The Gender Gap in College Scorecards and the Problem with 10-Year Earnings Data

November 23, 2015

In September, the Department of Education unveiled its new College Scorecard website. The scorecards are a major step forward in unleashing the power of educational data, as they allow prospective students to compare outcomes at different schools, providing data on costs, graduation rates, graduate indebtedness, and student earnings, among other metrics. But the scorecards have shortcomings, too, ranging from the limits of the data sample—the tax returns of students who received federal aid—to their reduction of a college education to a purely economic exchange. Among its most significant oversights, the scorecard data fails women


In September, the Department of Education unveiled its new College Scorecard website. The scorecards are a major step forward in unleashing the power of educational data, as they allow prospective students to compare outcomes at different schools, providing data on costs, graduation rates, graduate indebtedness, and student earnings, among other metrics. But the scorecards have shortcomings, too, ranging from the limits of the data sample—the tax returns of students who received federal aid—to their reduction of a college education to a purely economic exchange. Among its most significant oversights, the scorecard data fails women.

Gender isn’t entirely missing from the scorecards, of course; The New York Times flagged that the ten-year earnings data, highlighted by the scorecards, shows large wage gaps between women and men, especially those who graduate from top schools. This is hardly shocking: We know that there is a persistent gender gap in wages throughout the country. In 2014, women’s median earnings were 78.6 percent of men’s. In fact, it would be surprising if a gender gap didn’t exist after graduation. What is surprising, though, is how little the federal data tells us about the gender gap—and how much it could, if it took seriously women’s professional life cycle and the needs of female students.

Ten-year earnings data, while quite useful for understanding male graduates’ economic prospects, simply does not tell us that much about women’s earnings. We know that women’s earnings take an immediate hit when they have children—that inadequate maternity leave policies and child care options force many women to take time out of the workforce or at least reduce the hours they work. And 10 years out of college is exactly when most college-educated women are having children.

To understand women graduates’ earnings, especially in relation to men, we need to see the earnings data not only at 10 years, but at five, 15, and 20 years as well. Otherwise, we have no idea what problem we need to address. One university might provide recent female graduates with equal employment opportunities yet also have alumni who overwhelmingly drop out of the workforce to raise children. Another might graduate committed career women for whom equal paying jobs are nonetheless unavailable. We might even find that certain schools manage the difficult—and essential—task of teaching women how to return to a successful career if they choose to take time out to care for their children.

While only the 10-year data is included in the College Scorecard itself, the Department of Education does provide mean student earnings at six years and 10 years in the underlying data. Even that single additional point in time provides vastly more information about the gender gap, and about how it manifests differently for different schools. For example, male MIT graduates make about 1.4 times as much as female MIT graduates six years out of college, but 1.6 times as much 10 years out. Their gender gap grows by 20 percentage points in just four years. In contrast, the wage gap between Rice alums shrinks rapidly in the same period. Although earnings among Rice graduates are less equal after six years than among MIT students, with men making almost 1.5 times as much as women, that number drops to just 1.2 by 10 years out. Male and female graduates at the University of Rochester begin with a very small gender gap that barely grows at all. Women who graduate from Yale make more than Cornell women six years after graduation, but at 10 years the opposite is true thanks to an ever-widening gender gap among Yale grads.

What are these schools doing differently? Are Rice women more likely to stay in the workforce when they have children than women at Yale or MIT? Are women at Cornell learning to ask for raises? This kind of data analysis can help colleges learn how to prepare their female graduates to overcome the gender gap.

This data can also help us understand not only school-level policy, but broader social questions as well. When does the gender gap emerge? How much of a hit do women take when they reach childbearing age? How does this vary across educational, socioeconomic, or racial backgrounds?

The College Scorecard data is meant to prompt these kinds of conversations for individuals and institutions alike. But for women—who, after all, are 57 percent of college students—those conversations have been cut short by the DoE’s choice of what data to collect and what data to highlight.

Government research has a long history of forgetting about women. As recently as the 1980s, the NIH routinely conducted all-male studies of common drugs. Choices in what we measure reflect, and then impose, gender bias on public policy. Unless data collection accounts for the many ways sex and gender shape our lives, policies can only continue to fail women.