Data Overview

The National Longitudinal Survey of Freshmen (NLSF) was developed to provide comprehensive data to test different theoretical explanations for minority underachievement in higher education. Rather than prejudging the validity of any single point of view, we sought to develop a broad database capable of testing each conceptual model, assessing its explanatory power, and specifying the circumstances under which it might apply. Specifically, the NLSF sought to measure the academic and social progress of college students at regular intervals to capture emergent psychological processes hypothesized by investigators such as Steele and Ogbu, while measuring the degree of social integration and intellectual engagement suggested by Tinto, and to control for pre-existing background differences with respect to social, economic, and demographic characteristics.

To realize this vision, we designed a survey that included equal-sized samples of white, black, Asian, and Latino freshmen entering selective colleges and universities. The baseline survey consists of face-to-face interviews that compiled detailed information about the neighborhood, family, and educational environments students experienced before entering college. The survey also assesses their attitudes, aspirations, and motivations at the time of entry. The follow-up surveys take place via telephone interview in the Spring of each academic year to gather information from the same students about their social, psychological, and academic experiences on campus.

By combining retrospective data captured in the baseline survey with prospective information compiled in years one through four, we seek to create a longitudinal database stretching from childhood through college graduation. This design provides a basis for linking pre-college experiences to behaviors and psychological states emerging in the course of higher education, and for sorting out the direction of causality between determinants and outcomes. Those dropping out of college or transferring to another institution are followed, interviewed, and retained in the survey to avoid building selection biases into the sample.


Sample and Sampling Frame

The institutions we chose to sample mirror those examined by Bowen and Bok (1998) in their College and Beyond Survey. Our principal modification was the addition of the University of California at Berkeley, which is not only a large and selective institution (currently rated as number one among public universities by US News and World Report), but also a school that recently abandoned its historical commitment to affirmative action (as a result of Proposition 209, which was approved by California’s voters in 1995). The other modification was to include historically black colleges and universities. 

We initially asked 35 schools to participate in the survey. The sample was stratified by the relative size of the black student body. Institutions with relatively large black student populations (1000+) were assigned a target sample size of 280 respondents (70 in each of four racial/ethnic groups); those with black student populations of 500-1,000 got a target size of 200 interviews (50 in each group); those with 100-500 black students had a target size of 80 respondents (20 in each group), and those with fewer than 100 black students were assigned a quota of 40 interviews (10 in each group). The historically black schools were given a target of 70 interviews per institution.

Although most schools were enthusiastic about participating, five schools declined the invitation outright (Duke, Vanderbilt, Wellesley, Hamilton, and Xavier). A major disappointment, however, was the response received from four historically black institutions we had targeted for study. Although only Xavier declined to participate outright, we were only able to secure a sample of freshmen in one historically black institution. Despite the fact that the Presidents of both Morehouse and Spelman agreed on behalf of their institutions to participate, the Registrars Offices at both colleges could not provide a list of freshmen from which we could draw a sample. This left only Howard University to represent historically black institutions. 

The final institutional participation rate was 80%. The loss of seven institutions out of 35 cut our expected sample size from a planned 4,160 to only 3,550 students. To make up for the lost cases we increased the number of interviews conducted at other institutions. In all, we approached 4,573 respondents across 28 institutions. Of these, 3924 completed the survey, for an overall response rate of 86%, which by the standards of survey research is very high, particularly for a long (2+ hours) face-to-face interview that for all intents and purposes was unpaid (respondents received a token payment of $15 for participating). The final sample included 959 Asians, 998 whites, 1,051 African Americans, and 916 Latinos. In order to be eligible for inclusion in the sample, a respondent had to be enrolled at the institution in question as a first-time freshman and be a U.S. citizen or resident alien. Foreign and returning students were excluded from the sample.