At a research ethics workshop at the 2015 CSCW conference (Fiesler et al., 2015), researchers in our community respectfully disagreed about using public social media data for research without the consent of those who had posted the material. Some argued that researchers had no obligation to gain consent from each person whose data appeared in a public social media dataset. Others contended that, instead, people should have to explicitly opt in to having their data collected for research purposes. The issue of consent for social media data remains an ongoing debate among researchers. In this blog post, we tackle a much smaller piece of this puzzle, focusing on the research ethics but not the legal aspects of this issue: how should researchers approach consent when including screenshots of user-generated social media posts in research papers? Because analysis of visual social media content is a growing research area, it is important to identify research ethics guidelines.
We first discuss a few approaches to using user-generated social media images ethically in research papers. In a 2016 paper that we co-authored, we used screenshots from Instagram, Tumblr, and Twitter to exemplify our characterizations of eating disorder presentation online (Pater, Haimson, Andalibi, & Mynatt, 2016). Though these images were posted publicly, we felt uncomfortable using them in our research paper without consent from the posters. We used an opt-out strategy, in which we included content in the paper as long as people did not explicitly opt out. We contacted 17 people using the messaging systems on the social media site where the content appeared, gave them a brief description of the research project, and explained that they could opt out of their post being presented in the paper by responding to the message. We sent these messages in May 2015, and intended to remove people’s images from the paper if they responded before the paper’s final submission for publication five months later in October 2015. Out of the 17 people that we contacted, three people gave explicit permission to use their images in the paper, and the remaining 14 did not respond. Though this was sensitive content due to the eating disorder context, it did not include any identifiable pictures (e.g. a poster’s face) or usernames. While we were not entirely comfortable using content from the 14 people who did not give explicit permission, this seemed to be in line with ethical research practices within our research community (e.g. (Chancellor, Lin, Goodman, Zerwas, & De Choudhury, 2016), who did not receive users’ consent to use images, but did blur any identifiable features). We ultimately decided that including the images did more good than harm, considering that our paper contributed an understanding of online self-presentation for a marginalized population, which could have important clinical and technological implications. Another paper (Andalibi, Ozturk, & Forte, 2017) took a different approach to publishing user-generated visual content. Because the authors had no way of contacting posters, they instead created a few example posts themselves, which included features similar but not identical to the images in the dataset, to communicate the type of images they referenced in the paper. This is similar to what Markham (2012) calls “fabrication as ethical practice.”
This opt-out approach is only ethical in certain cases. For instance, it is not in line with the Australian National Statement on Ethical Conduct in Human Research (National Health and Medical Research Council, 2012), which we assume was not written with social media researchers as its primary audience. NHMRC’s Chapter 2.3 states that an opt-out approach is only ethical “if participants receive and read the information provided.” In a social media context, people may not necessarily receive and read information messaged to them. Additionally, researchers and ethics committees may not agree on whether or not these people are “participants” or whether such a study constitutes human subjects research. When using non-identifiable images, as we did in our study described above, and when the study’s benefit outweighs potential harm done to those who posted the social media content, we argue that an opt-out approach is appropriate. However, an opt-out approach becomes unethical when sensitive, personally-identifiable images are included in a research paper, as we discuss next.
While issues of consent when using social media content in research papers remains a thorny ongoing discussion, in certain instances we believe researchers’ decisions are more clear-cut. If social media content is identifiable – that is, if the poster’s face and/or name appears in the post – researchers should either get explicit consent from that person, de-identify the image (such as by blurring the photo and removing the name), or use ethical fabrication (Markham, 2012). Particularly, we strongly argue that when dealing with sensitive contexts, such as stigmatized identities or health issues, a person’s face and name should not be used without permission. As an example, let’s say that a woman posts a picture of herself using the hashtag #IHadAnAbortion in a public Twitter post. A researcher may argue that this photo is publicly available and thus is also available to copy and paste into a research paper. However, this ignores the post’s contextual integrity (Nissenbaum, 2009): when taking the post out of its intended context (a particular hashtag on Twitter), the researcher fundamentally changes the presentation and the meaning of the post. Additionally, on Twitter, the poster has the agency to delete[1] the post at her discretion, a freedom that she loses when it becomes forever embedded into a research paper and all of the digital and physically distributed copies of that paper. Thus, we argue that when including identifiable social media data in papers, researchers should be obligated to receive explicit permission from the person who posted that content, should they wish to include that image in the paper.
[1] Though all tweets are archived by the Library of Congress and thus not fully deletable, they are not readily accessible by the public, and even by most researchers. Furthermore, Twitter’s Terms of Service require those who collect data to periodically check for and remove deleted tweets from their datasets, though it is not clear whether this applies to the Library of Congress (Twitter, n.d.).
References:
Andalibi, N., Ozturk, P., & Forte, A. (2017). Sensitive Self-disclosures, Responses, and Social Support on Instagram: The Case of #Depression. In Proceedings of the 20th ACM Conference on Computer-Supported Cooperative Work & Social Computing. New York, NY, USA: ACM. http://dx.doi.org/10.1145/2998181.2998243
Chancellor, S., Lin, Z., Goodman, E. L., Zerwas, S., & De Choudhury, M. (2016). Quantifying and Predicting Mental Illness Severity in Online Pro-Eating Disorder Communities. In Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social Computing (pp. 1171–1184). New York, NY, USA: ACM. https://doi.org/10.1145/2818048.2819973
Fiesler, C., Young, A., Peyton, T., Bruckman, A. S., Gray, M., Hancock, J., & Lutters, W. (2015). Ethics for Studying Online Sociotechnical Systems in a Big Data World. In Proceedings of the 18th ACM Conference Companion on Computer Supported Cooperative Work & Social Computing (pp. 289–292). New York, NY, USA: ACM. https://doi.org/10.1145/2685553.2685558
Markham, A. (2012). Fabrication as Ethical Practice. Information, Communication & Society, 15(3), 334–353. https://doi.org/10.1080/1369118X.2011.641993
National Health and Medical Research Council. (2012, February 10). Chapter 2.3: Qualifying or waiving conditions for consent. Retrieved December 13, 2016, from https://www.nhmrc.gov.au/book/national-statement-ethical-conduct-human-research-2007-updated-december-2013/chapter-2-3-qualif
Nissenbaum, H. (2009). Privacy in Context: Technology, Policy, and the Integrity of Social Life. Stanford University Press.
Pater, J. A., Haimson, O. L., Andalibi, N., & Mynatt, E. D. (2016). “Hunger Hurts but Starving Works”: Characterizing the Presentation of Eating Disorders Online. In Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social Computing (pp. 1185–1200). New York, NY, USA: ACM. https://doi.org/10.1145/2818048.2820030 Twitter. (n.d.). Developer Agreement & Policy —
Twitter Developers. Retrieved December 13, 2016, from https://dev.twitter.com/overview/terms/agreement-and-policy
The contributors:
Oliver L. Haimson (University of California, Irvine) – Email | Bio
Nazanin Andalibi (Drexel University) – Bio
Jessica Pater (Georgia Institute of Technology) – Bio
This post may be cited as:
Haimson O, Andalibi N and Pater J. (2016, 20 December) Ethical use of visual social media content in research publications. Research Ethics Monthly. Retrieved from:
https://ahrecs.com/uncategorized/ethical-use-visual-social-media-content-research-publications
2 thoughts on “Ethical use of visual social media content in research publications”
Everything is very open with a clear explanation of the challenges. It was definitely informative. Your website is useful. Many thanks for sharing!|
Appreciating the time and energy you put into your
site and in depth information you provide. It’s awesome to
come across a blog every once in a while that isn’t the same old rehashed material.
Great read! I’ve bookmarked your site and I’m including your RSS feeds to my Google account.