Throughout this study I aimed to maintain an ethical sensibility which responded to issues as they arose. Within the thesis, I highlighted those areas which triggered <#ethics> concerns and now return to summarise them. Here I’ll set out some of those observations and the way I responded.
One complex arena which tested my ethical sensibility was in the degrees of subtlety required when conducting online interactions. Simply asking a question, whether on Twitter or through a blog, obliged me to communicate my status as a researcher. Taking a cue from the norms of Twitter, I chose to include the hashtag #4MyResearch in each encounter, as described in an earlier post. It was later that I realised this could be a double-edged sword. Hashtags are of course searchable, so #4MyResearch could be one mechanism – albeit somewhat blunt – through which to bring one strand of my research together. Anyone could then check through the hashtag and view the way I had previously interacted with other people. That of course then makes each encounter more public than it otherwise might be, something of which a respondent might not initially be aware if they lose sight of potential audience in what Marwick and boyd (2010) term ‘context collapse.’
Before interacting on Twitter, my ethical sensibility became heightened. Responding to a single isolated tweet seemed different to jumping into a discussion which was already underway, even if that exchange was within the already discursive space of a hashtag chat. Imposing my research agenda in a space where the activity was altogether different felt intrusive. To ameliorate this, I instigated a ‘wait-time’ to allow the proceedings to conclude. I felt that asking my question(s) later was at least an act of politeness and hopefully meant I wasn’t interrupting someone active in other practice.
Guided by the norms and practices of participants (and following the guidance in Williams et al, 2017) the contributions of participants was acknowledged, rather than anonymised. Requesting permission to quote them in either a presentation or the thesis requires more thought than in a pre-planned interview where this can be incorporated within the consent-seeking process. I mostly avoided approaching participants in the public twitterstream to request permission to ‘publish’ their tweet, lest that bring undue pressure to bear. Whenever possible I made these requests through Twitter’s direct message service, or using other contact details if available. If I did have to make a request in public, I always offered the opportunity to reply in private.
Tweets feature within this thesis in a number of formats: as single tweets, brief exchanges and within timeline views. Permission for all single tweets and brief exchanges was sought from participants. In timeline views, identifying features were anonymised for two reasons. Firstly these views were presented to illustrate how they might support practices, rather than for the significance of their contents. Secondly, the larger number of participants made seeking permission from each of them impractical.
In presenting single, unaltered tweets, another concern became apparent. In some of the tweets, the author mentioned other individuals or unavoidably identified them when replying to tweets. I began to consider whether I also need to seek permission from that second level of participants. For this I found no guidance or precedence in the literatures and instead settled on a risk analysis which returned to some of the arguments in the ‘Ethics’ section of Chapter 4. I reasoned that those individuals mentioned in tweets were unlikely to suffer harm as a result of their name or Twitter handles appearing in tweets published in this thesis, given the subject matter being discussed. Had any tweets been discussing more controversial issues, I would have sought permission before publishing.
When new methods like the PLDbot emerged, I explored, rather than suppressed them for two main reasons: to attempt to open this area of study and reveal different knowledge from previous research, and to avoid producing a clean, purified account from practices which seemed anything but. As Fenwick and Edwards (2010) suggested,
Research methods are often designed to simplify the messy lumpishness and most interesting complications of the world, in well-intentioned efforts to know them and make things clear,’ and that our research may ‘distort or completely repress the very things we want to understand.’
For example, when the PLDBot was running in automated mode, it began to Like tweets from commercial providers of professional development. I hadn’t anticipated dealing with the <#ethics> of my bot becoming enrolled as a company’s advertising tool. Nor had I considered that anyone would want to interact with the bot by following it in the way that several people did. Although this concerned me at first, it also presented potential avenues for exploration, but which would also require a fresh <#ethics> submission, a route I couldn’t take within the time constraints of this study. Rather than ‘simplify’ the mess, the PLDbot seemed to amplify it, producing its own ‘interesting complications.’
These reflections highlight the need in unique circumstances such as these, to maintain an ongoing sensitivity to resolve issues at the stages they arise in a study. Furthermore, as Markham and Buchanan (2012) note, tensions may arise between top-down and bottom up approaches to ethics which ‘should be acknowledged and considered, even if there are no easy solutions.’
Fenwick, T., & Edwards, R. (2010). Actor-network theory in education. London: Routledge.
Markham, A. N., & Buchanan, E. (2012). Ethical decision-making and internet research: Version 2.0. Association of Internet Researchers.
Marwick, A. E., & Boyd, D. (2011). I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New Media and Society, 13(1), 114-133. doi:10.1177/1461444810365313
Williams, M. L., Burnap, P., & Sloan, L. (2017). Towards an ethical framework for publishing Twitter data in social research: Taking into account users’ views, online context and algorithmic estimation. Sociology.