Three main findings emerged from this metamethod: (a) missed opportunities for taking full and positive advantage of the social in social media; (b) differing assumptions about the social boundaries of public and private communication that occurs in virtual communities; and (c) the ethical dilemmas that emerge from these assumptions.
Missed opportunities
Some health researchers appear to be missing opportunities to capitalize on the social in social media. Social networking sites are, by design, a way to engage, collaborate, and interact with others in two-way or multi-dialogical conversations. Myspace originating in 2003, Facebook in 2004, YouTube in 2005, and Twitter in 2006 are common forms of social media for creating, sharing, and commenting on information, including health information. All selected studies took advantage of the accessibility of data that social media affords however only two took advantage of the unique opportunity to engage in dialogue with users in an established virtual community. One was Battles’ [
14] cyberethnographic study; the other was the study by Lee and colleagues [
17]. The latter study might be classified as interactive in that it encouraged users to post questions and receive responses. The authors drew conclusions about what type of information on HPV and HPV vaccines people sought when they consulted an internet site about HPV. However, we cannot know if posted questions represented an unknown piece of information crucial to HPV vaccination decision-making or a simple curiosity. It is unclear whether analyzing questions can adequately reveal a population’s understanding of and beliefs about HPV infections and HPV vaccination. With such specious assumptions, the value of the findings is questionable.
The remaining studies relied on traditional data collection techniques to generate findings. Data were analyzed using positive vs. negative scoring systems (sometimes including an ambivalent or neutral category) to describe HPV awareness and attitudes toward HPV vaccination. Generally the study designs were sound and produced credible findings about attitudes toward HPV vaccination. What was unclear was how valuable this knowledge is to those of us interested in developing public health policies and interventions. While the statistical surveillance used in the majority of these studies offers some sense of users’ opinions, the scoring systems strip data of the context that makes these beliefs and opinions meaningful and fails to probe users’ understanding of HPV vaccines that underpins the posts. Moreover, statistical analysis of such data that intends to measure trends and patterns (in a process similar to conducting a survey) does so without the benefit of randomized sampling and other design elements for ensuring generalizability or reproducibility of findings.f
The tone and quality of discourse fluctuates in dynamic ways across various virtual communities. Health researchers must be cautious in making assertions about the wider population based only on a website’s viewers and posters – groups that likely have a vested interest in a given topic and that are sufficiently comfortable with using technology to post their opinions on social networking sites. One cannot assume that data derived from social networking sites are representative of the broader population because of the self-selection bias of users. Thus, the relative value of knowledge generated from such studies seems limited.
We also wonder about the relative value of knowledge produced through uni-directional studies, including some of the selected articles reviewed in this metamethod that were designed to produce objective findings. Can such studies enhance our understanding of users’ subjective opinions, values and choices about an important health concern? Even the relatively well-executed statistical analyses of uni-directionally mined data – as was the case in the majority of the selected studies – cannot render more visible the complexities of HPV awareness or HPV vaccination decision-making. There was no explicit question posed to elicit a response as happens with a survey where parameters and variables are defined. This makes it unclear what precise stimulus participants were responding to when they posted a “thumbs up/thumbs down” or “like/unlike” to videos or blogs. A positive response could just as easily reflect appreciation of visual representations as it might a positive reaction to the content. Therefore, research that investigates public awareness of HPV and HPV vaccination (and other health issues) must reflect this complexity in ways that go beyond a tabulation of binary responses.
This metamethod suggests that traditional data collection techniques may require modification for use in a virtual environment. Moreover, traditional data mining fails to take full advantage of the dialogical nature of social media. Researchers who make full use of this feature might generate richer understandings of the complexities of users’ knowledge of HPV infections and HPV vaccination. Collaborative engagement with users, as in the Battles cyberethnography, acknowledges the shared authority and expertise of the researcher and the researched. At the risk of re-entrenching the quantitative-qualitative debate in health research [
20,
21], we are left to wonder if it might be more valuable to embrace and explore the subjectivity of communication in online communities rather than hold to a research paradigm that seeks an objective truth [
22].
Conceptualization of virtual communities: the public versus private binary
According to the Association for Internet Research, the internet is “a social phenomenon, a tool, and also a (field) site for research” [
23]. The same might be said of social media. The rapid growth of social media over the past decade has been nothing short of phenomenal. Users post commentary and share information about a range of topics. Organizations are increasingly using Twitter, Facebook, and other social networking sites to perform health research [
24]. What is less clear is how researchers conceive of this field site and if these conceptualizations are consistent with those held by those who operate in these spaces.
“Networked publics,” a term coined by boyd and Marwick, refers to “(1) the space constructed through networked technologies and (2) the imagined community that emerges as a result of the intersection of people, technology, and practice” [
25]. When considering the specific case of HPV awareness and HPV vaccination, social networking sites attract youth and those who make decisions for youth. Medical experts and others with varying levels of knowledge may also participate in these sites with a goal of shaping public opinion, effecting social change, or promoting a particular public health intervention. If the research goal is to map public awareness and attitudes about HPV infections and HPV vaccination by collecting data from networked publics, the methodological approach must reflect the complexity of knowledge consumption and decision-making. These complex processes require fact-gathering, critical evaluation of evidence, and thoughtful weighing of pros and cons in the context of one’s own life. This means taking advantage of social media as more than a tool for mining data while still utilizing it as a field site for research.
As with accessing any research field site, attention must be paid to understanding the social boundaries operating in that physical or virtual space. To quote boyd, “[i]t’s about a collective understanding of a social situation’s boundaries and knowing how to operate within them” [
26]. Community members may post information or commentary in a public virtual space while maintaining divergent expectations for how those postings are used by others. The Association of Internet Researchers Ethics Working Committee writes:
Individual and cultural definitions and expectations of privacy are ambiguous, contested, and changing. People may operate in public spaces but maintain strong perceptions or expectations of privacy. Or, they may acknowledge that the substance of their communication is public, but that the specific context in which it appears implies restrictions on how that information is – or ought to be – used by other parties [
23].
The Committee concludes that the “[s]ocial, academic, or regulatory delineations of public and private as a clearly recognizable binary no longer holds in everyday practice” [
23].
Ethical issues
How do health researchers engage in ethical research in virtual communities? This important question has been taken up by individual scholars [
22,
24‐
26] and by the Association for Internet Researchers [
23]. The answer seems to pivot on the ambiguity about social boundaries and the application of the public/private binary in networked publics. This metamethod suggests that some health researchers are not addressing these concerns. Five of the six studies were based on the assumption that social media was an accessible source of unfiltered information about public attitudes toward HPV vaccination as these studies aimed to characterize public attitudes about HPV and HPV vaccinations. The public accessibility of data seemed to justify its use even if no consent was obtained from users or service providers that data were being mined for research purposes. However, boyd and Crawford assert that “[j]ust because it [data] is accessible does not make it [using the data] ethical” [
22].
Traditionally, if research involved human subjects, it had to undergo ethics approval. Since the adoption of the Nuremberg Code in 1947, research ethics guidelines have required that human subjects must be protected in health research. The Code was drafted in reaction to the ethical implications of the Nazi medical experiments and emphasized the importance of informed consent [
27]. However, the Association of Internet Researchers contends that the concept of human subjects is not well suited to research that does not involve biomedical interventions or research that interacts indirectly with people through archival materials or other published texts [
23]. Instead, the Association recommends focusing on concepts of harm, vulnerability, personally identifiable information, and other such concepts [
23]. This recommendation is consistent with the Tri-Council Policy Statement 2 (TCPS2),
g which sets out three principles for ethical decision-making: concern for welfare, respect for person, and justice [
28].
The potential for harm is covered under the principle
concern for welfare, or “the quality of that person’s experience of life in all its aspects” [
28]. Under this principle, researchers have an obligation to both assess and minimize risk while balancing risks and potential benefits. Accordingly, those individuals whose data are included in a study have a right to make the final judgment about the acceptability of this balance. It cannot be assumed that using social media means that individuals have forfeited their right to make this judgment. In five of the six studies, individuals were not informed and did not consent to having their data collected for research purposes. Nor was there evidence that the researchers considered the potential harm to individuals whose data were used in a way they might not have intended or imagined. In the sixth study, Battles [
14] obtained informed consent acknowledging that attempting research in a virtual community is difficult because participants may wish to remain anonymous to the researcher and because the transient nature of social media participation may make it impossible to track users. This prompts the question: does mining data without informed consent constitute a risk of harm? This question seems especially salient given the sensitive (even taboo) nature of cultural attitudes toward HPV infections and vaccinations in some networked publics where data were mined.
The second principle identified in the TCPS2,
respect for person, involves the moral obligation to respect and protect the autonomy of research participants as well as those individuals whose data are used for research purposes [
28]. Obtaining free, informed, and ongoing consent is an accepted mechanism for doing this. The TCPS2 devotes minimal attention to research in virtual settings, equating these spaces with observational studies conducted in natural settings “where people have a reasonable or limited expectation of privacy” [
28]. In these situations the researcher must justify the exception to the requirement to obtain consent. Our metamethod indicates that ethics approval was only sought for the cyberethnography, leading us to think that researchers involved in the other five studies assumed that virtual communities were public spaces where users had no little or no expectation of privacy. There was an absence of discussion about the potential privacy incursion even when users’ comments were cited. Failing to engage users about the subsequent use or interpretation of data they posted may have unintended negative implications. For instance, Ache and Wallace directly quote negative and positive viewer-posted comments about HPV videos appearing on YouTube [
18], making it possible to identify the accounts of those individuals whose comments were cited.
When researchers publish users’ verbatim comments in scholarly articles, it is possible for readers to reproduce the quotations in a search engine that can link them to an individual’s account on the sampled social media website. Using this search method for an article that quotes YouTube comments, article readers can glean information about a willing or unwilling study participant by reviewing the “about” information in his or her public profile or by analyzing the videos followed by the individual. The potential exists for article readers to further identify such individuals by using the social media website as a platform to contact the person through messaging or review his or her posted content, which may include personal videos or photos.
On the other hand, researchers’ efforts to protect online study participants by not presenting direct quotations as part of the data set can negatively affect the trustworthiness and credibility of results. The study that paid most attention to the ethical protection of users (Battles [
14]) provided little evidence on which we could evaluate the quality of design or assess the rationale behind interpretations and conclusions about users’ understanding of HPV vaccination. To better maintain the principle of respect for person while presenting verifiable data, we recommend that researchers paraphrase users’ comments. If studies presented a topical but fictive sample quotation alongside the resultant paraphrase, then readers could more easily evaluate the credibility of the results by assessing how meaningfully authors can capture opinions in paraphrase.
Justice is the third principle listed in the TCPS2 and refers to fair and equitable treatment [
28]. Power, especially the power differential between the researcher and the researched, must be considered. Where the power imbalance is abused, there is a significant threat to justice and the potential for harm [
28]. In many jurisdictions, youth are considered vulnerable and deserving of special consideration because of their limited access to rights, opportunities and power [
28]. Youth are both eligible candidates for HPV immunization and a population that is technologically capable of using social networking sites to seek and share health information [
7]. Therefore it is reasonable to assume that youth may make up a portion of the users drawn to HPV information content. Youth and other vulnerable populations might not understand research in the same way and to the same degree as a researcher, particularly when privacy and self-disclosure may have implications beyond the immediate moment. Some cultural, social and religious groups should also be afforded special attention to ensure they are treated justly in research. This might be said of Korean users who visited the Q & A website that was established precisely because conversations about sexual behavior were considered taboo in a “Confucianism-based culture” [
17]. With the exception of the Battles study [
14], there is no evidence that researchers considered the expressed needs and concerns of the participants whose data they collected or whether mining data might harm potentially vulnerable users. Researchers are also at risk of misinterpreting or misrepresenting users’ comments. In the absence of dialogue it is difficult to assess if the participants would evaluate the researcher findings as accurate interpretations of their experience.
Given the relative newness of virtual communities, researchers and institutional ethics review boards must work together to develop expertise in evaluating the design of studies undertaken with virtual communities. Currently, ethics board review and approval is required only for research with “live” human subjects and is not required for research in traditional public spaces. As Zimmer points out, institutional ethics review boards lack the expertise to consider the potential privacy incursions or risk of harm conducted on social networking sites [
24]. The Association for Internet Research developed some guidelines in 2002 with a more recent revision in 2012 [
23]. However, if this metamethod is any indication, these ethical dilemmas and the guidelines for conducting ethical research in social networking sites are not widely known or implemented. Indeed, when we presented our initial findings at an academic conference in 2012, audience members dismissed the ethical concerns we identified calling up the public square argument. The public square argument used by research ethics boards assumes that ethics approval is not required when research involves the observation of people in traditional public spaces. When this guideline is extended to online spaces, it is used to assert that that individuals posting data on a social networking site are inviting public engagement and thus cannot reasonably hold an expectation of privacy. The users’ intended use of the public space (i.e., to share information and engage in public discussion) does not reasonably extend to consent for another purpose such as research. The public square argument does not address the concern that data are being used for a purpose other than what the users may have intended with an attendant risk of harm.
The application of the core principles of concern for welfare, respect for person, and justice must extend to networked publics (as both virtual spaces and the users who occupy them) with explicit justification provided for not seeking informed and ongoing consent if social boundaries are unclear. This ethical approach to engaged and collaborative research will help to build and maintain the trust of participants and the public in the research process and ensure that the benefits of knowledge production are shared [
28]. These findings seem especially relevant considering the public backlash against Facebook following their mood manipulation study [
29].