BACKGROUND: Studies of educational mobility in the United States report widespread persistence in the association between parental and offspring schooling over most of the 20th century. Despite this apparent persistency, many other studies report substantial improvements in the educational performance of historically disadvantaged groups. To reconcile these diverging trends, I propose examining educational mobility in terms of percentile ranks in the respective schooling distributions of parents and offspring. Using a novel estimator of educational rank, I compare patterns of mobility in the overall schooling distribution both over time and among population groups defined by race and gender.METHODS & DATA: To analyze educational rank mobility, I use quantile transition matrices known from studies on intergenerational income mobility. However, because schooling distributions are quite lumpy, particularly around 12 and 16 years of schooling, percentile ranks of interest may not always be defined among parents or offspring (e.g., the lower or upper quartile may not be given by the data). To deal with this issue, I use a cohort-adjustment that deflates the schooling distribution in proportion to average growth in schooling. The resulting within-cohort distribution of schooling is much smoother than the original distribution and, consequently, it has well-defined percentile ranks. Using about 30,000 parent-offspring pairs in the General Social Surveys, I study quartile-by-quartile matrices for individuals born 1903 through 1980.RESULTS: My analyses yield two key results. First, I find widespread equalization among groups in upward mobility: Among those born before Second World War, Whites are significantly more likely than are Blacks to be upwardly mobile out of the lower quartile of the schooling distribution (about 18 percentage points), but among those born after the War, the gap disappears. Second, I find widespread persistency among groups in downward mobility: Blacks are significantly more likely than are Whites to transition out of the upper quartile of the schooling distribution (about 18 percentage points), and this gap is virtually unchanged between cohorts born over the 20th century. Further analyses show that these differences in mobility with respect to the overall distribution of schooling cannot be explained by differences in the intergenerational transmission of schooling between Blacks and Whites, but rather by the fact that these groups regress to very different means.IMPLICATIONS: My study has at least two implications. First and most importantly, I can compare my estimates to corresponding estimates reported in studies of intergenerational income mobility. These studies show that for cohorts born after the War, Blacks have a much smaller probability of upward income mobility than do Whites. This would suggest that the equalization in upward educational mobility reported in my study has not brought about similar equalization in upward income mobility. Second, my study offers a generalized way to compare nonlinearities and group differences in educational mobility relevant to comparative research across countries.
Abstract presented at the Conference "Social Mobility in the Americas", Stanford University, April 24-25, 2015