“Real”/legal name communities behave better: where’s the evidence?

This is something I asked in comments here the other day:

So I think at this point we ideally would turn to research: how do people perceive the quality of communities on the real name required/encouraged/discouraged/forbidden axis and is this perception linked with or overwhelmed by their minority/exceptional/marginalised status in that community? Because it’s fairly clear by now that many people who personally have better perceived experiences with real names will universalise based on their own experience, and possibly people who prefer pseudonyms likewise (although pro-pseuds-allowed people don’t to me seem to turn it into pro-no-real-names-allowed as much).

Suw Charman makes a similar point:

I’d like to see the evidence that using real names changes people’s behaviour that much. Whenever I’ve been trolled/stalked online, it’s been by people using their real name. Dicks will, sadly, be dicks whether pseudonymously or eponymously. Whenever I bring this point up, people always point to 4Chan as an example of the sort of negative place that springs up when people are pseudonymous or anonymous. But 4Chan is a small corner of the web, and they are vastly outnumbered by all the pseudonymous people elsewhere that act perfectly nicely.

The ‘anonymity/pseudonymity = trollish behaviour’ meme has been doing the rounds for years, but it’s just not that simple. And it’s especially not that simple when the easy way to get round it is to use a pseudonym that looks ‘normal’ to Western eyes.

Are people aware of better-than-anecdotal evidence in either direction, that there’s a connection between permitted or encouraged pseudonymity and a decrease in civility or an increase in harassment? Or the other way around?

Note: when I say “better than anecdotal” I mean better than anecdotal. Comments along the lines of “well I don’t think I know of any evidence, but I have this anecdote/opinion that I really want to share” will be summarily deleted or possibly replaced with pictures of cranky cats.

Note 2: let me be even clearer (in response to stuff in moderation). What I want is data. Maybe not superb quality data, maybe self-selected or limited demographic or small sample or similar (obviously the better the quality of the data and the analysis the more convincing the research), but “my experience as a forum moderator” and “my friends’ experience online” is anecdote. I don’t have something against anecdotes in general—they are crucial in understanding lived experience—but in this thread I am asking for research.

(Tangent: I am hopeless at starting memes, but my G+ Share this if you’ve been harassed online by someone(s) using their legal name prominently in their email headers/profile/etc. is the second most successful ever, with 32 reshares.)

16 thoughts on ““Real”/legal name communities behave better: where’s the evidence?”

  3. Doctor Jay

    Here’s a treatment of the subject from the perspective of video games. The research explored is not specific to video games, though.

  4. cofax

    I think one might have to define “quality” first. You can have an active and thriving community that is, viewed from the outside, a festering hive of scum and villainy, but is creative and self-sustaining (i.e., 4chan). You can also have a really polite forum where the members drift away and no new folks arrive, and eventually it dies. One of them is more civil, but of what value is the discourse there? And all of that is unrelated to the question of naming protocols.

    Anyway, I don’t know if there have been any studies on group behavior, but Clay Shirky’s work, and maybe danah boyd’s, would be a place to start.

    1. Betsy

      I found this point really insightful and my brain wants to chew on it for a while.

  5. Alan Bell

    Cross site stats is hard, but I think I can do some analysis of twitter profiles. Starting from a sample of the full firehose stream (random sample of people tweeting now – biases towards frequent tweeters) I can grab tweets coming past and look at the profile of the tweeter and compare the number of people they follow to the number of people that follow them – my theory being that the people who add more value to the conversation should have a greater proportion of followers. I am now bumping up against square 1 of the bingo board, it is kind of hard to programatically work out what a real name is. Having looked by eye at rather a lot of twitter profiles it seems that a quick and dirty but reasonably effective method would be to assume that two or more words in the name field is a real name, one word is a pseudonym. It could also look at what they are saying, identify the language then run a spelling and grammar checker over the tweet contents to see if what is said is on average more or less coherent if they have a multi-word name. Does this sound like it would reveal any useful results, should I code it up and see what happens?

    1. Allison

      Code it up.

      I’d turn it backwards though.
      So predict which accounts have real names using you mechanism and which accounts are aliases. Then extract the top 100/bottom 100 and randomise the list order.

      Then without looking at the score vs, id just the account name manually assign them real name fake name. Run it back through a the list to compute the p-value of our hypothesis which if I’m not mistaken is that there is no difference between the realname/psuedoyms population in the two lists.

    2. Mary Post author

      Having looked by eye at rather a lot of twitter profiles it seems that a quick and dirty but reasonably effective method would be to assume that two or more words in the name field is a real name

      This introduces a massive cultural bias, if you do that at the very least I’d suggest limiting your search to accounts that post in English, and obviously still state it as an issue in your research. Separate issue: do I post under my legal name? On Twitter I use my ID given name, but not my family name.

      Moderator note: given the purpose of this thread (see my comment below), I’d rather you took the experimental design discussion off this thread to another site.

  6. Mary Post author

    Hi folks,

    I’m really sorry to have to delete comments from legitimate commenters, but I really want to collect links to research, if there is any, not ideas for what such research would look like, or general speculation on what you think it will find. The idea is to be able to point people to this thread so that they can examine the actual data/findings in order to further discussions. It’s clear people want a What Makes Communities Useful/Pleasant Theories thread, but this isn’t it.

  8. Dorothea

    Some desultory soc-sci database searching turned up a few things. I had trouble finding empirical research, so had to resort to more thought-pieces than I think you want. Nevertheless:

    Chen et al. Online privacy control via anonymity and pseudonym : Cross-cultural implications. Behaviour & information technology (Print), vol. 27, no. 3, pp. 229-242, 2008.

    Kavakli et al. Incorporating privacy requirements into the system design process: The PriS conceptual framework. Internet Research: Electronic Networking Applications and Policy, vol. 16, no. 2, pp. 140-158, 2006.

    Brennan and Pettit. Esteem, Identifiability and the Internet. Analyse & Kritik, vol. 26, no. 1, pp. 139-157, Dec 2004.

    Rowland, Diane. Privacy, freedom of expression and cyberslapps: fostering anonymity on the internet? International Review of Law, Computers and Technology; 17 (3) Nov 2003, pp.303-312.

    Phillips, David J. Negotiating the Digital Closet: Online Pseudonymity and the Politics of Sexual Identity. Information, Communication & Society, vol. 5, no. 3, pp. 406-424, 2002.

    Teich et al. Anonymous communication policies for the Internet : Results and recommendations of the AAAS conference. The Information society, vol. 15, no. 2, pp. 71-77, 1999.

  9. Sarah

    Tangential Dartmouth study on Wikipedia registered v. anonymous contributors:
    http://www.dartmouth.edu/~news/releases/2007/10/17.html
    “By subdividing their analysis by registered versus anonymous contributors, the researchers found that among those who contribute often, registered users are more reliable. And they discovered that among those who contribute only a little, the anonymous users are more reliable. The researchers were most surprised to find that the reliability of Good Samaritans’ contributions were at least as high as that of the more reputable registered users’ contributions.”

    And the only reason I’m not +1-ing your G+ meme “Share this if you’ve been harassed online by someone(s) using their legal name prominently in their email headers/profile/etc. ” is because I’ve been harassed *face to face* by someone using their legal name, so I’m not sure that qualifies as the kind of +1 you’re looking for.

  10. jon

    Removing anonymity won’t stop the online flame wars discusses research from psychology lecturer Dr Ros Dyer:

    In fact, contrary to expectations, her experiments demonstrated that students who were familiar with each other took more liberties on screen, not fewer. “There was four times as much flaming when they knew each other than when they didn’t,” Dyer says. There was also – dating sites take note – more flirting when people used their own names.

  11. Fox

    This paper (PDF) is from the study referenced in ABC’s War on Anonymity piece. It looks at the effects of South Korea’s Real Name Verification Law on online participation and behavior.

    Basically it finds that participants were less inhibited without the real name law. The paper measures disinhibition in terms of “swear words and non-normative expressions” which just sounds… kind of creepy? And like it’s probably not the best way, or a complete enough way, to measure disinhibition.

    But it’s the only study I’ve found so far that seems directly relevant, and “non-normative expressions” is something I could see in big letters on a t-shirt…

    What I would really like to find is something that looks holistically at multiple factors in a community’s culture that contribute to respectful interactions. My guess would be that in lots of contexts the risks of harms from pseudonyms outweigh the benefits of blocking them.

  12. Fox

    Ack! Meant to say something more like “the risks of harms from BLOCKING pseudonyms outweigh any benefits”

Comments are closed.