Gender Gap in Educational Outcomes

Definition

The gender gap in educational outcomes refers to the systematic difference between boys and girls in behavioral (suspensions, absences), cognitive (test scores), and attainment (graduation, readiness) measures. In developed-country contexts, the modern gap runs against boys: boys are more likely to be suspended, absent, and to fail to graduate high school on time. The key empirical finding from Autor et al. (2019) is that this gap is not fixed — it is substantially larger at the bottom of the socioeconomic distribution and shrinks as family socioeconomic status (SES) rises, with the mechanism operating postnatally through behavioral pathways rather than cognitive ones.

Key Ideas

The boy-girl gap in behavioral outcomes is largest among disadvantaged families and narrows as SES rises. Each standard deviation (SD) increase in a composite family SES index (mother's education, age, Medicaid birth, marital status) reduces the boy-girl suspension gap by $\approx 2.2$ pp and the absences gap by $\approx 0.075$ pp.
The gap in attainment follows the same SES gradient: $+1.15$ pp in kindergarten readiness per SD SES, $+1.5$ pp in on-time high school graduation.
Postnatal mechanism confirmed: The boy-girl birthweight gap is $\approx 3$ grams per SD SES — effectively zero. This rules out prenatal biology (differential fetal sensitivity to maternal stress, nutrition, cortisol) as the mechanism. The gap is socially produced, not hormonally determined.
Behavioral, not cognitive, pathway to dropout: Suspensions account for $\approx 90.6\%$ of the behavioral/academic mediator path from family disadvantage to high school (HS) non-completion. Test score gaps account for negligible share. Boys in disadvantaged families fall behind mainly because they are removed from the classroom or disengage from school, not because they develop lower cognitive skills.
Family operates independently of neighborhood/school: School and neighborhood quality controls reduce the family SES effect on behavioral outcomes by $\leq 25\%$ . This means that at least $73\%$ of the excess boy-girl gap explained by family disadvantage operates within rather than between schools and neighborhoods — family is a distinct input, not merely a proxy for where you grow up.
Race heterogeneity: The protective effect of SES on suspensions is weaker for Black children (three-way interaction $\beta_5 = +0.97$ pp). This is consistent with Black boys facing differential punitiveness in schools: even when family SES rises, suspension rates do not fall as much, implying a school-level constraint on the SES gradient.

How It Works

The Double-Difference Design

The identifying variation in Autor et al. (2019) is a double difference: the boy-girl gap in outcomes is regressed on family SES, so the coefficient captures how much the gap widens as SES falls. This differences out both (a) the gender-neutral effect of SES on outcomes and (b) any level difference between boys and girls not interacted with SES. The identifying assumption is that the latent boy-girl gap is uncorrelated with family disadvantage conditional on observable controls — i.e., that nature does not independently make boys more sensitive to observable SES proxies for reasons unrelated to the postnatal environment. The fetal origins test provides the key validation: if the mechanism were prenatal (e.g., testosterone-mediated fetal programming responding to maternal cortisol, which is elevated under poverty stress), then birthweight would show the same SES gradient. It does not.

Within-Sibling Robustness

The authors also use a sibling fixed-effects design: comparing siblings of different genders within the same family ( $N \approx 301{,}298$ ). This controls for all unobserved family-level characteristics. The sibling estimates are close to the ordinary least squares (OLS) estimates with SES controls, confirming that the double-difference result is not driven by unobserved family-level heterogeneity correlated with gender composition.

The SES Index

A principal component analysis (PCA) of four birth-record variables: mother's education, mother's age, Medicaid birth indicator, and marital status. The first principal component has eigenvalue $2.26$ (explaining $57\%$ of variance). The Black–white SES gap $= +1.44\sigma$ (mean Black $= -1.06$ , mean white $= +0.37$ ), implying that race heterogeneity in this gap partly explains differential educational outcomes by race.

Behavioral Pathway Mechanics

The mediation analysis (using school suspensions and absences as the behavioral channel and Florida Comprehensive Assessment Test (FCAT) math/reading scores as the cognitive channel) finds that behavioral outcomes account for the overwhelmingly dominant share of the mediator path. Suspensions are particularly important because they mechanically remove boys from the classroom during critical instruction periods, compounding initial disadvantage.

Why It Matters

For Education Policy

The finding that school controls explain only $\leq 25\%$ of the family SES effect has a direct policy implication: improving school quality or moving disadvantaged boys to better schools does not fully close the gap. The family environment itself is an independent input. This does not mean schools are irrelevant — it means that school-only interventions face structural limits and must be paired with family-support interventions.

For the Racial Achievement Gap

The Black–white SES gap ( $+1.44\sigma$ ) combined with the SES-sensitivity of the boy-girl gap implies that the Black-White gap in male educational outcomes is partly a family disadvantage gap, not purely a race effect. However, the smaller SES gradient for Black children in suspensions suggests that additional school-level factors (differential discipline) operate independently of family SES for Black boys.

For Intergenerational Mobility

Educational attainment — particularly high school graduation — is a primary input to adult earnings. A $1.5$ pp boy-girl graduation gap per SD SES, combined with the $+1.44\sigma$ Black-white SES gap, implies that Black boys from disadvantaged families face a compounded graduation disadvantage that feeds directly into the intergenerational income gap. See Intergenerational Income Mobility.

For Marriage Market Research

The family structure variable (marital status) is a component of the SES index and is also one of the strongest downstream determinants of child outcomes. Single-parent family structure is both an input to the SES index and a plausible mechanism: fathers' presence/absence may differentially affect boys' behavioral regulation. See Marriage Market.

Convergence with the Father Absence Literature

The behavioral pathway identified by Autor et al. (2019) aligns precisely with McLanahan, Tach, and Schneider's (2013) finding that father absence causally increases externalizing behavior — more pronounced for boys than girls — and reduces HS graduation through that behavioral channel rather than through cognitive deficits. The two literatures (family disadvantage × gender gap, and father absence causal effects) thus converge on the same mechanism: male children in disadvantaged or single-parent families are disproportionately harmed because they show elevated externalizing behavior, which removes them from the classroom and disengages them from school. This behavioral sensitivity appears to be social in origin (postnatal) rather than biological, consistent with both studies' falsification designs. See Causal Effects of Father Absence.

Open Questions

Does the behavioral disadvantage of boys in low-SES families reflect differential parenting practices (less paternal involvement in behavior regulation for boys), peer effects (boys in disadvantaged neighborhoods form peer groups with higher delinquency norms), or school sorting (boys tracked into less supportive classrooms)?
Why does the SES gradient flatten for Black boys in suspensions? Is this differential teacher/administrator punitiveness (implicit bias), peer composition effects, or something else?
Do the Florida findings generalize? Florida has a high share of Black and Hispanic children and a distinctive school governance structure; the SES gradients may differ in other states.
What happens in adulthood? The paper measures outcomes through HS graduation. Whether the behavioral gap persists into early-career earnings, incarceration, and mortality is an open research question (though the Chetty-Hendren Opportunity Atlas provides some evidence).