GSE Qualifying Papers

Validity studies of college admissions tests have found that black and Hispanic students typically perform worse in college, as measured by their freshman grade point average (FGPA), than predicted based on college admissions test scores. These studies, however, conflate student and school characteristics when estimating differential prediction. The differential prediction of racial minorities may arise in part because they attend high schools in which all students, regardless of race, perform worse in college than predicted. I examined this using data from the City University of New York. There was no overprediction based on race/ethnicity among students within schools when school characteristics were included in the prediction model. Instead, there was overprediction of FGPA based on school proportion black and Hispanic. Differential prediction associated with schools’ racial composition was greater in models that included high school test scores than in models with college admissions test scores.

Even though boys have consistently scored lower than girls on reading assessments, relatively little is known about the nature of this gap. The present study explored this gender gap in reading in the U.S. using the Early Childhood Longitudinal Study, Kindergarten Class of 2010-11 (ECLS-K:2011; n=7,780). Hierarchical linear modeling was performed in order to explore whether this gap is present at Kindergarten entry and how it develops throughout second grade; whether the gap holds when controlling for theoretically motivated factors that may differ by gender; and whether the magnitude of the gap differs by sociocultural factors. Girls had significantly higher reading scores than boys, on average, at Kindergarten entry, and their reading scores grew at a faster rate than boys through the end of second grade. The gender gap at Kindergarten entry was fully mediated by children’s age of first word, disability status, independent reading frequency, externalizing behaviors, and approaches to learning, and the gender difference in reading growth was partially mediated by children’s externalizing behaviors and approaches to learning. The magnitude of the gender gap was found to differ by school geographic region, but only in the model containing controls. No other sociocultural factors were found to moderate the gender gap. These findings suggest that the gender gap is already present at school entry and grows throughout second grade; it does not disproportionally affect certain groups of students; and children’s externalizing behaviors and approaches to learning may be of particular interest for future applied research.