Consensus days

Our research centre conducted two consensus sessions, which were constructed to discuss and debate the complex ethical challenges embedded in SM research. The first session involved seven researchers from within the unit who had a range of experiences of using SM sites for personal and professional purposes; usually, dissemination of team activities, although with no prior experience of using such sites to carry out research. The discussion was guided by questions generated by a review of the literature, which was conducted by the group facilitator and circulated to participants in advance of the first meeting. The second meeting further explored the issues raised in the first session, and also included external researchers with a strong background in SM research. Our concern did not mirror the recruitment focus of Comabella and Wanat,3 but instead reflected those issues raised in Zimmer's work9 around the ethics of using SM, specifically in relation to privacy and anonymity. This meant that the two consensus-day discussions placed far greater emphasis on establishing our position over the debated status of the internet and internet data (public/private), which encompassed a focus on the intentions of SM users, and how these should be incorporated into what we considered to be ethical research. Considerable debate and discussion took place during these consensus meetings, addressing the pros and cons of SM research. An overview of guidelines constructed for the conduct of ethical internet research informs the conclusion of the paper.

The internet, a public or private space?

The spatial way in which the internet is viewed has far-reaching consequences on the way that the intentions of the SM user are perceived and any research subsequently conducted. The questionable spatial status of the internet is widely debated and no distinction can easily be drawn between what can be considered public and that which is perceived as private. The accessibility of online communities has been likened to that of television or newspaper interviews10 which suggests that online communities can be considered to reside in the public domain. The journalistic media certainly do not hesitate in publishing publicly available microblog uploads from celebrities and politicians, and comments from the general public on healthcare debates are used in numerous articles and reports. Alternatively, the expectations of where and what internet users are communicating is considered to be the dominant ethical consideration for researchers.10 Comabella and Wanat3 addressed this point when they highlighted a subtle difference between the internet forum and other SM platforms such as YouTube or Twitter where the latter are created with the intention of reaching a large public audience. However, this can be contrasted with Frankel and Siang's position, that although communications may be easily accessible, many SM users would not consider their posts as potential research material, and accordingly, for research purposes, their posts should be considered private.10

Further complicating this debate are those scholars who suggest SM be approached as neither public nor private, but instead, a space that exists somewhere between the two; a ‘publicly private’ and ‘privately public’ space. Moreover, it can be argued that the internet should not be viewed exclusively as a site of human communication and interpersonal interaction but as a means of cultural production that can be considered in its textual form.11 In line with that approach, researchers need not consider the internet in ‘spatial’ terms at all but can, instead, focus on the dialogue that is extracted from the internet as text that is available for reproduction and public consumption. Concerns with the text-based approach can, nevertheless, be seen in the need for researchers to navigate their way through issues of ownership, removal and copyright,11 however, the considerations seen in spatial arguments are less apparent within this framework.

The way in which SM are understood, be they public/private and/or text based, will have a direct impact on the way that researchers frame the position of the participants themselves, and this will, accordingly, shape the way that the research is conducted in its entirety.3 The impact of these spatial/textual discourses is, therefore, embedded in subsequent methods of recruiting participants; approaches to obtaining informed consent; ensuring confidentiality; the questionable need for authenticating participants’ identities and the data they construct; and dissemination of the work that retains participants’ confidentiality. Comabella and Wanat3 urge researchers to establish their position in relation to these points prior to beginning research in order to ensure the protection and privacy of participants.

Our consensus groups also considered the spatial position of the internet to be crucial to the way that the researcher would perceive the forum they were approaching, the users of that forum and the data that could potentially be drawn from that particular form of SM. This meant that the questionable spatial status of closed groups, open groups, historical posts and bloggers were issues that all required individual discussion, as were the intentions of the SM within those groups.

Status of open and closed groups

Closed groups were defined as those that a user needed to join prior to gaining access (such as Facebook, user support groups and so on). This kind of group was initially considered to occupy a private space for two reasons:

• Owing to the privacy settings that administrators/users can employ to make communications available to a specific group of people;

• Because users within closed groups are liable to consider their communications private and will conduct their posts accordingly.

However, contrary to the idea that privacy settings create a private forum for its users, changes to security settings implemented by internet providers (eg, Facebook, Yahoo and so on) necessitate that both users and group administrators maintain constant awareness of the security status of the groups to which they belong. It would be naive to presume that this is undertaken by all SM users, and thus, closed and private groups may at different junctures become more accessible to a public audience. Similarly, the necessity to join a group does little in the way of creating a barrier to public access as anyone, regardless of their appropriateness to the group, can create an identity with which to join the online forum and view the ‘private’ communications with relative ease. The true extent to which any ‘private’ communications can be considered to take place in a private space is, therefore, once again raised. Closed groups would, for this reason, be best understood to occupy a perceived private space that is potentially vulnerable to a public audience.

Open access forums were considered by the consensus groups to be firmly grounded in the public space. Bloggers were perceived to be focused on disseminating their communications to a public audience, and for that reason their posts are widely accessible to a public and sometimes publicised space. It is important to recognise that a perceived intent for public dissemination does not automatically equate to a willingness to become a research participant, thus whether it is ethical to reposition this data in such a way remains arguable. An alternative approach to the public/private debate is the textual approach described above which suggests that spatial concerns about the internet could be avoided if one considered SM communications as neither public nor private, but instead as textual data. The consensus groups largely discussed this approach in relation to the use of historical data on the internet.

Use of historical data

Across the consensus delegates it was agreed that the text-based approach enables internet communications to be repositioned as text that stands alone from its author. Theoretically, it can be argued that text is not vulnerable to the same person-centred concerns that spatial perceptions involve, and thus, the text-based approach avoids complicated ethical quandaries. In support of this approach, one can identify parallels with its physical world equivalent where researchers draw from previously conducted research; however, physical world research is not vulnerable to the same complexities as that conducted online where cyber data is traceable to its original forum and author. Although a textual approach theoretically circumnavigates person-centred concerns (as it is simply text that is being discussed), for us there remains a discomfort about removing the authors from the text, and thus, concerns about anonymity and consent remained unresolved.

Illustrating this point was a concern around the lack of consent in the textual approach to historical data (or any other forum). It was suggested that in the equivalent ‘real world’ historical research, there would have been a consent process undertaken in the collection of the original material. That consent could, in principle, be seen to extend to the use of the data in future research projects. However, for a text-based approach in SM research where there is no original consent process, a physical world equivalent cannot be drawn; subsequently, the (internet) author's intentions and wishes would remain unknown. It would seem fair to presume that people do not communicate across SM with any consideration that their communications might be repositioned as data, and used for research purposes. Subsequently, out of consideration for the author and the sensitivity of the communications of interest in palliative research, the consensus groups concluded that a textual approach does not avoid the person-centred complexities embedded in the spatial perspective, and therefore, should not be considered an appropriate approach or way forward.

Recruiting participants and obtaining informed consent

Online recruitment can be achieved in two primary ways; prospectively, with the researcher making contact with participants directly, or retrospectively, with the researcher making a general request to an internet forum/wall for people to opt out if they do not wish to be included in the study. Both approaches are problematic; locating valid contact details for the recruitment of prospective participants can be an excessively time-demanding process, and due to the fluidity of the internet forum (where community users come and go), people may not be aware of retrospective recruitment relating to their online communications.7 In palliative research, retrospective consent is particularly problematic due to the likelihood that the person may no longer be alive. The method used to recruit participants will have a subsequent impact on the way that informed consent is viewed and approached.

Comabella and Wanat3 suggest that the direct recruitment of participants may not always be appropriate or necessary, and they cite the use of data from blogs and YouTube as evidence of this. We took this a step further and considered whether this would be the case where the researcher spends a period of time passively observing online communities (lurking). As previously suggested, where the internet is understood as a public domain, then postings and communications may be considered ‘fair research game’12 that do not require specific consent from community members. However, legal issues relating to ownership of data stored on the servers of private companies complicate this matter,3 and the sensitive nature of palliative research might make it ethically appropriate to seek consent from all research participants, regardless of the status of the space in which their communications took place.7 ,11 ,12 Where/if this is deemed to be the case, a comprehensive systematic process must be constructed in order to gain informed consent from online participants that aims to overcome the challenges embedded in this process.

The process of consent in the physical world involves researchers ensuring that participants receive all relevant information, (normally in the form of paper/electronic copies of information sheets) and making efforts to check that they have a satisfactory understanding of that information. Attempts to ensure that participation is voluntary, and to clarify comprehension of the study, can be measured to some degree by physical cues and markers; however, there is no equivalent of this in gaining online consent. The equivalent of a signature on a consent form would be the click of a box next to a statement agreeing to consent to the study,10 yet people regularly click statements of agreement without having read the information provided, and thus, the true extent to which their consent is ‘informed’ is difficult to establish.

The delegates in our groups were in agreement that researchers have a duty of care in the online research process, and it is our responsibility to ensure that this is in place, is appropriately adhered to (ensuring possible harms are avoided), and that our data-gathering intentions are transparent. For this reason, the delegates of our consensus meetings were in agreement that although SM could and indeed should be used as a research tool, its spatial position would best be understood to straddle both public and private spheres; private in that communications may have been constructed with a perception of privacy; public in that those communications are liable to be publically accessible. For this research centre, this means that the privacy of internet users should be a significant consideration in the conduct of SM research, and should therefore involve consenting internet users prior to the collection of their posts for research purposes. Where this is not possible, the research centre felt that the appropriateness of its use should be seriously questioned.

Anonymity in dissemination

When composing reports, qualitative research tends to draw heavily from the original transcripts (the SM equivalent being text from blogs/forum pages and so on), which is usually presented in the form of quotes. This is unproblematic in physical world research where the potential for identifying participants should be minimal; but for internet data, where there exists a very real potential for quotes to be traced back to the original author, it is considered to be an issue of significant concern. Three potential ways of approaching this problem were discussed in our consensus meetings:

• Rewrite what people say;

• Ensure the potential for identification is disclosed in the consent process;

• Consider the construction of online identities to be sufficient barrier to identification.

The benefit of rewriting online communications is that the original communications and forum cannot be traced through a search engine, thus, a degree of distance can be afforded between the text and the author. Although some researchers may not perceive this to be a problematic way of dealing with data, it can be argued that removing data from its original frame and rewriting it to suit research requirements will obliterate the original impact of the experience that was narrated, albeit in a textual form.13 This mirrors Ricoeurs concept of distanciation where every retelling, rewriting and reframing of the story creates an increasing distance from the essence, meaning and impact of the person's actual experience.13 As a research centre, we are keen to remain sensitive to the nature of our palliative research and want to ensure that, as far as possible, participants’ experiences are not lost within the reframing and retelling of their communications. For these reasons, the consensus meeting drew the conclusion that where possible, distanciation should be kept to a minimum and, instead, participants should be made aware of the potential for identification within the consent process.

Rather than attempting to avoid the issue of anonymity, it was suggested that it may be more appropriate to highlight the problem within the consent process, giving the participant the opportunity to decide the extent to which their online identity needs protecting, and thus, giving them the time to respond appropriately to their decision. It was also suggested that where there was a forum created specifically for research purposes, participants could create an identity that was unique to that forum alone. However, the creation of multiple identities is not always permitted across SM forums, and as such, this may not be a permissible means with which to proceed. Subsequently, our researchers conceded that issues of anonymity needed to be highlighted in the consent process, but online identities would then be considered to provide a sufficient level of anonymity to the participant's real-world identity.