In lieu of an abstract, here is a brief excerpt of the content:

Appendix A Data Sources and Methods M ost of the analyses presented in this book draw their data from two large surveys, one of which we call the CUNY survey, and the other the NLSY (National Longitudinal Survey of Youth). Both studies are longitudinal, following representative samples of women for decades from roughly their late teens until they had reached middle age by the year 2000. This substantial time span (CUNY’s was longer than the NLSY because it was started earlier, in 1972) proved especially important because many of these women dropped out of college , but some later returned and completed their degrees after a hiatus of a decade or more (when the degree is eventually completed, these sorts of interruptions are called “stopouts”). Those dynamics would have been lost if our data had ended earlier, in the years shortly after college. Both surveys collected multiple indicators of women’s current wellbeing (earnings, household income, home ownership, and occupational , educational, and family situation). As we report in chapter 3, different indicators may paint incongruent pictures; in particular, women’s earnings and household incomes often diverge, depending on marital history. These two surveys contain rich data about the families in which the women grew up—their parents’ income and education and occupation, and the women’s high school background—allowing us to distinguish between the effects on women’s lives of college attendance versus other advantages or disadvantages such as the woman’s class background or having a good high school education, or race or marriage. Finally, both studies were unusual in containing data about three generations: the parents of the women, the women themselves, and the children of these women. The surveys collected information about the children’s educational progress and academic skills. Other variables captured how parents raised their children: the kinds of cognitive stimulation and cultural enrichment they provided. Together these data allow us to examine how maternal education influenced the life chances of the offspring in these families (see chapters 4 and 5). In this methodological appendix we discuss the main features of these two surveys, and also explain a statistical technique that we employed, 202 Appendix A 203 knownasthecounterfactualmodelofcausalinference,or,moreinformally, as “propensity score matching.” The CUNY Longitudinal Survey Data Set Our survey, undertaken in 2000, was carried out especially for this book. It was based on a cohort of undergraduate women who first entered the various campuses of the City University of New York (CUNY) between 1970 and 1972. CUNY is the nation’s largest urban public university and enrolled about 250,000 undergraduates in that period. Prior to the early 1970s, CUNY had developed into an academically quite selective institution, and few students of color attended. In 1970, bowing to political pressures, the university undertook a radical shift in policy and opened its doors to all graduates of New York City’s high schools. On the basis of their high school records, some entering students were only admitted into CUNY’s community colleges, which offered two-year associate’s of arts, or A.A., degrees, whereas better-prepared students were allowed to enter the four-year colleges to work toward bachelor’s of arts, or B.A., degrees. Open admissions, as the new admissions policy became known, quickly made a huge difference to the demography of the CUNY student body: far more black and Hispanic students were admitted than in years past, and the representation of students from poor and workingclass backgrounds soared. Because this policy represented such a big shift, CUNY undertook a large research effort at that time to study the new students. With support from the Exxon and Ford foundations, researchers systematically sampled entering cohorts of students in the early 1970s, collecting information on their high school records, race and ethnicity, and family background (including parental education, occupation , and income), the students’ educational and career aspirations, and so on. That baseline study also tracked college grades and graduation through the year 1975 (Lavin, Alba, and Silberstein 1981). The CUNY baseline data set provided an unusual window into the effects of increased access to higher education. It is especially well suited to studying a population of working-class and poorer college students, many of whom had relatively weak high school preparation. The baseline sample included over 18,000 women, and their social security numbers were recorded. By 1975, some of the women had graduated, others had left CUNY without completing a college degree, and yet others had transferred to other...

Share