For the first time ever, LGBTQ+ populations are being invited to “come out” across a wide range of large, population-representative datasets. This new visibility not only provides sociologists with an historic opportunity to study queer people; it also invites us to study all people in more queer ways. In this talk, I present a new method to support this effort: what I call a “gender predictive” approach. Applying the tools of supervised machine learning, a “gender predictive” approach identifies new spectrums of gendered experience using old sources of survey data. To motivate this approach, I examine one pressing site of gender inequality in America today: the rapidly growing gender gap in higher education. Analyzing seven nationally representative high school cohort studies, spanning six decades, I present a new picture of how sex, gender and sexuality intersectionally shape school success.