• Introduction
• Adjustments to Datasets
• Conclusion

A number of NCES surveys can be used to estimate enrollment levels in postsecondary education. Generating consistent enrollment estimates across surveys, however, is complicated by differences in surveys that lead to different enrollment counts. This R&D report describes the process of generating comparable estimates of undergraduate enrollment in postsecondary institutions across four NCES datasets—the National Postsecondary Student Aid Study (NPSAS), a sample survey of postsecondary students; the Integrated Postsecondary Education Data System (IPEDS), a universe survey of postsecondary institutions; the National Household Education Surveys Program (NHES) Adult Education Survey, a sample survey of adults in households; and the October school enrollment supplement to the Current Population Survey (CPS), a sample survey of adults in households. The purpose of the report is to highlight differences across surveys that may affect postsecondary enrollment estimates and to describe how largely comparable estimates can be derived, given these differences.

For each dataset, the analysis estimated the number of individuals enrolled in postsecondary education in the 1989–90, 1995–96, and 1999–2000 school years, or the closest available time period to those dates. Enrollment counts were estimated for the traditional college age group, ages 18 to 24, as well as for those individuals ages 18 to 64. Each estimate was placed over the relevant population age group to obtain an estimate of the percentage of the population enrolled in postsecondary education, using resident population counts (for April 1990, 1996, and 2000) provided by the U.S. Census Bureau.

Preliminary estimates for undergraduate and graduate students combined, with minimal corrections for survey differences, revealed inconsistencies in enrollment levels within years and in trends across years from one survey to another. Survey differences that may contribute to these inconsistencies include the following:

• Sources of information: Whether a student, postsecondary institution, or household member provided enrollment information, and whether proxy respondents are allowed. For example, CPS and NHES collect information from household members, NPSAS collects information from students, and IPEDS collects information from postsecondary institutions. In CPS surveys, an adult member of each household serves as a proxy respondent, providing information for all members of the household.
• Reference period: Whether the survey asked about enrollment at one point in time or over an entire school year or calendar year. For example, while NHES collects full-year enrollments, CPS collects fall-only enrollments.
• Definition of enrollment: Differences in the types of enrollment counted in the survey, such as whether students had to be in for-credit courses (e.g., NPSAS) or courses leading to a degree (e.g., NHES). Also, differences in the target population (e.g., whether military personnel are included in the population).
• Definition of postsecondary institution: Which postsecondary institutions were included in the survey and how eligible institutions were defined. Some surveys set specific criteria in defining postsecondary institutions (e.g., IPEDS), whereas others rely primarily on respondent perception (e.g., CPS).
• Variations in survey administration: These differences are assumed to be largely corrected by sample weights, and include factors such as telephone-based sampling, time of year of administration, and differences in response rates.

The remainder of the report focuses on undergraduate enrollments only. The following adjustments were made to the datasets to obtain undergraduate enrollment estimates that are as comparable as possible.

NPSAS

To make NPSAS estimates comparable across time, the three waves of NPSAS data were restricted to Title IV eligible institutions (i.e., institutions eligible to participate in the federal student financial aid program) and excluded institutions in Puerto Rico. Because of inconsistencies in the inclusion of students in less-than-2-year institutions in the IPEDS and CPS datasets, students enrolled in less-than-2-year institutions were excluded as well. In cases where student age was missing, these data were imputed.

IPEDS

Because IPEDS generally does not collect enrollment by age categories from less-than-2-year institutions, these schools were excluded from the analysis, as were institutions in areas other than the 50 states and the District of Columbia. IPEDS provides both full-year and fall-only enrollment counts; however, because IPEDS full-year enrollment data are not disaggregated by age, this analysis used IPEDS fall-only enrollments. Age was imputed when missing.

NHES

The 1991 administration of NHES was not used to examine undergraduate enrollments because in the 1991 survey these enrollments could not be separated from graduate enrollments. In the remaining years, data were restricted to adults working on either an associate's or a bachelor's degree; cases in which adults indicated they were working on "another degree" were individually examined and recoded into these degree categories as necessary.

CPS

CPS includes separate questions about enrollments at a "regular" school and enrollments in "business, vocational, technical, secretarial, trade, or correspondence courses." Because the second question potentially includes a wide range of courses outside of postsecondary education, only responses to the first question were used in this analysis, effectively restricting the estimates to those enrolled in 2- or 4-year institutions. No other adjustments were made to the CPS estimates.

After making these adjustments, levels of enrollment were generally not significantly different for those surveys with similar reference periods (i.e., full-year NPSAS and NHES vs. fall-only IPEDS and CPS). As one would expect, full-year enrollments were often higher than fall-only enrollments. The remaining differences across surveys can be reasonably attributed to factors such as the population surveyed, the survey methodology, and the time of year in which the survey was administered.

Because of the potential effects of survey differences on postsecondary enrollment estimates, it is important that the analyst examining participation in postsecondary education note the reference period, levels of degrees, and institution types covered by the analysis, and the effects of this coverage related to other possible analyses and/or data sources. Which data sources to use, and which adjustments to make, will depend in large part on the questions the analyst wishes to answer.