Savage Minds welcomes guest blogger TAZ KARIM
In the past five years, Twitter has become a mecca for social science researchers: the number of topics, informants, and networks waiting to be analyzed are limitless (here are some examples). With the help of a nifty program like Tweet Archivist, you could literally collect thousands of micro-narratives about people’s ideologies, behaviors, and relationships around a search query – all from the comfort of your office. This was the utopian vision I had of Twitter research when I started designing my final project for the Cultural Heritage Informatics (CHI) Fellowship at Michigan State University (link has been fixed!).
Over the last year, I have become interested in how Americans are sharing experiences with prescription drugs through social media. My dissertation look at one drug in particular, Adderall, a treatment for ADHD which is being illegally bought and sold by college students for academic and recreational purposes. At first, I was completely shocked by the sheer number of individuals who are openly admitting their illicit drug use online – after all, many twitter names are publically attached to an individual’s real name. More amazing was how many are intentionally categorizing their tweets using hashtags like # adderall or #adderallproblems, so that people interested in the topic (like myself) could easily find and share their tweets. There are even entire accounts dedicated to the “adderall lifestyle” like @adderallavenger @adderallnation @adderalltalking @addiestories… and the list goes on. I felt like I had hit a goldmine of data – now all I had to do was figure out how to harvest it.
One of the biggest issues with collecting twitter data is that you can only go back so far before Twitter clears its history– which means time is your enemy. Initially, I had no idea how to capture these tweets so I began taking screenshots and collecting the most interesting ones. As you can imagine, this got really tiring and I thought to myself “there must be a better way to save these”. Being relatively new to Twitter, I looked around and discovered that you could “favorite” a tweet and save it to your account. I started favoriting tweets left and right for about a week, with the intention of screen-capturing them later on when I had the time to work on it. What I failed to realize was that when I favorited a tweet, the author of the tweet received a notification about it. All of a sudden, I went from being a silent observer sitting in my office in East Lansing, MI, to becoming a tangible, visible part of each of their worlds… all because I thoughtlessly hit a button.
In general, most authors of the tweets I favorited didn’t respond. Several of them started following my account and retweeting some of the articles I tweeted about. A few even messaged me and told me how cool they thought my research was. However, there was one individual whose tweet I favorited who had a different reaction. I remember favoriting the tweet because I thought it contained really important data on how the desire for Adderall was facilitating doctor-patient relationships:
The next morning, the author of this tweet had sent me a direct message indicating that she felt like I was judging her as a “drug addict” and that somehow my intention was to use her tweet to argue that Adderall should be banned… she inferred all of this from the simple fact that I had favorited her tweet and taking a quick glimpse at my own Twitter account. I quickly sent her a message apologizing and trying to clarify the intent of my project and that I was in no way passing judgment on her. She didn’t respond so I couldn’t tell you if she understood or accepted my explanation.
Initially, I was really upset by this experience for several reasons. Why would this person assume I was anti-Adderall or that somehow I was trying to take away her access to medication by doing this project? More importantly, didn’t she know that these tweets are public? If she didn’t want people reading them or “judging” her, why did she put them online in the first place?
At the following CHI meeting, we discussed this incident as part of several broader issues. To begin with, how do we establish ethical boundaries in social media research, especially around sensitive topics like mental illness, academic dishonesty, and illicit drug use? One project that helped us think through these issues was the site NoHomophobes which does an excellent job visually representing the frequency in which people use homophobic language online. Despite the importance of the project, is it ethical to decontextualize a single tweet and attach an individual’s identity to a site dedicated to exposing homophobia?
The positionality of anthropologists to their informants was another important consideration. While participant observation is a key method in anthropology, it raises concerns of how the act of observation itself can influence the subjects of study. Once the author of a tweet knows they are being observed, will they change their tweeting habits? Will they now be more careful about what they say online? Is this necessarily a bad thing?
This leads me to my final point which is that social media research really highlights the ambiguity between public and private spaces. While I am not being personally invited into their corner of Twitter and I may not be the intended audience for the tweet, I am not illegally invading it either because tweets are public. This concern forces me to weigh the value of collecting unadulterated data against a gut feeling that my participants should know and understand their role in my larger project, even if it is a small one.
As a result of these concerns, I stopped favoriting tweets and moved to other methods of data collection which allow me to harvest tweets both quickly and anonymously. While I have circumvented some of the issues mentioned in this post, I am still struggling to navigate the waters of social media in a way that leaves both my informants and myself feeling positive about my research.
Stay tuned for my next post on which will elaborate on the methods I am currently using to collect data from Twitter and Instagram…