In 1954, Brown v. Board of Education was decided in the US. The judges ruled that racial segregation in public schools was unconstitutional, leading to modern integration of public schools. One of the most well-known arguments leading to the integration of American public schools was that which came from the famous doll tests. These tests were thought up by psychologists, Kenneth and Mamie Clark and essentially let Black students and White students select their favourite from a set of dolls. If the students selected a doll of the opposite race as them, the psychologists concluded they had lower self-esteem.
So, the psychologists took off to Southern, segregated schools and performed the doll test on a group of children. These tests showed exactly what they expected. Black children were more likely to pick different-race dolls than White children, indicating that segregated schools led to feelings of lower self-worth in African Americans. These feelings of low self-worth were argued to explain the worse school performance of Black children at the time. While this argument was nowhere near fundamental in the Supreme Court decision (in fact, its only mention outside of the transcript is a footnote), most Americans recognise it as stunning, amplifying their negative feelings about segregation.
Good news! The study is wrong. The Clarks had various theoretical and practical problems in assessing the effect of segregation on self-esteem. First of all, there was no control group for this study. The Clarks simply took the results from segregated schools and said they had proven segregation lowered self-esteem. This is difficult to say unless they compared it to non-segregated schools. Lani Guinier pointed out Black children in non-segregated schools have far more contact with allegedly racist Whites. If this is the case, why should Blacks in schools without Whites have less self-esteem? Not only was their study lacking a control group, but the sample it had was incredibly small, making it more difficult to generalise.
The Clarks also commit the composition fallacy in the theoretical design of their study. Their argument goes as follows: connection to ethnic identity is correlated to self-esteem, connection to ethnic identity is hurt by segregated schools; therefore, self-esteem is hurt by segregated schools. But this is illogical at best. Even if X and Y are correlated, and Y and Z are correlated, this does not mean X and Z are correlated.
The doll tests are no better a measure than typical self-esteem measures. In fact, it may be worse. In the 1980s another set of doll test reports came out showing the self-esteem differences in Northern schools. These showed that Blacks preferred white dolls in the North even more than Blacks in the South. This would indicate either integration hurt Black self-esteem or that the doll tests are completely invalid. The famous Coleman Report suggested Blacks in the rural South were more likely to classify themselves as bright and able to learn than those in the North. Other major literature reviews using standard measures of self-esteem showed that Black children had greater self-esteem than Whites, and that this disparity was actually the smallest in integrated schools. Abbas Tashakkori wrote: ‘As expected, African American children had higher general self-esteem scores. Their scores were also higher on another index of self-attitude obtained from pooling the SDQ-M subscales. These findings are consistent with previous reports with larger, national samples of older adolescents.’
More recently, large longitudinal studies have been done which show the opposite of the doll tests. Researchers Ruth Erol and Ulrich Orth examined race differences in self-esteem over ages 14 to 16. The data included over 7,000 individuals. Across all ages, Blacks had the highest self-esteem. White self-esteem levels rose slightly, moderated, and slightly declined by age 30. In the end, a significant difference was found between Blacks and Whites, in the favour of Blacks. Across the entire sample, self-esteem would raise and stabilise with age. This creates further uncertainty in the doll tests: self-esteem is not a stable trait in childhood, so there will be unreliability in estimates of child self-esteem levels.
To this day, the doll tests are curiously massively cited in psychology. But, as we see, the original study did not even compare segregated schools to integrated schools. Without knowing this, there is no way to know that segregation had an impact on African American self-esteem. Furthermore, there is no theoretical reason to believe segregation hurts Black self-esteem more than integration does. Indeed, from the data we have, we can either say the doll tests are invalid or that integration had a negative effect on Black self-esteem. So, what do we make from this? Broadly, psychologists and teachers should stop putting as much emphasis on the doll tests. Despite their lack of validity, as well as unimportance in the Supreme Court decision, the famous doll tests have stolen their way into relevance, making it impossible for students to not be given a false narrative.
Zeke Jeffrey writes about human intelligence, drug policy, and psychology. You can connect with him on Twitter @zekes_tweets.