Association for Public Policy Analysis & Management

During the 1960s, there were essentially three career choices for women: nurse, secretary, or teacher. Admissions quotas discriminated against women who wanted to pursue a different career path. Title IX of the Education Amendments of 1972 partially removed these barriers by making gender discrimination in admissions illegal. This allowed women to pursue degree fields previously open mainly to men, like law and medicine. Granting women access to high-skilled occupations, which were also the highest-paid occupations, not only reduces gender disparity in graduate degrees and occupations but also increases the expected returns of their human capital investment, further encouraging college matriculation. This paper uses a novel measure of distributional change to empirically assess whether legislation was effective in eliminating discrimination in graduate education.

The striking and new finding of my analysis is the sharp and dramatic convergence of female versus male graduate-degree fields coincident with the passage of Title IX. I use two different methods to measure gender convergence: the Segregation Index and the Earth Mover's Distance algorithm (EMD). Although the segregation index is a popular method for measuring distributional change, EMD is a better measure when studying discrimination as it takes into account which bins people move out of and into, and, more importantly, the distance between bins (using expected wage of the different fields as a measure of distance). This is an important detail because women were barred from entering certain fields prior to Title IX, and those fields were precisely the more lucrative fields. My results do not change by convergence measure, though EMD estimates greater convergence. This distributional convergence was driven by sex-composition effects as females predominantly moved into male-dominated fields, but it does not seem to be driven by gender-specific preferences. Further, alternative explanations were gradual changes and cannot explain the large, national shift in graduate-field distribution that occurred between 1972-73 and 1974-75.

In addition to providing evidence of successful anti-discrimination, this paper sheds new light on the factors responsible for the college gender gap reversal. More importantly, it adds to the empirical toolbox of convergence measures. Measures of distributional change have broad applications, ranging from studies of residential segregation (Massey & Denton, 1988) to occupational segregation (Blau, Brummund, & Liu, 2013) to income-achievement gaps (Nielsen, 2015). As such, the literature on convergence measures is long and ever-growing (Duncan & Duncan, 1955; Taeuber & Taeuber, 1976; Cowell, 1985; Massey & Denton, 1988; Ruber, Tomasi, & Guibas, 2000; Reardon & Firebaugh, 2002; Reardon, 2009). I contribute to this literature by introducing a well-known measure in computer science (EMD) and using it to answer an important policy question. The advantage of EMD is that it takes into account the distance between bins that people are moving into and out of when measuring convergence. This is something the segregation index does not do, but is an important detail when studying discrimination. With EMD, I am able to conduct a robust quantitative analysis of Title IX.