A new working paper by American University public policy professor Seth Gershenson examines whether a “match” of students and teachers by race has any effect on teacher expectations of students. What is the result, for example, of white instructors teaching black students versus white students? What about other racial combinations?
Gershenson used nationally representative survey data from the Educational Longitudinal Study of 2002 (ELS) for U.S. students who were in tenth grade in 2002. There were over sixteen thousand student-teacher matches, which included various demographic data about the students and teachers. And each student’s tenth-grade math and English teachers reported their expectations for that student’s educational attainment, with possible responses ranging from those not finishing high school to those completing a four-year degree.
To ensure that any differences were systematic rather than random—which would suggest that teacher beliefs are at least partly explained by student demographics—Gershenon designed his study carefully. For example, he made use of various demographic variables to rule out systematic sorting (whereby, for instance, low-ability math students may be routinely assigned to white math teachers). The ELS administration was also set up so that a student’s two teachers offer their assessments at the same point in time. And the design built on a methodology that has been used in prior studies by other respected scholars.
The key finding is that non-black teachers have significantly lower educational expectations for black students than do black teachers when evaluating the same students. For example, when a black student is evaluated by a black teacher and a non-black teacher, the latter is about 30 percent less likely than the former to expect that student to complete a four-year college degree. These effects are larger for black male students and math teachers. When looking at the average effects across all students, there are small differences relative to racial mismatches; but these small average effects are mostly driven by the much larger effects among black students.
Gershenon writes in a related blog post that biases in expectations are generally unintentional and an “artifact of how humans categorize complex information,” pointing again to the need for a more diverse teaching force.
It’s a strong study with a startling finding, but it also underlines the need for a longitudinal study to measure, for example, which teacher’s prediction holds true for a given student. Are the lower expectations unfairly pessimistic, or are the higher expectations unrealistically hopeful?
SOURCE: Seth Gershenson, Stephen B. Holt, and Nicholas Papageorge, “Who Believes in Me? The Effect of Student-Teacher Demographic Match on Teacher Expectations,” W.E. Upjohn Institute for Employment Research, Working Paper 231 (July 2015).