1On November 5, 2020, the Google Play app store counted 5,707,840,304 WhatsApp downloads on Brazilian smartphones since its launch in 20091. The platform announced, in February 2020, that it had reached the milestone of two billion users distributed in more than 180 countries in which it operates2, while the most recent official data points to the existence of 127 million active accounts per month in Brazil3.
2In this scenario, WhatsApp, as the most popular instant messaging application, occupies a central place in Brazilian communication mediation. It is through this app that individuals connect with friends and family, seek and offer services and products, follow discussion groups on topics of personal interest, and debate political issues. A survey conducted by DataSenado in 2019, for example, showed that, for 79% of respondents4, WhatsApp is the main means of accessing information, ahead of television and radio, as well as digital channels such as YouTube, Facebook, and news websites.
3The adoption of WhatsApp into the daily lives of citizens, associated with the strategic use made by parties and interest groups, significantly alters the logic of contemporary political communication. Analysts and researchers attribute President Jair Bolsonaro’s electoral success to the coordinated articulation between WhatsApp groups capable of activating a base of supporters and promoting political mobilization based on the dissemination of false, misleading, and inaccurate information (Machado & Konopacki, 2019; Machado et al., 2019; Chagas, Modesto & Magalhães, 2019; Baptista et al., 2019; Evangelista & Bruno, 2019; Pancho, Bastos & Silva, 2020).
4Despite its growing importance in the political debate, the implications of the use of the messenger in contemporary sociability remain little explored. As this is an emerging topic, research on WhatsApp still faces a series of methodological, epistemic, and ethical challenges. Studies whose research locus is WhatsApp have adopted different methodological approaches and strategies, such as netnography, participant observation, and content analysis. However, an element that these approaches share is the need to observe ethical parameters for social research in digital environments mediated by WhatsApp (Townsend & Wallace, 2016; Barbosa & Milan, 2019). The challenge, therefore, is to structure investigation designs and monitoring strategies that respect ethical requirements.
5This article aims to shed light on procedures, guidelines, and risks involving observation of human behavior in WhatsApp groups in Social Sciences field research. For this, this manuscript is organised as follows: in the first section, we present a mapping of WhatsApp research and an analysis of the areas of knowledge that have produced studies on the application. Then, we focus specifically on research that deals with the implications of WhatsApp in the political communication field, starting from a literature review of Brazilian production on the aplication. In a second part, we systematized a series of characteristics that involve research stages and ethical research questions on the use of the messenger for political purposes.
6To understand the global frequency and evolution of research that use WhatsApp as an object and/or analysis environment globally, we performed three searches as a first exploratory mapping effort. First, held on October 29, 2020, we used the Findpapers5 application to search for scientific articles available in five digital libraries: ACM, arXiv, IEEE, PubMed, and Scopus. As the purpose of the mapping is to explore the entire field of academic production about the messenger, we searched for studies that contained the term “WhatsApp” in the title, abstract, or keywords. This collection covered studies published from 2010, the year following the launch of the instant messaging application, to 2020. We identified 2,896 publications indexed in the five scientific databases, and Scopus, which covers all areas of knowledge, delivers 71.7% of this corpus (Table 1).
Table 1. Results of collection of WhatsApp references by digital library
|
Library
|
N
|
Percentage
|
|
Scopus
|
2.076
|
71,7%
|
|
PubMed
|
406
|
14%
|
|
IEEE
|
202
|
7%
|
|
ACM
|
166
|
5,7%
|
|
arXiv
|
46
|
1,6%
|
|
Overall
|
2.896
|
|
7In addition to Scopus, which tends to cover a wide variety of knowledge areas among the indexed titles, we identified four databases with more focus on specialized areas. ACM is a repository with all publications of the Association for Computing Machinery, including scientific articles, conference proceedings, technical notes, etc.6, in the field of computer science. arXiv is a pre-printed archive of scientific papers from nine disciplines: physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, electrical engineering, scientific systems, and economics7. The IEEE, from the Institute of Electrical and Electronics Engineers, brings together publications in the fields of engineering, computer science, and electronics8. PubMed’s search engine, in turn, focuses on the areas of life science and medical science9.
Figure 1. Distribution of academic production on WhatsApp over time (2012-2020)
8It should be noted that some titles are in more than one of the libraries used, which required the removal of duplicate references in the research sample. By eliminating the duplicates, we reach the number of 2,217 articles, which are temporally distributed as above (Fig. 1). As can be seen, no publications were identified in 2010 and 2011. After this initial mapping, a second search (November 5, 2020) was conducted only in the most expressive digital library, Scopus, this time with the terms “WhatsApp” and “Brazil”, to better understand the references that debate national issues. In this new search, the quantity of publications was reduced to 51 titles. Research on this app in the Brazilian context began timidly in 2015 with an article focused on the analysis of digital networks. The paper “Intensity Use of Social Networks in Southern Brazil” examined, through a survey, the use of social networks by the population of southern Brazil (Wojahn, Oliveira & Souza, 2015). Most publications, however, focus on the years 2019 and 2020, with 15 and 24 respectively. These studies are in areas such as health and communication, with emphasis on disinformation. Figure 2 represents the distribution of titles by knowledge area, considering the area of concentration of researchers and the fields to which the annals or indexed journals are localized.
Figure 2. Distribution of studies about WhatsApp in Brazil per area of knowledge
9In the third step, we use this Scopus base to filter the searches that relate to WhatsApp and politics in Brazil. With this number in hand, we ran Google Scholar search cycles to complement the database, as we identified important absences that were not indexed in Scopus. This systematization resulted in 67 titles in Portuguese and English of Brazilian researchers or about Brazil, which will be the object of a brief literature review in the next topic. This sample reveals the same growth trend in publications between 2019 and 2020. We observed a clear evolution in the number of publications each year. The numbers show that research within and/or with WhatsApp has grown at a strong pace at least since 2013, peaking in the year 2020. The concern involving scientific ethics reflects the application’s leading role in understanding phenomena in various scientific fields. Data processing, privacy issues, and user consent are demands raised by the use of the messenger.
10The adoption of WhatsApp as a communication and social interaction tool has unfolded into a political phenomenon. From 26 scientific articles in different languages, Schaefer, Barbosa, Epitácio, and Resende’s (2019) systematic literature review offers some clues on electoral political communication on WhatsApp between 2010 and 2019. Schaefer and colleagues (2019) identify a relatively balanced distribution between studies that adopt a qualitative approach and those that invest in quantitative methods to explain the effects of the use of WhatsApp on numeral results. According to the authors, there are four units of analysis in studies on the messenger: parties and candidates, voters, WhatsApp groups, and other types of organizations.
11The techniques and methodological procedures adopted in the research are directly related to the object under analysis. The authors find that studies investigating, for example, the adoption of WhatsApp as a campaign tool for political parties and candidates tend to adopt a more qualitative approach; studies focused on groups, on the other hand, take a quantitative approach based on descriptive or inferential statistical analysis. Schaefer and colleagues (2019) also find that most Brazilian research on WhatsApp tends to focus on WhatsApp groups as units of analysis, which further reinforces the need to debate the ethical parameters adopted by this kind of research carried out in the country.
12Regarding the themes studied, in Brazilian academic production, it stands out a significant number of studies focused on electoral processes, more specifically on WhatsApp as a tool used for disinformation actions (Baptista et al., 2019; Muzell et al., 2020; Santos et al., 2019) or digital populism (Cesarino, 2020). From the analysis of 90 public WhatsApp groups of supporters of six of the main candidates for the 2018 presidential elections, Santos and colleagues (2019) point out that the characteristics and topologies of the network formed by the WhatsApp groups can contribute to an in-depth understanding of disinformation processes in an electoral context. From the intersection between network characteristics, such as the centrality of modularity, the authors propose an approach for mapping disinformation flows in the network constituted by WhatsApp groups.
13In addition to network analysis, the messages shared in these WhatsApp groups is also an important unit of analysis. Mont’Alverne and Mitozo (2018) investigate 213 public WhatsApp groups throughout the 2018 election period to identify how news was disseminated through links. Chagas, Modesto, and Magalhães (2019), in turn, monitored 150 public WhatsApp groups to understand how voters mobilize the meme “Brazil will become a Venezuela” and what types of framings were involved in these messages. Almeida and colleagues (2019) investigate the dissemination of fake news during the 2018 Brazilian elections, focusing on the analysis of images shared in public WhatsApp groups during the campaign.
14Another way to understand the effects of WhatsApp on the political process is to look specifically at users. Based on a survey of internet users in Brazil, Baptista et al (2019) examine how the use of WhatsApp contributes to the formation of public opinion about politics. They investigate which user profile is more likely to use WhatsApp to access or share political information, also seeking to understand how these users believe they are exposed to information they consider false in these digital environments. Another example is developed by Lemos and Oliveira (2020) in which the authors examine how an individual’s perception of their information competence affects how this same user understands the sharing of false information on WhatsApp. From a discussion on the third-person effect theory, they argue that respondents tend to perceive that those who share false information do not do so for self-interest, but because of inappropriate informational behaviors, which, according to the authors, would reduce the perceived social distance, since those who share deceptive content are supposedly not immoral, but a victim.
15Some researchers seek to understand the correlated effects of communication mediated by WhatsApp through its reflections on the media or the legal system. Jesus and Holanda (2019), for example, examine a news portal fact-checking initiative to understand what kind of strategies these media outlets use to guide its readers about the dissemination of false news on WhatsApp during the 2018 election for the Rio de Janeiro government. Pancho and colleagues (2019), on the other hand, analyze the positions adopted by the Superior Electoral Court to signal the theoretical insufficiencies of the agency’s actions in understanding key aspects of contemporary communication, more specifically, the idea of the audience as a commodity. In addition to this, Brazilian researchers have observed cases studies of WhatsApp as an online activism tool (Weber, Bastone & Bastos, 2017; Santos, 2017; Silva & Nunes, 2019), as a means of promoting political participation (Chagas et al., 2019) or as a platform for political talks.
16The accelerated development of digital devices, platforms, and resources brings new challenges to political communication research. The unpredictable way in which digital communication has come to mediate different spheres of sociability today makes it difficult to formulate a canonical document on ethical requirements for social research in the digital age. The most interesting attempts seek to escape the “checklist” model, indicating general guidelines that should be evaluated according to the logic of each platform. Thus, research on WhatsApp has often been framed as netnography or digital ethnography, and should follow ethical, innovative, and up-to-date parameters (Barbosa and Milan, 2019; Franzke et al, 2020). In Brazil, some regulations can serve as a guide for developing ethical parameters for research on WhatsApp. The resolution of the Brazilian National Health Council No. 196/96 establishes the regulatory standards for research involving human beings and the ethical requirements of scientific research, that include: I) the free and informed consent of individuals involved in the research; II) balancing between risks and benefits, minimizing or, if possible, eliminating harm and risk, maximizing individual and/or collective benefit; III) social relevance of the research, focusing on its socio-humanitarian nature.
17We argue, therefore, that studies on WhatsApp should consider the ethical parameters of social research that can be systematized into four central points:
-
strategies adopted for data collection;
-
concerns regarding privacy and access to sensitive participant data;
-
debates about the effectiveness of informed consent forms (ICFs);
-
protocols on the posture of researchers in groups on instant messaging platforms.
- 10 API is an interface developed by the platform itself to allow access to its functionalities or some (...)
18WhatsApp is a closed interpersonal communication platform, which means that there is no API (Application Programming Interface)10 that allows researchers to access data about interactions between users and groups. Despite being considered a notoriously opaque platform (Caetano et al., 2018), researchers have adopted different strategies and practices to obtain information about users’ actions in groups on WhatsApp. The main challenge of the methodologies employed is to structure research designs that mitigate the biases generated by the closed architecture of the mobile app. In general, it is possible to divide the means of data collection between those that accumulate messages through collaborators’ sharing and those that extract messages directly from groups in which the researchers participate.
19The voluntary sending of messages by users who participate in groups is one of the means of obtaining data for research on WhatsApp. From this perspective, the researcher does not directly become a member of the groups observed but has access to the material for analysis through the collaboration of participants who forward the researcher content shared in their particular groups. This approach can be operationalized through a number registered in the application that receives messages forwarded voluntarily by collaborators, or by ceding the message archives of different users. The first case was used in research during the electoral period in Brazil, as in the study by Cruz, Massaro, and Borges (2019) on the spam that circulated through WhatsApp, and in Piaia’s (2019) monitoring of communications through the mobile app. As for the second case, there is research by foreign authors such as Petitjean and Morel (2017), who, from messages from 43 collaborators, analyzed the uses of laughter in WhatsApp messages, and such as Rosenfeld et al (2018), who examined communication patterns on the app through messages from 111 volunteers. Both of these researchers received the conversation archives from these contributors and performed the analysis without becoming participants in the surveyed groups.
20This type of approach allows the formation of highly heterogeneous databases. This means of collection favors analyses centered on users or the contents in circulation, even though it is not possible to identify the sharing flows. Regarding permissions, an active relationship is established between the collaborator and the researcher, with the former previously consenting to data exploration. Although procedures of anonymization of the interlocutors are carried out, it is possible to point out tensions in relation to the consent of third parties involved in individual or group conversations provided for research.
21On the other hand, the direct extraction of messages from public groups implies that the researcher enters the analyzed groups. Public groups are considered those that are accessed through an invitation link circulating on the web, on social media, or other public groups. The collection procedure consist, in general, in the following steps: search on Google or on pages of social media websites such as Facebook and Twitter for invitation links to join groups on specific themes, filter groups of interest for the investigation from specific parameters (minimum number of participants, for example), collection of general information about the group (name, description, number of participants, number of moderators, etc.).
22The methodologies adopted and data analysis depend on the number of groups in the sample. A first set of research encompasses efforts that seek to map a small number of groups, through qualitative methodologies such as netnography and participant observation. The work of Cesarino (2019) exemplifies this type of approach, with the analysis of groups supporting President Bolsonaro accessed from September 2018, during the Brazilian election campaign. The smaller number of groups makes possible an approach that resembles that of classical ethnographies, in which the researcher’s presence is communicated and assented to by all (or at least the administrators of the groups).
23There is also research that deals with a large amount of groups (Resende et al. 2018; Caetano et al., 2018; Mont’Alverne & Mitozo, 2018; Piaia & Alves, 2018; Resende et al., 2019; Santos et al, 2019); Chagas, Modesto, & Magalhães, 2019). Collecting in many groups imposes some methodological challenges for extracting the messages, which causes most work to employ programming techniques to streamline and structure the process. Content analysis, quantitative approaches, and network analysis methods are often applied to the data collected. Public groups have been the means of data collection that most informs research on WhatsApp in Brazil, and the uses of these environments in scientific research have raised debates involving researchers’ attitudes towards the medium, the data circulating in them, and the participating users. These issues will be discussed more carefully in the next section.
24It is important to point out that all these forms of collection are crossed by the challenge of selection biases since the closed architecture of the application does not allow the measurement of a reliable sample from a universe of interactions. WhatsApp social research, therefore, should be careful when making generalizing statements.
25In addition to data collection, another issue that emerges in internet studies is the ability to store, manage and preserve data online. Although digital traces have been a rich source for social research in recent decades, the massive amount of data coming from digital platforms poses new challenges for scientific research. The ephemeral nature of certain online content, the high costs of the technical infrastructure for maintaining datasets, and the difficulties of creating a secure digital environment for data storage represent new challenges for the production of knowledge and the reproducibility of scientific research. Barats, Schafer, and Fickers (2020) argue that these challenges are not limited to the technical dimension (software, tools, and datasets), but also include wider institutional contexts, epistemic traditions, and social practices in which the doing of research in humanities and social sciences.
26To face such challenges, some researchers suggest the elaboration of a data management plan as a strategy to describe the entire life cycle of a research project, including the selection, curation, description, analysis, and interpretation of data but also its forms of publication and long-term storage. The main objective of this document would be to ensure the applicability of the FAIR principles (Findable, Accessible, Interoperable, and Reusable) in the field of humanities and social sciences.
27The second point is about the exposure, privacy, and processing of data of the WhatsApp group’s participants. According to manuals on ethics in research on social media, the identity of those observed must be kept confidential (Townsend; Wallace, 2016). It is also important to extend this protection to the names and telephone numbers of WhatsApp users. Thus, the large-scale monitoring protocol of WhatsApp groups’ studies must contain: i) anonymization of users from the source of the data; ii) presentation and manipulation only of aggregated data, which means analyzing the behavior of groups and users together, without detailing, possibly invasively, the performance of some specific users. For these procedures, the strategies and technical infrastructure for data collection and processing become central for the anonymization actions.
28In this sense, the WhatsApp studies need to describe the technical means of collection and anonymization, with an emphasis on the moment of anonymization and to inform whether the process takes place in an automated or manual way. Technical means of extracting and anonymizing data become even more relevant given the sensitive data protection requirements in the General Data Protection Law, Brazil. The Brazilian regulation model on data protection defines that anonymized data does not receive the protection granted to personal and sensitive data. Thus, the holder’s consent is waived for the data to be collected and processed, which allows them to be the object of research. The same article includes the caveat that the technical means of anonymization cannot be easily reversible, so that techniques used must have a high degree of security, at the risk of future liability of the researchers involved.
29The third point about ethical issues in research on WhatsApp is the complexity in obtaining the ICF from the participants observed by the research. It is a document that details the rights, procedures, risks, and benefits associated with choosing to participate in the research in question. By signing this form, the individual being observed becomes aware of how the researcher will use the data collected through the study. According to the Resolution nº 510/2016, published by the Brazilian National Health Council, the record of consent or assent of research participants can be made on paper or by other formats such as audio, video, or electronic media, being the form of record chosen from the individual, social, linguistic, economic, and cultural characteristics of the research participant and due to the methodological approaches applied. In 2020, the Brazilian National Health Council, more specifically, under the National Research Ethics Commission, published a guide that systematizes the rights of research participants concerning the information of the study clearly, the opportunity to clarify their doubts, the time necessary to make an autonomous decision, the freedom of refusal to participate in the study, the freedom to withdraw their consent at any stage of research, the ensuring of the confidentiality on their data, the ensuring of their privacy, among other rights11.
30As mentioned, there is an important difference concerning the procedures adopted according to the number of groups accessed. In surveys involving a large number of WhatsApp groups, however, the constant in and outflow and a large number of members makes it impossible to delimit a controlled universe for consent to be sought. In a public group - that is, a group with an open-access link - with a large number of members, it is common to observe dozens of entries and exits per day. In this sense, even authorization from the group administrator would not be equivalent to individualized consent.
31The waiver of informed consent is, therefore, a precondition for large-scale research in WhatsApp groups, being validated by national and international ethics committees (Garimella and Eckles, 2020). Beyond this issue, the observation of WhatsApp group dynamics is similar to the observation in public spaces, so the unit of analysis is the location (in this case, the group) and not specific individuals.
32There is a relevant discussion about the public or private nature of public WhatsApp groups. For this definition, the following question is usually used: Does the user expect to be watched? (Williams et al. 2017). WhatsApp groups are divided into those that are completely closed (such as family groups) and groups that have open invitations for a limited general public. Groups with open invitations circulate on the web, social media, and in other WhatsApp groups, allowing access to any user who has the entry link - within the limit of 256 members. In addition, to use WhatsApp, it is necessary to agree to the platform’s terms of use, which expressly states that messages exchanged in groups will be stored on the devices of all participants (Garimella & Tyson, 2018). There is even an option provided by WhatsApp itself for this data extraction. Thus, it is understood that users have enough information to know that they may be being watched.
33The fourth point about ethics in research on WhatsApp permeates all previous aspects and concerns the researchers’ attitude towards the group’s observation. It is still unclear how researchers should behave when they choose to research interaction environments focused on interpersonal chats. There are two usual researchers’ attitudes. First, on the one hand, some present themselves when joining conversation groups, announcing that the research has been carried out, and requesting consent from all group members. On the other hand, some researchers choose not to notify or request consent from group members. The justification for this is that the behavior of individuals targeted by the research would be altered if they knew that their actions were being the focus of a scientific investigation.
34This second type of approach is based on a protocol from the health sciences area that deals with the so-called “covert research”. Covert research is a type of observation “conducted without the participants being informed about the objectives and procedures of the study, and without their consent being obtained before or during the research”12. This type of approach is subject to the approval of the research by the Research Ethics Committee of universities and institutes, following Resolution No 510 (Brazil), of April 7, 2016. The objective is to ensure data confidentiality, seeking the consent of the participants later when possible. In the case of the adoption of the covert research method, the absence of interactions between the researcher and the participants is recommended, under the risk of incurring in experiments and inductions without the consent of the individuals observed. One solution to prevent the research from being attacked is to insert the name and logo of the institution or research group that coordinates the research. These parameters enable the development of research in places hostile to the work of researchers, such as radicalized political conversation groups and negationists.
35This paper aims to present a set of understandings and procedures required in WhatsApp studies. This is not a definitive framework but a systematization effort to support researchers in social sciences fields. Thus, we compiled the main characteristics of this app to indicate which ethical issues must be considered in research planning, regarding scientific methods. The influence of interactions established in the application for understanding social and political dynamics–especially in groups–makes the Brazilian case fertile for reflection on research techniques and procedures. Dialogue with international ethical standards also allows the elaboration of reflections that contemplate and debate the critical aspects for social research in these types of platforms.
36Even though further critical elaboration is necessary with regard to the ethical limits of research, especially in the context of social hyperconnection, in which the domain of information, in general, is restricted to some groups of power. It is noteworthy that those who have resources, technical domain and therefore hold information are very differently socially positioned in relation to those who are the object of observation in these digital environments and, to some degree, susceptible to the influence and control of these groups, specifically governments and companies. In this way, the complexity of different context-aware social realities in which very unequal power relations can be crossed, must be taken into account when questioning the limits of public and private, for example. Furthermore, the practices of transparency and even democratic guidelines that should govern these processes are other relevant factors to be considered by researchers interested in understanding these social phenomena that often shape behaviors, attitudes, and preferences which consequently also shape the political configuration more broadly.
37As a private and open app, WhatsApp’s studies always have several limitations in empirical analysis. For this reason, the research results are also a consequence of collective and accumulated knowledge from previous studies, especially in dialogue with different academic perspectives. In addition, the constant and non-informed app’s updates impact their affordances, architecture, and functionalities, which create new challenges for the establishment of rigorous and responsible research proposals.
38Future studies may focus specifically on the research practices adopted by scholars who have developed studies on WhatsApp. For example, by conducting surveys or semi-structured interviews with the authors with the aim of understanding the theoretical, epistemological, and methodological challenges of studying communicational processes mediated by WhatsApp. The studies may also focus on the differences between WhatsApp and other increasingly prominent instant messaging apps in the Brazilian context, such as Telegram. What are the implications of the different affordances of instant messaging platforms on social research and what kind of ethical issues emerge from this comparison are questions that may be subjects of future research.
Flux RSS
