Projections of Education Statistics to 2011: Appendix E 1999 IPEDS (Fall Enrollment)Survey Methodology

• Go to the Latest Version of Projection of Education Statistics •

Appendix E
1999 IPEDS (Fall Enrollment)
Survey Methodology

Overview

Fall 1999 enrollment data collected through the National Center for Education Statistics (NCES) represent 3,958 degree-granting postsecondary institutions that are eligible to participate in Title IV programs (financial aid) in the United States. Table E1 includes only those institutions in the 50 states and the District of Columbia. The 1999 Fall Enrollment survey (either EF1 or EF2) was sent to the universe institutions accredited at the collegiate level and to all other institutions offering a bachelor's, master's, doctor's or first-professional degree. The Fall Enrollment (EF) survey is conducted annually as part of the National Center for Education Statistics' Integrated Postsecondary Education Data System (IPEDS). Eligibility status of institutions was obtained from the Office of Postsecondary Education's 1998 Postsecondary Education Participants System (PEPS) file.

The 1999 Fall Enrollment Survey data currently available in a peer tool on the NCES web site do not permit national estimates. The peer tool includes only those institutions that reported data that passed various edit checks. To calculate national totals, data for nonresponding institutions were imputed using procedures described in this appendix and added to data reported by institutions.

Students included in the Fall Enrollment survey were students enrolled in courses toward a degree or other formal award; students enrolled in courses that are a part of a vocational or occupational program, including those enrolled in off-campus centers; and high school students taking regular college courses for credit. Students excluded from the Fall Enrollment survey were students enrolled exclusively in courses not creditable toward a formal award and not in a postsecondary vocational program, students enrolled exclusively in remedial courses; students exclusively auditing classes; students studying abroad (e.g., at a foreign university) if their enrollment at the institution is only an administrative record and the fee is only nominal; and students in any branch campus located in a foreign country; and students earning continuing units (CEU's) only.

There are two versions of the Fall Enrollment survey. The most extensive form, EF1, was sent to all 4-year institutions. The EF2 form is less detailed and was sent to 2-year postsecondary institutions that grant an associate's degree (degree-granting).

Universe, Institutions Surveyed and Response Rates

A universe of postsecondary institutions was initially established as being eligible to participate in Title IV programs by the IPEDS 1998-99 Institutional Characteristics Survey. Fall Enrollment Survey forms were mailed in July 1999. The survey results were collected from November 1999 through June 2000. During this time period, some institutions determined to be out-of-scope were deleted from the universe. These deletions resulted from formal notification by IPEDS state coordinators, the Department of Education eligibility notices, and from follow-up telephone calls. Included in the deletions were (1) duplicates of other institutions on the file; (2) institutions that no longer offered postsecondary programs; or (3) schools that did not conform to the IPEDS definition of an institution or branch. At the end of the process, 3,958 institutions were in the final 1999 fall enrollment universe. The final universe was also adjusted to reflect institutions that changed from one sector to another subsequent to survey mail out.

Table E2 shows the number of institutions that responded to the mail out of the 1999 Fall Enrollment survey by level and their enrollment. It also reports the total number of institutions in the survey universe, and the final imputed enrollment. The table shows the response rate as the proportion of the survey universe that reported to the survey both in terms of counts of institutions and in terms of their enrollment.

Survey Conduct and Editing

The 1999 Fall Enrollment survey was due November 15. Survey data were collected via paper or transmitted NCES via the Internet. Some institutions submitted data on diskette. Data for nonresponding less than 2-year institutions were collected through the Postsecondary Education Telephone System (PETS). All data, whether received on paper forms, diskettes, electronically via the Internet, or through the PETS system, went through the same editing process. Extensive follow-up for survey nonresponse was conducted from November 1999 through April 2000. Initially, reminder letters were mailed, encouraging nonresponding institutions to complete and return their forms; and subsequently, the PETS was used to collect critical data by telephone from an institutional representative.

Survey responses were edited for internal and inter-year consistency. The following editing procedures were used:

Part A: Enrollment, by Sex and Race/ethnicity

Addition checks were performed by adding down the columns and comparing generated totals with reported totals. If the reported total differed from the generated total but was within a designated range, the reported total was replaced by the generated total and the cell was flagged with the proper impute code. If the difference exceeded the designated range, institutions were contacted for verification/correction. Addition checks were also performed by generating totals for men and women by adding across columns (racial/ethnic group) in each line. Generated totals by sex were compared to the reported totals. If they differed but were within a designated range, a balance field was created containing the difference by sex and placed on the data file. Reported detail and totals were not altered. If the difference exceeded the designated range, institutions were contacted for correction. Editing of the racial/ethnic data is explained in detail in the section on raking of racial/ethnic data. Comparisons were also made with the number of students reported for the selected items for the prior year. If the differences were sufficiently large to trigger an edit flag, institutions were contacted for further verification.

Part B: Enrollment, by Age

Part B data were edited in a similar manner. Addition checks were performed by comparing reported totals in Part B to corresponding totals in Part A. Differences between the totals were calculated and balances containing the differences were generated. If the balance was outside a certain range, the institution was contacted for verification/correction. If the balance was within a certain range, the Part B total was replaced by the corresponding total from Part A. The balance was then allocated among the age categories using the institution's reported age distribution.

Data Management and Imputation

The response rate for institutions was approximately 97 percent, and these institutions accounted for more than 99 percent of enrollment. Because the response rate was so high for the enrollment data, a very straightforward process was followed for imputation. For nonreporting institutions, data from the institution's prior 1998 Part A response was used as the imputation for fall 1999. In some cases, the prior response was also an imputation. Because of the extremely high representation of the larger institutions in all major sectors of degree-granting institutions, any bias caused by this procedure was considered to be minimal.

Data for Part B was imputed by using the distribution from the 1997 enrollment by age survey, and using that age distribution for each institution to distribute their (reported or imputed) fall 1999 enrollment by age.

Raking of Data

When data were reported by racial/ethnic categories on a detail line and the generated sum of these enrollments did not equal the reported total enrollment, the difference between the generated total and the reported total was calculated. If the difference exceeded a certain designated range, the institution was contacted for correction. If the difference was within the designated range, a "balance column" was created. The balance column as well as the "race unknown" column was then distributed in the same proportions as the reported racial/ethnic data for that detail line. When the racial/ethnic numbers were adjusted to full counts, there were often lines that failed add checks because of rounding. The largest figure in each row was adjusted by one or two, so that the line added to the original reported total for that line.

Further information on tables E1 and E2 may be obtained from: