Do demographic differences lead to biased expectations of teachers?

Responsible researcher: Adriano Valladão Pires Ribeiro

Article title: WHO BELIEVES IN ME? THE EFFECT OF STUDENT–TEACHER DEMOGRAPHIC MATCH ON TEACHER EXPECTATIONS

Article authors: Seth Gershenson, Stephen B. Holt and Nicholas W. Papageorge

Location of intervention: United States

Sample size: 16,810 teacher-student pairs

Big theme: Education

Type of Intervention: Demographic differences

Variable of Main Interest : Teachers’ expectations

Evaluation method: Experimental Evaluation (RCT)

Policy Problem

Differences in educational levels in terms of income, race and gender are a problem, especially when it means that the minority part of these groups invests less in education than they should. One of the sources of asymmetry is teachers' expectations regarding their students' training, since teachers can directly influence students' decisions regarding whether or not to pursue further education.

Assessment Context

On many occasions, the teacher is one of the few adults with higher education that the student has contact with, playing a fundamental role in shaping their academic aspirations and affecting their performance. The perception that the teacher has low expectations for the student can negatively affect their performance and discourage them from studying. Furthermore, this low expectation can influence the teacher's own work, being reflected in worse classes, assessments and counseling, which also results in lower student achievement. Therefore, a low teacher expectation of a student's academic potential is a problem, especially for minority groups in a way that perpetuates educational difference.

In this context, the process of forming teachers' expectations gains importance, that is, which teacher's biases are responsible for linking a certain prediction to a student. A factor that generates bias that can be explored is the difference in demographic characteristics between teachers and students, such as race and gender.

Intervention Details

The 2002 Education Longitudinal Study (ELS) is a nationally representative survey of tenth-grade students (standard age 15 to 16) in the United States. In the survey, math and reading teachers provided their expectations for what each student's final level of education would be. There is also information on the demographic and socioeconomic characteristics of students and teachers. In total, the sample has 16,810 pairs between teachers and students, that is, two observations per student.

Demographic differences between teachers and students can be divided into four mutually exclusive categories: i) same race and same gender; (ii) different race, same gender; (iii) same race, different gender; or (iv) different race and gender. Same race and gender comprises 1/3 of the observations, same race and opposite sex another 1/3 and the last part is split half and half between the remaining two categories.

Initially, different predictions from two teachers about the highest level of education to be achieved by the same student could be random if they represented idiosyncratic opinions. Or even if that student had discrepant skills in math and reading subjects. However, neither of these two reasons can explain the relationship between the difference in the demographic characteristics of teachers and students and their expectations according to the following evidence.

Teachers have systematically lower expectations for the maximum education of black students compared to white students. On average, expectations for female students are higher than for male students. Furthermore, because most teachers are white, black students are more likely to encounter teachers with different racial profiles. Finally, because women are overrepresented in teaching roles, classes with teachers of the same gender are more likely for young women than for young men.

In a similar way, some standards can be described for teachers. White and female teachers tend to be more optimistic about their students' educational future when compared to black and male teachers, respectively. Furthermore, black teachers are more likely to teach students of the same race and students with lower performance. On the other hand, white teachers are 8 percentage points more likely to have graduated in higher education. As for gender, there is no big difference between teachers.

Methodology

The process of forming teacher expectations in the context presented is influenced by the characteristics of the teacher, the demographic differences between teacher and student and, also, the characteristics of the student. The problem in obtaining the effect of demographic differences on expectations is not accounting for all of these factors, especially when some characteristics of students, such as their motivation and innate ability, are not observed. By not considering unobserved factors, their effects would be partly credited to demographic characteristics. However, information about two different teachers' predictions for the same student allows us to correct this problem by comparing the variation in expectations between teachers with similar and different demographic characteristics for the same student. Therefore, a causal relationship is obtained between the demographic differences of teacher and student and the expectations of teachers.

Results

Teachers, on average, are more likely to assign lower educational expectations to students with other racial profiles. This result may be almost entirely reflected as the expectations of non-Black, primarily White, teachers of Black students. Specifically, non-black teachers attribute a 12 percentage point lower probability that black students will complete high school at most compared to black teachers. A deeper investigation shows that this prediction of completing at most high school is 5 percentage points less for young black men than for young black women. As for the gender difference, the result of expectations between students and teachers, on average, was the same. A detailed analysis, however, reveals that mathematics teachers give female students a greater chance of obtaining at most secondary education.

Public Policy Lessons

Since students' decision-making to seek a higher educational level can be influenced by teachers, the expectations formed about them become relevant. Especially when these expectations are related to demographic differences between students and faculty. Two lessons can be learned from the study. First, teacher hiring must be even more diverse and representative in order to reduce demographic differences. Second, the formation of faculty expectations of students should be included in training programs so that bias in expectations can be reduced.

Reference

Gershenson, Seth; Holt, Stephen B.; Papageorge, Nicholas W. “Who Believes in Me? The Effect of Student–Teacher Demographic Match on Teacher Expectations.” Economics of Education Review, vol. 52, p. 209-224, 2016.