Stereotype Inaccuracy: A Belief Impervious to Data
When are liberals anti-scientific?
Posted Aug 01, 2014
Before continuing, I request that you start with this disclaimer, if you have not already read it:
But we agree that being impervious to data is a bad thing, right? Liberals routinely rail against conservatives' supposedly anti-scientific stands, right? Liberals, in sharp contrast, don't ever oppose data and science, do they?
Great! In that case, you will be interested to discover that:
1. Stereotype accuracy is one of the largest and most replicable effects in all of social psychology
2. The fact that this is true has had almost no effect on the frequency with which social scientists claim, assume, or imply that stereotypes are inaccurate.
You probably find this hard to believe. After all, you have been told, over and over and over and over, that stereotypes are inaccurate. This has been part and parcel of the liberal project of fighting oppression and prejudice.
So, I can’t blame you for finding this hard to swallow. But it is true nonetheless. Let’s first start with some definitions:
Social perceptual accuracy: Correspondence between a person’s belief and social reality, with one exception. If that person has caused a self-fulfilling prophecy, then the belief was not initially accurate. (Self-fulfilling prophecies are initially erroneous beliefs that create their own reality).
Social reality: What people are actually like. Their beliefs, attitudes, attributes, accomplishments, behaviors, etc.
Empirical question: A proposition whose truth can only be evaluated against data. “The Yankees will win the World Series,” “the Earth is getting warmer,” and “women earn 75% of what men earn” are all empirical questions.
Perceiver: Person making a judgment about a group. A perceiver group is a group of perceivers making a judgment about a group.
Target: A person being judged. A target group is a group being stereotyped.
For example, I might ask my introductory psychology students what they believe about the social status and achievements of Jews, African-Americans, and Asian Americans. My students would be the perceivers, and Jews, African-Americans, and Asian Americans would be the target groups.
Stereotype accuracy is the extent to which people’s beliefs about groups correspond to those groups’ actual characteristics. In this blog, I focus on two types of stereotype accuracy:
- Consensual stereotype accuracy correlations.These are the correlations between a group of perceivers’ average beliefs about a target group, and the target group’s actual characteristics.
- Personal stereotype accuracy correlations. These are the correlations between an individual perceiver’s beliefs about a target group, and the target group’s actual characteristics.
(see Jussim, 2012; Jussim et al., 2009, in press, for more detailed explanations of these).
Correlation: A statistic indexing how strongly one variable relates to another. Correlations range from -1.0 to 1.0, but for this blog entry, I will mostly be concerned with correlations between 0 (two variables are unrelated) and 1.0 (two variables are perfectly related). Correlations are often used as one way to measure an “effect size” – how big an effect one variable has on another.
How big does a correlation have to be for social psychologists to consider it BIG? Correlations of .40 and higher, are generally considered large. Social psychology is a field of modest effects. The overall average in the field, corresponds to a correlation of about .20 (Richard et al, 2003). By most standards, that is a pretty small effect. If teachers have a .20 effect on their students, it means they are having a substantially effect on about 10% of their students – which is the same as saying they are having little or no effect on 90% of their students.
Stereotype accuracy is an empirical question. You can claim anything you want. Your interpretation of your experience is whatever you believe. But combating the well-established flaws and limitations of subjective interpretation of experience is exactly why science was developed.
Which gets us to, not your personal experience, but the science. What has scientific research found about the accuracy of stereotypes?
Stereotypes are (Usually) More Valid Than Most Social Psychological Hypotheses
Over the last 40 years, there has been a ton of research assessing the accuracy of stereotypes. The findings are astonishing, at least if you have bought the longstanding line that “stereotypes are inaccurate.”
The following data are from my recent review of this area of research (Jussim et al, 2014). It gives the proportion of results for various types of research that are greater than correlations of .30 and .50, respectively, because Richard et al (2003) provided these figures for all of social psychology, which then constitutes an excellent standard of comparison.
Which is more accurate, social psychology or social stereotypes?
Proportions of correlations that are: >.30, >.50
All of social psychology: 24%, 5%.
Race, consensual stereotype accuracy: 95%, 95%.
Race, personal stereotype accuracy: 47%, 18%
Gender, consensual stereotype accuracy: 100%, 94%
Gender, personal stereotype accuracy: 79%, 58%
These results are based on over 20 studies of stereotype accuracy conducted by multiple independent researchers and laboratories (see Jussim, 2012; Jussim et al, in press, for reviews). Results for other stereotypes (e.g., age, occupation, politics, etc.), are similar. As such, stereotype accuracy is far more replicable than many far more famous “effects” in social psychology (large effects are inherently more replicable, but understanding why that must be involves an arcane statistical discussion that is beyond the scope of this blog entry).
To be sure, there is some evidence of inaccuracy in stereotypes, especially national stereotypes of personality. There is also good evidence that political ideologues exaggerate each others' views. Nonetheless, the BIG picture remains intact: Stereotype accuracy is one of the largest and most replicable findings in all of social psychology.
Why, then, have social scientists been declaring and decrying the inaccuracy of stereotypes for nearly a century? The data don’t now, and never have, supported such a claim.
Social scientists don’t go around making stuff up to advance their leftish narratives of oppression. Do they?
Disclaimer II: I AM a social scientist. There are LOTS of other social scientists out there who go to great lengths and do a good job of not allowing their politics to distort their science. I admit that making claims that are unhinged from data does no credit to our field and, if taken out of context, can lead people to dismiss the field’s value and importance. However, the solution to bad science is not to kill science. It is to pressure and advocate for, and push, enhance, and support good science. Such efforts, which include exposing bad science, should count as a CREDIT to the social sciences.
Jussim, L., Crawford, J. T., Anglin, S. M., Chambers, J. R., Stevens, S. T., Cohen, F. (2014). Stereotype accuracy: One of the largest and most replicable effects in all of social psychology. Draft chapter prepared for T. Nelson (ed.), Handbook of Prejudice, Stereotyping, and Discrimination, Hillsdale, NJ: Erlbaum.
Jussim, L. (2012). Social perception and social reality: Why accuracy dominates bias and self-fulfilling prophecy. New York: Oxford University Press.
Jussim, L., Cain, T., Crawford, J., Harber, K., & Cohen, F. (2009). The unbearable accuracy of stereotypes. In T. Nelson (Ed.), Handbook of prejudice, stereotyping, and discrimination (pp.199-227).Hillsdale, NJ: Erlbaum.
Richard, F. D., Bond, C. F. Jr., Stokes-Zoota, J. J. (2003). One hundred years of social psychology quantitatively described. Review of General Psychology, 7, 331-363.