Monday, March 24, 2014

The Science of Anonymity: Why We Should Flood, Not Boycott, Yik Yak

It cannot be said that anonymity tends to bring out the best in people.  The KKK’s slitted hoods, the masks warriors wear into battle, and the lonely basements occupied by Internet trolls encourage depths of nastiness and aggression that their perpetrators would not dare to display if they were individually identifiable.  And now we have Yik Yak.   

Like most of the Lehigh community, I had not heard of Yik Yak until I received an email about it from the Council for Equity and Community (CEC) on Sunday evening.  Yik Yak, it turns out after a quick download, is an anonymous and location-based version of Twitter.  People can post short thoughts (Yiks?) without an account or username, which are then broadcast to all other Yak-ers within a certain radius.  And like most anonymous things on the Internet, many of the thoughts so broadcast are solipsistic, illogical and intentionally offensive.  Some are racist, and a great many more are misogynistic and homophobic.  Not to mention scatological.

The postings I have seen on Yik Yak (emanating from Lehigh) are not the products of great minds.  Many are, further, hurtful and divisive, and they should not go unchallenged.  To best formulate a challenge, it might be helpful to turn to the science of crowd behavior.

The effects of anonymity are actually rather more subtle than a simple “anonymity causes antisocial behavior” storyline implies.  Certainly that can be the outcome.  A classic study in social psychology, for example, observed that children who were anonymous (vs. identified by name) stole more candy while trick-or-treating, particularly if they were in a group. 

It turns out that the nature of the group is the key.  Anonymity does not seem to increase antisocial behavior per se, but rather increases conformity to what people believe is normative behavior in that situation.  Anonymity reduces feelings of individual identity, but replaces them with a heightened sense of collective identity.  As the individual self becomes less influential, people start looking to the behavior of others to determine how they should behave.  As a result, anonymity can, on occasion, actually increase prosocial behavior – if that’s what the group is doing and is what the situation seems to demand.  In many cases, of course, group norms are more antisocial and anonymity increases asshole-ishness (to use the scientific term).

Anonymity is not, in other words, inevitably disastrous.  And the behavior of some of our students – which is currently pretty vile - cannot really be blamed on Yik Yak.  I suspect that in response to the CEC’s email, many people have, like me, taken a look.  A few find it attractive and have joined in. Most people, however, quickly and reasonably decide that it is not a place they want to be.  

There have been calls for a boycott of the app.  This is of dubious feasibility and moreover is a lost opportunity.  If the effects of anonymity on behavior are driven by conformity to group norms, then the response must be to change those norms.  The answer is not to boycott or avoid Yik Yak, but to flood it.  To flood it with users who call out the sexists, the racists and the gay-bashers, who mock the stupid potty jokes, and who change the conversation.  Two hundred new Lehigh Yak-ers – students and faculty – would drive out the diehard jerks.  More importantly, it would change the behavior of the rest who have responded to anonymity by going along with the crowd.  It will take some planning and organization, but let’s give them a new crowd to follow.

For further reference see:

Diener et al. (1976). Effects of deindividuation variables on stealing among Halloween trick-or-treaters. Journal of Personality and Social Psychology. 

Reicher, Spears & Postmes. (1995). A social identity model of deindividuation phenomena. European Review of Social Psychology.

Sunday, February 16, 2014

Strategic Biases & Evaluative Tuning - The SPSP talk that wasn't...

Much to my chagrin, a massive snow storm prevented me and most of the lab from traveling to Austen for #SPSP2014.  In an effort to salvage something from the wreckage, I recorded a shortened - 10 minute - version of the talk I would have given at the Attitudes Preconference.

Strategic Biases and Evaluative Tuning
(or Can Batman Reduce Intergroup Biases, including Rapid Evaluative Responses?)

This accompanying poster with ShiangYi Lin provides more detail about our evaluative tuning data, which investigate how cooperation-facilitating social institutions can reduce implicit racial biases. 

Wednesday, November 27, 2013

The Cooperative Contingencies Model - New & Improved Working Paper

Matthew Kugler and I have posted a revised version of our working paper on the Cooperative Contingencies Model of strategic intergroup bias - the CCM, for short.  The manuscript is currently under peer review, but we would love to hear any comments, suggestions, critiques, supporting or contradictory evidence that you might have.

"What's the Cooperative Contingencies Model?" you ask.

Despite substantial declines in overtly prejudicial attitudes, massive disparities (e.g., in education, health, employment and justice) persist between groups. The CCM offers a partial account for how and why intergroup disparities and segregation endure despite positive shifts in intergroup attitudes. Life in human societies hinges on cooperation, but decisions to cooperate with others are often risky. The CCM posits that because groups facilitate cooperation, decision-makers often have an instrumental incentive to preferentially coordinate with ingroup (over outgroup) members. Importantly, this incentive operates independently from other contributors to bias (e.g., social identity, ideology) and thus exists even for egalitarians. The CCM further posits, however, that the incentive to preferentially coordinate with ingroup members is contingent on the full set of cooperative requirements (how much people need to cooperate) and cooperative affordances (mechanisms available to facilitate cooperation) perceived in current contexts.


Specifically, we hypothesize that decision-makers are more likely to preferentially rely on ingroup members in situations where cooperation is important for goal attainment.  Importantly, however, strategic reliance on ingroups can be attenuated by the presence of alternate cooperative affordances, including social structures or institutions that facilitate cooperation between individuals (e.g., rule of law, third party punishers).  Some early evidence for this prediction can be found here.  Interpersonal relationships and reputations also represent cooperative affordances that can reduce intergroup bias when social networks include outgroup members. However, given that people are more likely to know and know about ingroup than outgroup members, these personalized affordances often tend to increase intergroup bias. 
The CCM can be used to derive specific and real world hypotheses about when people will exhibit strategic intergroup biases and how they can be reduced.  We predict, for example, that informational asymmetries and power differentials will tend to increase strategic biases.  The model sheds light on the stubborn persistence of intergroup disparities in societies with increasingly egalitarian values, and promises to illuminate a variety of interesting issues, including:
  • Why gangs are often racially segregated and wear highly visible gang colors or tattoos
  • Why non-prejudiced people nevertheless laugh at or even tell racist jokes
  • Why con artists and fraudsters are likely to selectively target ingroup members
  • Why bias is likely to be greater in markets for used vs. new goods
  • Why employees may often be more biased in their choice of an employer than employers are in their choice of employees

It's a model, and thus imperfect.  Bits of it will be right, and other bits of it will undoubtedly be wrong.  However, as George Box wrote, "All models are wrong, but some are useful."  We have found this to be a useful model because it is directing our attention to novel questions and raising provocative hypotheses. We hope that others will find it similarly useful.