From 1972-1992, total fall enrollment in college grew from 4.4 to 5.9 million in four-year public
colleges and from 2 to 3.3 million in four-year private colleges (Digest of educational statistics,
2002, table 173). This study examines how higher education access and opportunity changed
over these three decades. Specifically, we examine changing influence of SES and test score
effects and the role college planning play in the overall educational attainment process. We also
examine whether there are observable increases in the probability of higher education outcomes
(i.e., college plans, college attendance, and college graduation) for various groups of students.
Our study takes the approach of Sewell and Shah (1967), but extends it in several ways.
First, we study national samples of the high school graduates over the last three decades: the high
school class of 1972, 1982, and 1992. While past studies have examined trends in the
educational attainment process, none has examined data from the 1990s cohort. With a few
exceptions, most studies are based on cross-sectional data and do not examine trends in a
consistent manner. Second, this study uses three major national longitudinal surveys which
allow comparisons of the determinants of students' college plans, college attendance, and
ultimate degree attainments 8 or more years after high school. Third, these analyses examine
changes throughout the entire spectrum of SES and test score. Examining race and gender groups
separately, we determine “who gains” at each segment of the SES and test score distribution. Our
analyses replicate Sewell and Shah’s (1967) model using logistic regression and procedures that
account of the sampling design of the surveys. In addition, we can examine which group
experienced significant gains via the inclusion of confidence intervals at all portions of the
logistic curves.
Literature Review
Studies that attempt to explore the educational attainment process usually review
literature in the area of status attainment. Researchers of these studies often base their analytic
2
Convention