Thursday 1 May 2014

Community feedback is likely to perpetuate undesired member behavior

The feedback of other community members is considered to be an important corrective factor in moderating community discussions. Liking a post, voting on a comment, rating are the most common feedback mechanisms. Theoretically, feedback would lead users to behave in ways that benefit the community.


But investigating four online sites,
  • CNN.com (general news),
  • Breitbart.com (political news),
  • IGN.com (computer gaming), and 
  • Allkpop.com (Korean entertainment
Cheng et al. [1] found some quite counter-intuitive results.

One commonality of these sites is that users post comments on (news) articles, where each comment can then be up- or down-voted by other users.

From a behaviourist point of view (here: operant conditioning), positive ratings should act as a “reward” and negative ratings as a“punishment”. So one would predict that feedback encourages users to generate better content in the future, and that users with negatively evaluated content will contribute less than rewarded users.

The impact on posting behavior
  • Negative feedback: Authors of negatively-evaluated content contribute more, their future posts are of also lower quality, and perceived by the community as even worse. Further, these authors are more likely to subsequently evaluate their fellow users negatively, percolating these effects through the community. 
  • Positive feedback neither encourages rewarded authors to write more, nor does it improve the quality of their posts. 
  • Users that receive no feedback are most likely to leave a community. 
Interestingly, evaluations polarize the community the most when positive and negative votes are equally split.

The impact on voting behavior

User behavior is largely tit-for-tat: Users with predominantly negative/positive evaluations will negatively/positively evaluate others. But very negatively evaluated people actually respond in a positive direction: The proportion of up-votes they give is higher than the proportion of up-votes they receive. And users receiving many up-votes appear to be more “critical”, as they evaluate others more negatively.

What are we to make of these findings?

The authers conclude that community feedback does not automatically drive behaviour in a direction that is beneficial to the community. Instead, it is likely to perpetuate detrimental behaviour. This, of course, raises the question whether the content evaluation mechanisms currently implemented in social media systems have effects contrary to the interest of the community.

Of course one could blame the theory: Despite of being a fundamental framework in behavioral psychology, there seems to be only limited empirical evidence that operant conditioning has noteworthy effects on human beeings. [2]

But did Cheng et al. investigate communities after all? Probably not. In a very general sense, the term community designates people gathering on a virtual site, even though the term audience would be  more appropriate. In a strict sense, a community comes into existence when users develop a significant sense of (virtual) community (SOVC). According McMillan and Chavis [2], sense of community consists has four components:
  • membership (feelings of emotional safety with a sense of belonging and identification),
  • influence (exertion of one's influence on the community with reciprocal influence of the community on oneself),
  • integration and fulfillment of needs (beeing supported and giving support, thereby reinforcing one to behave in a manner acceptable to the community),
  • shared emotional connection (positive affect related to community membership, shared history).
So I would expect, that members with a high SOVC but negatively rated contributions are likely to try to do better next time. Members with a low SOVC, on the other hand, may behave like described by the authors in their study or they may leave the community.
--------------------------------------------------------------------------------------------------------------------------

[1] Cheng, J., Danescu-Niculescu-Mizil, C., & Leskovec, J. (2014). How Community Feedback Shapes User Behavior. Retrieved May 1, 2014, from http://www-cs.stanford.edu/people/jure/pubs/disqus-icwsm14.pdf

[2] Baron, A., Perone, M., & Galizio, M. (1991). Analyzing the reinforcement process at the human level: Can application and behavioristic interpretation replace laboratory research?. The Behavior Analyst, 14(2), 95. Google Scholar.

[3] McMillan, David W. and Chavis David M. (1986). Sense of Community: A Definition and Theory.Journal of Community Psychology Volume 14, 6-23. Google Scholar.

For further information about the SO(V)C concept and possible ways to measure it, see the post: "Sense of community in virtual communities" (12.8.2013)

No comments:

Post a Comment