded in 2004 by Mark Zuckerberg with his college roommates and fellow students at Harvard University, Facebook is one of the most popular commercial social media platforms with an estimated 1.44 billion monthly active users. Although there is a gradual decrease in daily users, specifically among teenagers in Western countries who are opting for private-facing social media platforms like SnapChat, WeChat and WhatsApp, the youth remain the most active users of Facebook globally (Miller et al., 2016). Facebook’s business model is based on systematically monitoring and harvesting of user data (such as the way people describe themselves in their profiles, who they are connected to, their interests and hobbies and their online activities) which the company eventually sells as a commodity to advertising clients (Fuchs, 2012). This suggests that Facebook acts as an advertising agent linking advertisers with a huge pool of users on its database.

Facebook is based on a programming code that produces a specific digital environment (boyd, 2014). Like the architecture of physical spaces, Facebook’s architecture simultaneously enables and restricts particular modes of interaction. Users can only express themselves within the limits of a platform’s design (Jungerns and Jungherr, 2016). The affordances of the platform’s programming interface (API) also shape the kind of data that researchers are able to retrieve. On Facebook, the collection of digital trace data is limited to what is accessible through the official interface (Jungerns and Jungherr, 2016). Given the operations of Facebook algorithms, it is important to note that data does not ‘naturally’ occur but is ‘constructed’ and programmed by code. Social media data is not like oil to ‘naturally’ mine or raw data, but are both framed by and framing other contexts (Halford, 2015).

When this particular research was conducted between 2011 and 2013, Facebook’s features allowed users to friend other users, ³ share text updates (status updates), photos and private messages, with a large emphasis on interacting online through ‘liking’, ‘tagging’, ‘sharing’, ‘following’ and commenting on the ‘status’ of friends. In terms of privacy settings, Facebook’s architecture is designed in such a way that users are able to control which parts of their profiles are visible to others. Users’ profiles could be viewed by friends only, friends of friends, friends and networks and everyone on the internet.

The site also allowed users to create and join groups based around common interests and activities. Once a user joined a Facebook group, he/she received a message whenever someone posts something in the group. There were three kinds of Facebook groups: secret, public and private. A secret group did not appear in group search results or in members’ profiles and requires an invitation from an administrator to join. Group information could only be viewed by members. A public group allowed anyone to join and to invite others to join. Information in this group could be viewed by anyone with a Facebook account. A private group required approval from Facebook administrator(s) for one to join.

Noteworthy to highlight that Facebook is a dynamic environment which has changed significantly since 2013 and has upgraded its social infrastructure [including features, EdgeRank algorithms and analytic tools] over the years. For this particular study, the choice to focus on Facebook was also shaped by the fact that the platform compared to others has a far larger penetration in Africa. Although it has been overtaken by WhatsApp in terms of subscriptions, Facebook is the second most popular social network site in Zimbabwe and South Africa. The study acknowledges that people on Facebook are not representative of the entire national population. The site is overly used by the middle class and urban-based youths which suggests that those from rural areas and lower classes are under-represented. In the next section, I briefly discuss the national contexts which influenced significantly the choices I made throughout the data collection and analysis stages.

Research context: Zimbabwe and South Africa

Zimbabwe and South Africa provided two extreme cases for a richer comparative study on how and why youth activists deployed Facebook for mobilization and activism. The former has been classified as ‘not free’ (meaning there is no respect for political rights and civil liberties) while the latter has been characterized as ‘free’ (meaning there is sufficient respect for political and civil liberties) though its media context is characterized as ‘partly free’ (meaning there is limited respect for media freedom and freedom of expression) (Freedom House, 2016). In terms of political transitions, Zimbabwe and South Africa have experienced ‘elite continuity and renewal’ (Sparks, 2011) rather than genuine political and media transformation. In the case of Zimbabwe, the transition from an authoritarian colonial state led to a post-colonial state which could be described as a hybrid regime (a combination of authoritarian and democratic tendencies). South Africa is considered a ‘model’ of electoral democracy—irrespective of the existence of major social and economic inequities (Sparks, 2011). In both countries party alternation has not occurred since the regime transitions—with the Zimbabwe African Nation Union-Patriotic Front (ZANU-PF) and the African National Congress (ANC) (in South Africa) dominating the political sphere. These liberation parties have, however, seen their support bases waning in urban and peri-urban areas in the last couple of years.

Notwithstanding similarities and differences in terms of political transformations, both countries have been riddled by different levels of constitutional, accountability and electoral conflicts. In South Africa, accountability conflicts have taken the form of community protests which are largely fuelled by a range of issues like corruption, e-tolling, service delivery backlogs, youth unemployment and the influx of foreign migrants. These protests are largely born out of the frustration with the continued high levels of inequality and a revolt against a government that is increasingly seen as uncaring and not listening. Zimbabwe has experienced constitutional, accountability and electoral conflicts which have manifested themselves through disputed elections, poor service delivery, deteriorating economic situation and a worsening human rights record (Mare, 2015). Discontent especially amongst out-of-school and school going urban youth has been singled out as a key factor in community protests in both countries. In 2015 and 2016, both nations witnessed a resurgence of youth protests as evidenced by #ThisFlag and #Tajamuka citizen movements in Zimbabwe and #RhodesMustFall and #FeesMustFall student movements in South Africa (Mare, 2015).

ICTs and activism research: revisiting the methodological debate

Several researchers (see Baker, 2013; Miller, 2011; Pink and Postill, 2012) have begun to focus on Facebook as an ethnographic object and area of inquiry. Consequently, there has been a methodological shift from focusing on geographical spaces towards online ethnographies. This shift has generated ‘big data’ which contributes to new ways of understanding the social world. ICTs are also providing methodological opportunities for researchers to adapt traditional research approaches to virtual spaces as well as creating entirely new data collection tools for these uncharted territories.

An ever-growing body of ethnographic studies (Gerbaudo, 2012; Mosca, 2014; Postill and Pink, 2012) focusing on the use of ICTs by social and political activists in Western and non-Western democracies has begun to emerge. These studies amongst others have proposed the use of offline and online data collection methods. Gerbaudo (2012) used in-depth interviews and online and offline participant of public gatherings in Egypt, Spain and the US. Based on multi-sited fieldwork, these ethnographic studies (Miller, 2011; Postill and Pink, 2012) employed extensive and intensive periods of offline and online participant observations in protests, qualitative content analysis of websites and discussion boards as well as interviews with activists. These examples illustrate the fact that ethnography constitutes a flexible, adaptable and an enduring methodological approach suitable for studying new technologies and activist practices.

A highly polarised debate on how best to study new media technologies has surfaced, pitting at least two extreme camps—those who believe there is no need for new inventions in methods (Van Selm and Jankowski, 2006) and those who believe that a whole set of new methods are required (Hine, 2005). The first camp in this methodological debate maintains that traditional methods are adaptable and flexible enough to meet the demands of studying internet sociality. Old research methods are viewed as retaining much of their salience in contemporary internet studies. The argument is that as long as the internet is strongly embedded in people’s everyday lives and not radically divorced from offline activities, traditional data collection methods can be easily imported, remediated, and simply digitalized without facing particular challenges (Mosca, 2014). In other words, online and offline spaces are not necessarily different because the online environment remains of this world and not, therefore, beyond existing knowledge or method.

Some scholars (Miller and Slater, 2000), however, recommend the re-sharpening of existing research methods to fit new research environments in which ICTs challenge existing research assumptions and premises. In this vein, Baym and Markham (2009: 26) argue that ‘the internet changes the way we understand and conduct qualitative inquiry’. Proponents (Howard, 2002; Murthy, 2008) of the ‘modernizing’ perspective call for a different methodological orientation on the part of the researcher, in order to speak to the ever-changing digital fields. They observe that conventional techniques must innovate and transform to accommodate the blurring nature of offline and online field sites. Attempts to modernise ethnography has seen the fieldwork in ‘the field’ being substituted by ‘fieldwork in and of networks’ (Howard, 2002). This school of thought advocates for the use of hybrid research techniques, like face-to-face and online interviews.

Other scholars (Bruns and Burgess, 2012) have advocated for the shift towards big data in social research. Big data entails the use of network analysis, computer algorithms and data visualisation techniques to map out large-scale communication patterns and network structures (Stephansen and Couldry, 2014). However, critics (boyd and Crawford, 2012; Stephansen and Couldry, 2014) of big data argue that by privileging large-scale quantitative approaches, it sidelines other forms of analysis and limits the kinds of questions that can be asked. They (boyd and Crawford, 2012) propose the use of qualitative and mixed research methods which foreground textual analysis and qualitative interviews with social media users. In keeping with the big data turn in social research, Rieder (2013) has developed an online tool known as Netvizz which is a Facebook application that allows for content scraping and generating statistics of public profiles and pages. It allows researchers to glean information about users from their profiles without their active participation.

According to the second camp, traditional research methods are now moribund, hence new ones are urgently required (Hine, 2005). New technologies are castigated for spawning a ‘crisis for the ethnographic project’ (Horst, Hjorth and Tacchi, 2012) or ‘crisis of empirical sociology’ (Savage and Burrows, 2007), thereby destabilizing the epistemological, ontological and methodological roots of social research. These technologies are also seen as unleashing ethnography from the traditional bounded field site or place towards what Marcus (1995) calls ‘multi-sited fieldwork’. Multi-sited fieldwork involves the moving out from the single sites and local situations to examine the circulation of cultural meanings, objects, and identities in diffuse time-space (Marcus, 1995). Their argument is that traditional methods were designed for the study of physically bounded social interactions; hence there is need for newer methods to understand deterritorialized social interactions. This school of thought advocates for the adoption of qualitative and quantitative digital methods like online questionnaires, online interviews and virtual ethnography.

Despite the afore-mentioned never-ending debate, Jankowski and van Selm (2005) posit that modifying existing methods is a more common practice than radical reconstruction. This is partly because online cultures cannot be divorced from the offline positions, values and meanings that users bring to the online environment (Kozinets, 2010). Cognisant of the inseparability of online and offline realms, scholars (see Hine, 2005; Murthy, 2008; Miller and Slater, 2000) have proposed the expansion of ethnography to embrace and address mediated venues. Unlike traditional ethnography, online ethnographic studies foreground the use of digital practices like email communication as well as covert and overt participation in chat rooms (such as Facebook groups and profile pages) to conduct forms of participant content analysis. This study triangulated offline and online data in the sense that it used in-depth interviews (offline), social media ethnography and qualitative content analysis. Online data was collected before offline because it was important for constructing unstructured questions which were asked during the in-depth interviews. It provided a rich background upon which to conduct offline interviews. For instance, Mosca (2014: 413) observes that offline phenomenon would be impossible to understand without seriously studying the online environment. In cases, when conflicts between data occurred the researcher returned to the respondents on Facebook or offline for clarification. It was a back and forth process where online data was verified and cross-checked through online and offline interviews with activists in Zimbabwe and South Africa.

In the next three sections, this article focuses on how I negotiated entry in multi-sited environments, the ‘nuts and bolts’ of social media ethnography and ethical dilemmas associated with tracing and archiving ‘constructed’ data on Facebook.

Negotiating entry into a multi-sited research context

Negotiating entry in a multi-sited ethnography that cut across the online and offline boundaries is riddled with serious challenges. This is largely because building relationships with interviewees requires trust that is not easy to generate online because disembodied, anonymous, and textual settings do not facilitate conditions of mutual confidence (Hine, 2005). The situation is even more complicated when dealing with sensitive topics like political activism in authoritarian contexts where respondents are suspicious of communication surveillance by intelligence officers. As Hine (2005: 20) reminds us, “[e]stablishing one’s presence as a bona fide researcher and trustworthy recipient of confidences is not automatic, [it] varies depending on the cultural context under investigation”. At the time when I started my fieldwork in August 2011, there was a conspicuous absence of literature focusing on gaining entry on Facebook groups and fan pages. Despite a few books (Miller 2011; Marichal, 2010) having been published, extant research methods are silent on the mechanics of negotiating entry on the site.

As part of my pre-fieldwork preparation, I spent a significant amount of time on Facebook trawling through groups, fan and profile pages as a “passive observer” in an attempt to get an ‘intimate familiarity’ of the research site. This approach to gaining entry known as “mental access” (Gummesson, 2000) enabled me to understand what is happening, delineate what to observe and from whom to gather information. I engaged in background listening in some of the vibrant Facebook groups and fan pages to acquaint myself with online political practices. Although background listening as an ethnographic strategy is replete with ethical challenges (which I will discuss below), ⁴ it allows for a period of cultural familiarization in order to facilitate a relatively smooth entry into the field. This strategy finds support in Milan (2014) who suggests that simple online observation in the early stage of research can be the best way to familiarize oneself with the phenomenon and to facilitate entry into the field.

Officially, I started my fieldwork on 1 August 2011 after receiving informed consent from individual participants and gatekeepers from the case organizations under consideration. I disclosed my status as a PhD student at Rhodes University undertaking a research project on the use of Facebook for political purposes by youth activists in Zimbabwe and South Africa. I made it clear to the gatekeepers that I was not an ‘activist’ although I sympathised with their struggles. My journey towards gaining entry began when I was granted an ethical clearance from Rhodes University’s ethics committee. My supervisors also wrote me a release letter addressed to the gatekeepers at the six case organizations which introduced me as a PhD student and also briefly explained the objectives of the study.

Because ‘multi-sited’ (Marcus, 1995) fieldwork involves knocking at multiple gates of entry, I found myself having to deal with gatekeeping at an individual (youth activists), Facebook group or page (administrators) and organizational (leaders of the six social movements) levels. Being accepted as a member-cum-researcher in a Facebook group required negotiating access with the administrator(s) (often someone who is not the leader of the organization). In order to circumvent this conundrum, I resorted to knocking on physical and virtual gates manned by different gatekeepers. These gates were sometimes closed, partially closed or opened. As intimated earlier, this is largely because Facebook offer sophisticated settings that allow users to restrict access to certain pieces of information.

Respondents were selected based on active participation on Facebook and online protest actions. Active participants refer to users who create, transform, distribute, and consume content on the web (van Dijck, 2009). I had minimal success recruiting respondents through emails (sent to their professional addresses) and private messages on Facebook. Most of my emails were either ignored or sent back with a message of refusal in South Africa. I kept on knocking persistently until some ‘gates’ were opened. Referral through friends and Facebook group administrators also proved worthwhile, as some respondents who initially turned down my requests ended up agreeing to take part in the study. This illustrates that the process of gaining entry is never linear but involves negotiation, persistence and continuous re-negotiation.

Compared to South African social movements, I found those from Zimbabwe extremely difficult to access because of the inherent fear harboured by gatekeepers that researchers can infiltrate their movements using research as a cover up. Some of these movements have been subjected to both physical and communication surveillance by the state which accounts for their schizophrenic attitude towards researchers. I noticed that some youth activists in Zimbabwe were suspicious of my identity and research objectives beyond academic interests. Their suspicions towards me are summed up in the following conversation:

In order to negotiate physical access, I used Patton’s (2002) ‘known sponsor approach’ which entailed relying on the leaders of the purposively sampled case organizations to introduce me to the rest of their members. This kind of snowball sampling technique allowed me to gather research subjects through the identification of an initial subject who was used to provide names of other participants). The ‘known sponsors’ were briefed about the nature of the research, data collection tools, fieldwork timescale and ethical obligations. Support at the leadership levels of the social movements was crucial to my success in gaining access to the rest of the members. This chimes with Stake’s (1995) assertion that individuals often immediately acquiesce if a superior has granted permission.

The fact that I ‘shared’ Facebook friends and group affiliations with some of the respondents opened physical and virtual gates for me to conduct fieldwork. As Ekdale (2013) observes, social media provide a digital archive for the researcher to become the researched. Because of the shared friends on Facebook, some respondents treated me as a fellow ‘activist’ in both South Africa and Zimbabwe. I was constantly greeted using the title ‘comrade’ on Facebook and during face-to-face interviews. Be-friending my respondents on Facebook allowed me to track their online activist practices as well as maintaining social relationships with them. These encounters also underscore the fact that ethnography is not just about observations and interviews, but that it also involves ‘informant cultivation’ and the exploitation of personal relationships forged prior to the research.

Social media ethnography

Social media ethnography as a variant of (online) participant observation enables researchers to learn about the activities of the people under study in their natural (and ‘constructed’) settings through observing and participating in those activities (Hine, 2005; Postill and Pink, 2012). This data collection technique allowed me to go beyond taken-for-granted assumptions and dig deep into youth activists’ everyday (online) life contexts or what Malinowski (1922: 18) calls the ‘the imponderabilia of everyday life’. Given the ‘general tendency for people to disclose more about themselves online’ (Hine, 2005: 18), social media ethnography allowed me to observe young people’s use of Facebook for political purposes which could not be gathered through qualitative interviews alone. This practice finds support in Lichterman’s (1998: 401) suggestion that [online] participant observation ‘can teach us much about the everyday meanings of doing social [and political] activism’.

As noted above, formally online participant observation took place during a period of two years. Although I did not keep records of how much time I spent combing through Facebook groups, fan pages and profile pages, on average I scanned at least 10 profile pages a day in an effort to keep track of my respondents’ posting behaviour. As the next few paragraphs will illustrate, social media ethnography is not necessarily a single research method, but rather a blend of formal and informal interviews, observations, interactions, and content analysis. Building on Postill and Pink’s (2012) typology, I expand on seven overlapping routines at the heart of social media ethnography: background listening, friending/ liking, observing, catching-up, exploring, interacting, and archiving. Although these stages do not necessarily follow any chronological order, they feed off and into each other in complex ways.

The first routine comprised of background listening which entailed monitoring Facebook fan pages and groups without necessarily disclosing my identity to the gatekeepers. As a metaphor for attending to discussions and debates online, listening more usefully reflects the fact that everyone moves between the states of listening and disclosing online; both are necessary and both are forms of participation (Crawford, 2009). The architecture of Facebook groups and fan pages enabled me to engage in background listening without revealing my identity in the initial stages of the research. This was important to get a sense of how the social movements had incorporated Facebook into their repertoires of communication. Whilst background listening has its own ethical pitfalls, it is important to highlight that it allowed me to conduct some kind of online ‘pilot’ observation before plunging into the deep end. Data were gathered by friending people on Facebook after debriefing them about the purpose and objectives of the study. Informed consent was given first after the friending process.

The second routine consisted of friending/liking: Similar to offline spaces, Facebook allow researchers to friend/like their respondents and see what they are doing online on a daily basis. I used my own Facebook profile page to friend and like my respondents (ethical issues will be teased out below). Through friending and liking, I was able to elicit some modicum of trust amongst respondents especially those who didn’t know me in offline spaces. In total, I befriended and observed at least 102 Facebook profile pages in Zimbabwe and South Africa. Friending on Facebook also proved important in breaking the great enclosure or walled garden associated with profile page and group privacy settings. As my ‘friends’ on Facebook, I had access to the private and public information especially on profile pages. As Pascoe (2012) writes, these spaces provide a window into youth social worlds that might not otherwise be available. Friending also involved ‘liking’ fan pages run by purposively sampled case organizations on Facebook. Liking of fan pages was chosen as a mode of reaching out because it ensured that I became part of the discussion threads through periodic notifications and access to posted material.

The third routine entailed observing conversations, interactions, posting and commenting behaviours of individuals and group members on Facebook. Taking into consideration ethical guidelines, a decision was made to observe and identify relevant ‘political’ rather than ‘personal’ postings on Facebook groups and profile pages,^{6, 7} although I acknowledge that there are blurred lines between them. Whilst I also recognised that personal lives are not insulated from the repercussions of various socio-political events and that personal posts have political implications, I chose to focus on political posts which reflected the youth activists’ relationship with the chosen social movement. This allowed me to draw a flexible checklist of how young people use Facebook for political purposes. Directly observing what youth activists from both case nations ‘shared’, ‘updated’ and ‘commented’ on in their Facebook groups and profile pages enabled me to get insight into how they used the site to advance their political objectives. It also allowed me to account for the kind of activist work that young people do ‘behind the screen’ (Orgad, 2005: 58). Because looking at what people talk about with others ‘is an ideal unit for examining social behaviours (…) and reveals aspects of groups that are not evident from aggregations of individual behaviours’ (Haythornthwaite, 2005: 127). As part of the observation phase, I took field notes (through saving relevant status updates, conversations, images and so forth) based on my personal impressions. I took into consideration Facebook’s fair usage policy when reporting online data. ⁸ As a corporate entity, Facebook discourages copyright violations on its website. Hence, my online data only used a limited portion of posts, pictures and videos from Facebook pages and groups.

Besides merely observing and identifying relevant political postings and discursive interactions on Facebook groups and profile pages, the fourth routine I engaged in was catching-up with my respondents. I kept track of the happenings on my research participants’ Facebook groups and profile pages through periodically checking my ‘notifications’ function. ⁹ I also used the Facebook news feed function as a ‘semi-public notice board’. ¹⁰ This means I read and got timely updates on individual and group discussions through regularly checking my Facebook news feed. Instant messaging features on Facebook provided opportunities for me to engage in real time chats in an ongoing fashion. The administrator(s) of the Facebook debriefed the members and fans about my research project.

My fifth routine involved exploration. Through this routine, I regularly explored and followed external website links, listened to audio and watched video messages posted by youth activists on their Facebook groups and profile pages. Following and monitoring youth activists’ trace data on Facebook allowed me to track patterns of communication, posting behaviour and to quantify the levels of participation on the platform. Most of the external website links posted by the young people often took me to online newspapers, blogs, social movements’ websites, motivational quotations and book reviews.

As a fore-runner to in-depth interviews, my sixth routine involved interacting with my respondents through Facebook chat, private messages, WhatsApp and mobile phones. ¹¹ Facebook’s features (email, instant messaging, texting, and Skype) can be used to facilitate interviews. As Pascoe (2012) notes, these features allow researchers to modify traditional interview techniques by addressing challenges such as young people’s physical immobility, their distance from the interviewer, or the difficulty of accessing a dispersed population. I also used my mobile phone to schedule interviews, to conduct interviews and to pose follow-up questions. As Pelckmans (2010: 31) suggests, ‘telephone calls to and from the field (“phoning the field”) serve as a reminder of the open-ended and seemingly placeless nature of contemporary fieldwork’. Besides enabling me to develop an extended set of ‘weak-ties’ with the respondents, informal conversations on Facebook allowed me to ask follow up questions intended to allow respondents to expand their answers on individual subjects. I engaged in some kind of ‘reciprocal listening’ (Crawford, 2009) with friended respondents on Facebook. ¹² Dubois and Ford (2015) call these informal conversations ‘trace interviews’ . ¹³ Trace interviews allowed me to use data gathered from online participant observation as probes thereby serving as ‘instructive way[s] of stimulating the interviewees’ memory and encouraging [them] to elaborate’ (Orgad, 2005: 61). Below is one such trace interview via Facebook messaging system:

Informal conversations though chats on WhatsApp also facilitated phatic communion with geographically dispersed respondents. This mobile instant messaging platform enabled me to conduct interviews during times when my respondents were free to chat. During these informal interviews with my respondents, I often talked about their profiles and group discussions to get a sense of what they thought about the political discourses and conversations they engaged in with their friends on Facebook. Facebook chat proved the most popular form of interviewing as some respondents found it hard to set up physical meetings. As one respondent from South Africa remarked, ‘I prefer that we conduct the interview on Facebook because it allows me to do other things while I am chatting with you’. Consequently, I resorted to Facebook’s real-time chat system which gave me an opportunity to communicate with several respondents whilst simultaneously observing their online activities. The advantage of this mode of interviewing is that respondents write in their own words which reduces data transfer errors and the time required to transcribe recorded interviews.

The seventh routine consisted of archiving the qualitative and quantitative data gathered from Facebook groups, fan pages and profile pages. Archiving of online data is required ‘in order to have stable object to study and refer to when the analysis is to be documented’ (Bruggler, 2011: 24). Social media has rendered web-archiving a more challenging task (Mosca, 2014). The problem is that social media data can be only partially automated while mining of big data requires huge hardware space and processing power. Although using Facebook archiving technologies for archiving purposes was not in the researcher’s right, it is crucial to note that as part of the debriefing process the respondents were offered the right to withdraw from the study. In fact, some respondents deleted some posts after warnings from their friends and relatives. Instead of ‘field notes’ (Malinowski, 1922) associated with traditional ethnography, I used a combination of manual (copying and pasting onto an MS word document) and electronic (relying on archived material by Facebook Inc.) archiving systems. I regularly copied and pasted onto MS word documents any status updates which suited my research questions. For the purposes of data storage and capture, archived data was classified according to the theme of the status update, comments, pseudonym of the author, publication date, group information and name of the event. Facebook is also handy for researchers because it archives data that are constitutively evanescent, rapidly changing and at risk of disappearing (Mosca, 2014). This is invaluable given that, as Hanna (2013) observes, that activists have the tendency to delete information they consider incriminating after a protest event. I experienced this in the Zimbabwean field site where some youth activists deleted some of their posts after the July 31, 2013 harmonised elections.

Ethical dilemmas associated with tracing and archiving data on Facebook

As this section will highlight, tracing and archiving ‘constructed’ data using and through Facebook groups and profile pages is dogged by several ethical challenges which require flexibility and sensitivity on the part of the researcher. As Halford (2015) aptly puts it, researchers are finding themselves at the centre of a perfect ethical storm especially at a time when university ethical governance is becoming increasingly bureaucratized and risk adverse (Halford, 2015). These ethical challenges ought to be navigated in order to produce rigorous and meaningful data while protecting the interests of participants and remaining within the limits of what is legally and logistically feasible (Mooney, 2015). Ethical codes often serve as safety valves to raise awareness amongst researchers as to the potential repercussions and outcomes of their actions upon the researched and at times the wider community. In light of this, Pascoe (2012) suggests that most institutional review boards are not yet equipped to review these online ethnographic techniques and researchers may have to educate their own institutions as to the ethics of these sorts of research practices. The situation has been compounded by the fact that some of the time-honoured institutional ethics can no longer provide robust ethical direction in the era of ‘data deluge’. Traditional social research ethical regimes and standards have become fluid and porous.

Because scientific research ethics exhort researchers to avoid harming their respondents as well as to avoid infringing on the rights of respondents, I had to navigate a minefield of ethical issues. Some of the ethical questions that arose during the fieldwork and reporting phases include: the exploitation of personal contacts within Facebook systems for research purposes; copyright associated with reproducing content for research and publication purposes and informed consent when qualitatively examining individuals’ activities on Facebook. Facebook has blurred the notion of individual and group privacy thereby complicating issues of confidentiality, accountability and informed consent. As such, it challenges the researcher’s ability to maintain a boundary between their public and private lives. Below I discuss how I navigated this ethical minefield as well as offer guidance on how online ethnographers can also negotiate these dilemmas.

Although I received ethical clearance from Rhodes University’s Ethical Standards Committee, I must point out that university ethical codes are insufficient to illuminate hidden aspects and offer practical solutions to researchers in online inquiry settings. Whilst I do not advocate for the total abandonment of academic scientific ethics, I argue that these guidelines are not cast in stone. These ethical guidelines should be thoroughly contextualized (Whiteman, 2010) in order to speak to the dynamics of multi-sited environments. Contextualized ethics do not exclude the relevance of general principles to the practice of research but rather explore the interpretation and relevance of these principles in specific research contexts (Whiteman, 2012). Cognisant of the risks associated with communication surveillance, I sought informed consent from my respondents before proceeding with online and telephonic interviews. I also ensured that sensitive questions were posed through relatively secure platforms like face-to-face conversations and WhatsApp chat in order to safeguard the security of my respondents. WhatsApp is also valuable because it provides end-to-end encryption and a third security layer designed to stop the possibility of man-in-the middle attacks.

In order to meet the ethical obligations of this study, I requested permission to be a researcher on the six Facebook groups from their administrators. Although some of these groups were public groups, meaning their information could be accessed by everyone on Facebook, I requested permission from the administrators to use messages contained therein. I also asked youth activists for their permission to use Facebook postings and to record them during the interviews. Given the intricate relationship between informed consent and privacy concerns in an online environment, I adopted ‘de-lurking’ as an ethical strategy. De-lurking entails making my presence known to the respondents before befriending and observing them on Facebook (boyd, 2014). This strategy enabled me to develop social rapport with the participants I observed and contacted for further conversation.

Another ethical problem associated with social media research is traceability. This refers to the fact that all communication which is typed rather than spoken leaves a physical trace known as a data trace that can be archived or preserved (Jungherr, 2015). ¹⁴ In order to deal with this ethical problem, data mined from Facebook systems was used sparingly to ensure confidentiality and privacy (Sveningsson, 2004). Due to the political volatility of the Zimbabwean context and the sensitivity of some of the responses provided by respondents, I decided to use the term ‘male’ or ‘female youth’ in place of real names (associated with Facebook policy) to preserve the confidentiality of my research subjects. The terms male or female youth were qualified with the use of the name of the social movement (for instance, female youth, NCA) with regards to the citation of interviews and online participant observations. All ‘Googleable’ postings were de-identified through a combination of paraphrasing and use of the term male or female youth. In order to meet copyright and Facebook’s fair usage policies, I ensured that all posts and comments gathered online included the following acknowledgement: name, date and timestamp. Because gaining permission to use online data especially from commenters and participants in discussion threads is complicated, it makes sense to acknowledge the date and timestamp of the post.

Conclusion

This article has, through a reflexive account of the authors’ fieldwork in Zimbabwe and South Africa, attempted to show that social media ethnography is a negotiated experience whose essence prescribes understanding people’s perspectives ‘from the inside’ while also viewing them more distantly to allow for in-depth analysis. Based on a seven-stage typology, this particular study has demonstrated how ethnographers can trace and archive constructed data. It has also explored the various options that researchers can experiment with in order to navigate challenges associated with gaining entry in messy field sites. It has dealt with the mechanics of online participant observation on Facebook and ethical dilemmas. Furthermore, the article has foregrounded how ethnographers can use trace interviews and online content to make sense of activist cultures in a hybrid media system. It has argued that online like offline participant observation has context specific methodological dilemmas which require innovative flexibility, reflexivity and ethical sensitivity on the part of the qualitative researcher.