Thursday, January 16, 2014

Stereotype threat vs self-fulfilling prophecy

Here's an experimental idea: How does stereotype threat affect people's behavior in the ultimatum game?  If you tell a group of people that the purpose of your study is to measure gender differences in cooperation, will this cause women to behave more cooperatively in a self-fulfilling prophecy?

Let me step back a bit and define our terms.  Stereotype threat is when you remind people that they're part of a negatively stereotyped group, and this reminder causes them to perform more poorly on a performance test, confirming the negative stereotype.  In the classic experiment, subjects are given an academic test.  When told that the test was diagnostic of intellectual ability, this negatively impacted the scores of African American subjects compared to other subjects, because they were reminded of negative stereotypes of their group.  (There is also an opposite effect, called stereotype lift, where people who are not part of the stereotyped group perform better when primed with the stereotype.)

The ultimatum game is a game often used in studies of social behavior.  Two players are offered a chance to split $100.  The first player decides how to divide the money.  The second player considers the first player's offer, and either accepts it or rejects it.  If the offer is rejected, then no one gets any money.  A more cooperative player offers more money to the other player.

The stereotype threat is often explained as a "self-fulfilling prophecy".  Mention a stereotype, and it causes the stereotype to become true!

However, every study on the stereotype threat seems to only look at performance tests.  The proposed mechanism for stereotype threat is that reminding people about their membership of a stereotyped group causes anxiety and taxes cognitive resources.  (Stereotype lift, on the other hand, reduces anxiety.)  This may cause people to confirm stereotypes when the stereotype is that they'll perform more poorly on a test.  But what happens when you study a stereotype that does not have to do with test performance?  For instance, could stereotype threat also "confirm" stereotypes about being more or less cooperative?  If so, could it also "confirm" other stereotypes, like gay men being effeminate or black people being religious?

That's why I think it would be interesting to see if stereotype threat affects how people behave in the ultimatum game.

Unfortunately, I don't have the resources to perform this experiment, but we can find answers in existing literature.  A brief literature search found the following paper: Social Identity and Preferences by Benjamin et al.

The paper finds that when Asian-Americans and (non-immigrant) black Americans are made to think about their own ethnicity, this affects their patience (with respect to receiving money rewards), and their risk aversion.  The authors infer that there are ethnic norms about patience and risk-aversion which are enhanced by priming.  The specific results of the paper aren't relevant here, but what's interesting is that it appears to be a self-fulfilling stereotype which is not about a performance test.

What's more, the authors do not use stereotype threat as an explanation.  Rather, they consider stereotype threat as one possible explanation, and reject it.  It cannot be stereotype threat, because they find that their method of priming people does not increase their anxiety.  Instead, the authors understand their results within self-categorization theory.  Priming people causes them to see themselves as more part of a group, and modify their behavior according to what is expected within that group.

This clarifies something I may have misunderstood about stereotype threat.  Stereotype threat is not just any mechanism by which stereotypes become self-fulfilling prophecies, it's just one particular mechanism.  Stereotype threat operates by causing anxiety and taxing cognitive resources, incidentally creating self-fulfilling prophecies when stereotypes are related to performance on certain tests.

The important thing about stereotype threat is not that it causes people to confirm stereotypes, the important thing is that negative stereotypes cause people anxiety, and this anxiety has measurable impact on their performance of real world tasks.

As for my experimental proposal, I predict that priming people with gender may in fact cause women to behave more cooperatively, but not by the mechanism of stereotype threat.  Rather, by self-categorization theory, men and women may see themselves as more part of their gender, and modify their behavior to better accord with gender norms.   But that's just my prediction, who knows whether it would carry out.