Working Paper: NBER ID: w18464
Authors: Prashant Bharadwaj; Giacomo De Giorgi; David Hansen; Christopher Neilson
Abstract: We establish the presence of a gender gap in mathematics across many low- and middle-income countries using detailed, comparable test score data. Examining micro level data on school performance linked to household demographics we note that first, the gender gap appears to increase with age. Indeed, the gap nearly doubles when comparing 4th grade and 8th grade test scores. Second, we test whether commonly proposed explanations such as parental background and investments, unobserved ability, and classroom environment (including teacher gender) explain a substantial portion of the gap. While none of these explanations help in substantially explaining the gender gap we observe, we show that boys and girls differ significantly in perceptions about their own ability in math, conditional on math test scores. Girls are much more likely to state that they dislike math, or find math difficult compared to boys. We highlight differences in self-assessed ability as areas for future research that might lead to a better understanding of the gender gap in math.
Keywords: gender gap; mathematics; education; low and middle-income countries
JEL Codes: I23; O15
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| self-assessed abilities (G53) | gender gap in math performance (I24) |
| parental investments in math homework (D29) | gender gap in math performance (I24) |
| societal stereotypes (Z13) | gender gap in math performance (I24) |
| gender (J16) | math performance (C12) |
| age (J14) | gender gap in math scores (I24) |