Educational attainment was historically severely limited among African-Americans, particularly in the South. By 1831, it was illegal for slaves to be taught to read or write in all of the Southern states [9]. Restrictions on education gradually lessened but remained during the Jim Crow era. During that period, schools were racially segregated, and the quality of black schools was markedly inferior to that of white schools [10].

Despite such legalized barriers to education, educational possibilities for African-Americans in the South increased markedly over the course of the 20^th century. This time-period effect engendered dissimilar educational opportunities, on average, for different age cohorts now alive [11]. For example, African-Americans who are currently in their 80s experienced a different educational climate when they were school-aged than did those who are now in their 40s.

Because education is an important determinant of an adult's access to resources and health, it is hard to overstate the importance of this transformation in opportunities. Social policies such as desegregation and civil rights expansion are likely responsible for some of the increases in 20^th century African Americans' educational attainment, but their influence has not been examined in detail.

Twin pairs are a potentially informative population in which to examine changes in educational attainment because each pair has the same parents and childhood socioeconomic status, two known determinants of educational attainment. Because genetic variance is likely constant over time, then an increase in phenotypic variance is likely attributable to increases in environmental variance. We therefore hypothesized that twins born later in the 20^th century would be less similar to each other in educational attainment than those born earlier in the 20^th century and that heritability estimates would therefore decrease over time. We hypothesized that the educational attainment of older twin pairs would be more highly correlated than that of younger twin pairs because of these changes in educational access in the South over time, potentially reflecting a “ceiling effect” associated with the Jim Crow laws and discrimination. The aim of this study was to test that hypothesis.

As expected, the mean educational attainment progressively increased with decreasing age of the groups: It increased from 5.67 for the 80–89 age group to 12.3, 12.6, 13.3, 13.6, 13.9, and 14.0 in the 70–79, 60–69, 50–59, 40–49, 30–39, and 20–29 age groups, respectively. Our main finding was that identical twins showed a correlation with their own twins in terms of educational attainment at the level of 0.93, 0.83, 0.89, and 0.70 for the 70–79, 60–69, 50–59, and 40–49 age groups, respectively; in contrast, fraternal twins showed correlations of 0.65, 0.54, 0.50, and 0.35, respectively, for the same age groups (see Table 1). An estimate of the contribution of genetic influences can be calculated using the formula: h²=2(rMZ-rDZ), where MZ indicates identical twins and DZ fraternal twins (c.f. Plomin[14]. Using this straightforward formula, the heritability values were 0.74, 0.56, 0.80, and 0.78 for the 70–79, 60–69, 50–59, and 40–49 age groups, respectively. The remainder of the individual variability was considered to be due to environmental factors.

We had also hypothesized that there would be a genetic influence on educational attainment. As predicted, we found that for every age group, the identical twin pairs had a greater correlation than did the comparable fraternal twin pairs. The results suggest that there is a substantial education-related genetic component affecting each age group. Several twin studies have examined differences in education between twins. Unpublished data from the Minnesota Twin Registry yielded heritabilities of 0.3 to 0.4; the participants in that study were predominately Caucasian. Other studies have analyzed intra-twin differences in education in relationship to divergent adult health [13], [15]. Such an analysis in our population did not generate significant results because of the high correlations in educational attainment. To our knowledge, ours is the only study to examine time-period effects on the correlation of educational attainment between twins.

The pattern we uncovered in decreasing intra-twin pair correlation in educational attainment was not statistically significant but we believe the pattern is noteworthy. Because educational attainment is usually achieved in the first three decades of life, education is a strong summary measure of childhood opportunities. In this group of study participants, such childhood opportunities changed dramatically over the course of the twentieth century. In this data, we have a unique exemplar of 422 African American twins spanning educational years of 1930 to 1980. This rich but moderate sized data set allows us to control for socioeconomic status and parental education by design but not to have enough twins to find statistical significance.

It should be noted that our analysis measured educational quantity, and not quality. A particular number of years of attainment does not necessarily confer the same benefit on all individuals in terms or cognition, physical health, or socioeconomic status because of differences in educational investment from state to state and county to county [16], [17]. A further complicating factor was identified by Whitfield and Wiggins (2003), who demonstrated that there were positive influences on healthy cognitive functioning of attending both segregated and desegregated schools.