The National Postsecondary Student Aid Study (NPSAS) is a comprehensive nationwide study conducted by the U.S. Department of Education's National Center for Education Statistics (NCES) to determine how students and their families pay for postsecondary education.1 It also describes demographic and other characteristics of students enrolled. The NPSAS study is based on a nationally representative sample of all students in postsecondary education institutions, including undergraduate, graduate, and first-professional students. Information is collected from institutions, student interviews, and government data files. For this study, only data for undergraduates from two administrations of the NPSAS survey were analyzed, NPSAS:93 and NPSAS:2000. Both surveys represent more than 16 million undergraduates who were enrolled at some time between July 1 and June 30 of the survey years. The NPSAS:2000 survey was selected for this study because it contains the most recent data available. NPSAS:93 was chosen because it is the earliest survey with data that is strictly comparable to the data in NPSAS:2000. In NPSAS:90 the non-tuition student expenses were derived from student-reported data rather than from the institution-reported student budgets that were used for need analysis. Consequently, the price of attendance, net price and need variables in NPSAS:90 are not comparable to those in the later NPSAS surveys.
For NPSAS:93, the institutional weighted response rate was 88.2 percent and the overall effective response rate for student interviews was 71.4 percent.2 For NPSAS:2000, the institutional response rate was 97 percent and the weighted overall student interview response rate was 65.6 percent.3 Because the student telephone interview response rate for NPSAS:2000 was less than 70 percent in some institutional sectors, an analysis was conducted to determine if Computer Assisted Telephone Interview (CATI) estimates were significantly biased due to CATI nonresponse. Considerable information was known for CATI nonrespondents and these data were used to analyze and reduce the bias. The distributions of several variables using the design-based, adjusted weights for study respondents (study weights) were found to be biased before CATI nonresponse adjustments. The CATI nonresponse and poststratification procedures, however, reduced the bias for these variables; and the remaining relative bias ranged from 0 to 0.35 percent.4