In this paper, I evaluate the extent to which colleges and universities of varying degrees of selectivity engaged in racial, ethnic and socioeconomic affirmative action for cohorts of students who graduated in 1972, 1982 and 1992. I find that a much wider range of institutions engage in affirmative action for African American students than previous analysts reported, and that a growing number of institutions extend the benefits of affirmative action to Hispanic students. Colleges and universities are less enthusiastic about engaging in affirmative action for socioeconomically disadvantaged students. To understand why postsecondary institutions might prefer students from particular minority groups over otherwise comparable white students, I introduce the concept of compensatory sponsorship (buidling on Turner’s ideal type of sponsored mobility). In a contest system perceived by many to unfairly disdavantage some competititors, college personnel engage in affirmative action both to right a perceived wrong and to preserve the legitimacy of the contest. The beneficiaries of compensatory sponsorship, however, are determined by historical and social forces that constrain how postsecondary institutions recruit, admit and fund potential matriculants.

Using eighteen years of data from more than 1,300 four-year colleges and universities in the United States, we investigate the extent to which institutional characteristics and contextual factors influence the propensity of colleges to indicate that they consider race/ethnicity in their admissions decisions. Consideration of race/ethnicity in admissions declined sharply after the mid-1990s, especially at public institutions. Rather than being shaped by specific historical and political contexts, consideration of race/ethnicity in admissions appears to be a widely institutionalized practice in higher education that has been tempered by changes in the policy environment over time.

Patterns of postsecondary attendance in the United States continue to be stratified by socioeconomic background and race/ethnicity. We suggest that inequalities in knowledge of the costs of going to college contribute to persistent patterns of stratification. We hypothesize that disadvantaged parents who believe their child will attend college are less certain of the costs of college attendance. As a result, they are less able or willing to provide an estimate of the costs of college attendance, more likely to over-estimate those costs if they do provide an estimate, and make larger errors in estimation than comparable middle class or white parents. Using nationally representative data, we find mixed support for these hypotheses. Socioeconomically disadvantaged parents and minority parents are less likely to provide estimates of college tuition and, when they provide estimates, tend to make larger errors. On average, though, parents provide upwardly biased estimates of cost that are uniform across race, ethnicity and socioeconomic status. We discuss implications of these findings for sociological theory and for inequality in postsecondary education.