[Air-L] Including screennames with tweets

QCENTRAL qcentral at indiana.edu
Sat Jul 14 05:44:37 PDT 2018 

Hey all,
Really enjoying this thread.

There’s a great presentation by Melissa Cheyney and Lisa Leventhal, offered by Public Responsibility in Medicine and Research, called “Reviewing Research on Data from Social Media Posts: When is it Within our Scope?”

https://youtu.be/hpqgTp9qlJY

Their presentation does a very nice job of outlining ways to think through when, how, if you deanonymize people’s social media posts.

TL;DR (from a person who spends a LOT of time thinking about this):
Funny to see this conversation seem so fresh…we’ve been thinking about this for a long time and there’s a ton of established norms at our fingertips, including the ethics docs written by AoIR. But, definitely, this stuff can be hard. There isn’t one right answer. Every research question will come with its set of ethical concerns.

For me, the quickest route to figuring out what to do with any online data is to ask myself “Who benefits from what I do?” For researchers doing scholarship in the public’s interest (e.g., if you’re a student or faculty member at an institution of higher ed or organization meant to serve the public interest or want your research to contribute to our understanding of the world), then, first and foremost, the people we involve in our research—willingly, we hope, but sometimes caught in our dragnets—get first dibs on benefits.

At the very least, our obligation is to respect people’s autonomy—their right to choose—and do our best not to introduce more risks to their lives because they participated in our research. Once their needs are met, all other stakeholders—from the individual researcher and her institution/company to Science (!) get to benefit. So, if I list a person’s Twitter handle because Twitter’s TOS says that I should, who benefits? Does the participant, first and foremost benefit or, at least, not suffer from my decision, especially if I had no way to give them a say in the matter? That’s one way to approach these questions. For me, it’s the best way to know that I’m always advancing the interests of the human participant over my own interests (and that’s my obligation as a researcher…if I want to advance my own interests first I should probably not be a researcher for the public interest).

For those who might say “but I can’t find out if someone wants to be a part of my research from scraping their tweets?” I’d say: yep. That’s an ethical dilemma that sprouts from your methodology. So, what are you going to do to at least try to respect this human being providing you awesome tweets for analysis?” One easy thing we can do: don’t name them in ways that set them up for Doxxing, even if you don’t think they should care about that or that it’s likely to happen (you don’t know if they have ex-lovers stalking them who’d love to have their name in your pub to add further flames to the fire). At the very least ask: “who benefits from me listing their Twitter handle?” If they aren’t at the top of the list, then there’s work to do to make sure they aren’t taking the fall when other people advance from the use of their names.

Stopping there.
: )

Mary


-- --
Mary L. Gray
Senior Researcher and Ethics Advisory Board Chair
Microsoft Research

Associate Professor
School of Informatics, Computing, and Engineering
Affiliate Faculty of Anthropology, Gender Studies, and The Media School
Indiana University

Fellow
Berkman Klein Center for Internet and Society
Harvard University


  1. Re: Including screennames with tweets (Craig Hamilton)


----------------------------------------------------------------------

Message: 1
Date: Fri, 13 Jul 2018 19:32:28 +0000
From: Craig Hamilton
This is a great discussion, and I?m glad to be part of it. It occurs to me that the fact people on this list in particular don?t have a hard and fast answers demonstrates what a tricky issue this is. I don?t have the answer either, but for what its worth I can share my own approach and thoughts.

I?ve collected from Twitter and other social media channels for a project dating back to 2013. The tweets in my particularly instance are volunteered to a specific, unique hashtag (#harkive), rather than harvested from keywords. Even so, I?ve taken the decision not to use usernames in articles/chapters/presentations. I provide a way via my project website for people to have their data permanently removed but, as I guess we all know, anyone who enters the text of a tweet into Google will almost certainly be able to find the original author if they want to. Once a tweet and/or a username is printed in a physical book, or a PDF made available online, then I would see that as beyond my control - in other words, removing a tweet from my database would not remove it from the copies in circulation. My feeling is that excluding usernames mitigates potential risks somewhat, but it doesn?t remove those risks entirely. If I were working with data harvested from keywords - and if consen!
t had not
 been given - then I would have misgivings about both verbatim quotes and usernames, and certainly if the subject matter of the tweets and/or article had any obvious potential to cause harm.

Kind regards
Craig
On 13 Jul 2018, at 19:13, Tarleton L. Gillespie <tlg28 at cornell.edu<mailto:tlg28 at cornell.edu>> wrote:

I think someone who administers IRB would disagree. Even if we take tweets to be public statements, and even if we the users in this case fully understood them as public, there are different values to think about. As researchers, our obligation is to recognize that including people in our work, however valuable to the research, can come at a cost for the people made into subjects of that research; part of the commitment to being a researcher then means respecting people's autonomy and treating them ethically, "not only by respecting their decisions and protecting them from harm, but also by making efforts to secure their well-being" (http://sphweb.bumc.bu.edu/otlt/MPH-Modules/EP/EP713_ResearchEthics/EP713_ResearchEthics3.html) Instead of "If you choose to exclude, you will have to explain why" I think IRB principles would say it?s the reverse: if you're including the identity of your research subjects, you will have to explain why -- and probably to them.

Paula, by the end your comment seems to imply that what's at issue is whether to quote the tweet itself ("very risky"), but I think all that is being discussed here is identifying its author. I totally have the gut instinct to want to credit people, it's deeply ingrained in how we work. But the tweets in this case are not secondary sources of scholarship, they're being selected and used as examples. Other tweets could similarly have been chosen to make the same point. Their value is not that they were posted by person X, but that they sound a certain way or address a certain topic. I would err on the side of protecting the person, because I can't anticipate the harm and can't anticipate their framework for understanding and can't anticipate how they'd feel about their words being used as an object of my research -- unless, of course, I sought their actual consent, as Casey suggested.

This conversation is really helping me think about this, thanks all.

Tarleton




?On 7/13/18, 1:46 PM, "Air-L on behalf of paula todd" <air-l-bounces at listserv.aoir.org<mailto:air-l-bounces at listserv.aoir.org> on behalf of paulatoddmedia at gmail.com<mailto:paulatoddmedia at gmail.com>> wrote:

  Hi, a few thoughts ...  Screen names are often creations/public projections
  and forms of agency, important/expressive in their own way. Signature
  selections are integral to the full social media message. Exclusion, then,
  is a form of censorship. If you choose to exclude, you will have to explain
  why. If the only reason is because someone who uses social media may not
  understand that their content is public and can therefore (legally) be
  reproduced or re-mediated on any platform, the same logic could apply to
  reproducing public content of any type. Very risky. The only cases I could
  justify such editorializing (selecting which parts of a communication to
  share) parallel general defamation and hate speech limitations. Social
  media is public, and of the public sphere; those who want to create
  private/privileged communication use peer-to-peer, offline, direct
  messaging, telephone *et al.*

  Paula


  *Paula Todd*
  B.A., LLB.(J.D.), PhD Can. (Digital Journalism)
  York & Ryerson Universities
  Toronto, Canada


  On Fri, Jul 13, 2018 at 2:23 AM, Hayes, Rebecca M <hayes2r at cmich.edu<mailto:hayes2r at cmich.edu>> wrote:

Dear All,
Can you please weigh in on the decision to include or not include
screennames
when we cite tweets in a book? The book is on new media and crime,
and we are using tweets in a few places as examples of some different
discussions.

We are back and forth on whether we should include the screennames and at
others or disclude them. The arguments we have seen thus far, are to
include them because it was made public and we are citing someones words.
The other argument is to disclude them
as the person did not consent to have it printed in that way persay, and
the screenname attached in our book could be used to find and harass the
person. What are your thoughts?

Thank you,
Becky
_______________________________________________
The Air-L at listserv.aoir.org<mailto:Air-L at listserv.aoir.org> mailing list
is provided by the Association of Internet Researchers http://aoir.org
Subscribe, change options or unsubscribe at: http://listserv.aoir.org/
listinfo.cgi/air-l-aoir.org<http://air-l-aoir.org>

Join the Association of Internet Researchers:
http://www.aoir.org/




  -
  _______________________________________________
  The Air-L at listserv.aoir.org<mailto:Air-L at listserv.aoir.org> mailing list
  is provided by the Association of Internet Researchers http://aoir.org
  Subscribe, change options or unsubscribe at: http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org

  Join the Association of Internet Researchers:
  http://www.aoir.org/


_______________________________________________
The Air-L at listserv.aoir.org<mailto:Air-L at listserv.aoir.org> mailing list
is provided by the Association of Internet Researchers http://aoir.org
Subscribe, change options or unsubscribe at: http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org

Join the Association of Internet Researchers:
http://www.aoir.org/


------------------------------

Subject: Digest Footer

_______________________________________________
The Air-L at listserv.aoir.org<mailto:Air-L at listserv.aoir.org> mailing list
is provided by the Association of Internet Researchers http://aoir.org
Subscribe, change options or unsubscribe at: http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org

Join the Association of Internet Researchers:
http://www.aoir.org/

------------------------------

End of Air-L Digest, Vol 168, Issue 19
**************************************

More information about the Air-L mailing list