Data element definitions have been formulated and tested to be relevant to all providers of postsecondary education and consistent among components of the system. A set of data elements has been established to identify characteristics common to all providers of postsecondary education, and specific data elements have been established to define unique characteristics of different types of providers. Interrelationships among various components of IPEDS have been formed to avoid duplicative reporting and to enhance the policy relevance and analytic potential of the data. Through the use of “clarifying” questions that ask what was or was not included in a reported count or total or the use of context notes that supplement the web collection, it is possible to address problems in making interstate and inter-institutional comparisons. Finally, specialized, but compatible, reporting formats have been developed for the different sectors of postsecondary education providers. This design feature accommodates the varied operating characteristics, program offerings, and reporting capabilities that differentiate postsecondary institutional sectors, while yielding comparable statistics for some common parameters of all sectors.
Only the data collected prior to 1993 from a sample of private less-than-2-year institutions are subject to sampling error. With this one exception, the HEGIS and the IPEDS programs include the universe of applicable postsecondary institutions.
The IPEDS data are subject to such nonsampling errors as errors of design, reporting, processing, nonresponse, and imputation. To the extent possible, these errors are kept to a minimum by methods built into the survey procedures.
The sources of nonsampling error in the IPEDS data vary with the survey instrument. In the Fall Enrollment component, the major sources of nonsampling error are classification problems, the unavailability of needed data, misinterpretation of definitions, and operational errors. Possible sources of nonsampling error in the Finance component include nonresponse, imputation, and misclassification. The primary sources of nonsampling error in the Completions component are differences between the NCES program taxonomy and taxonomies used by colleges, classification of double majors and double degrees, operational problems, and survey timing. A major source of nonsampling error in the Graduation Rates components is the correct identification of cohort students (full-time, first-time, degree/certificate-seeking undergraduates); for Human Resources, difficulties in classifying employees by primary occupation; for 12-month Enrollment, definitional difficulties with calculating instructional activity. For Student Financial Aid, institutions often must merge enrollment and financial aid databases, and face difficulties in placing students in the various groups for which data are collected.
Coverage Error. Coverage error in IPEDS is believed to be minimal. For institutions that are eligible for Title IV federal financial aid programs, coverage is almost 100 percent. Schools targeted as “possible adds” are identified from many sources, including a review of the PEPS file from OPE, a universe review done by state coordinators, and the institutions themselves.
Nonresponse Error. Since 1993, all institutions entering into PPAs with the U.S. Department of Education are required by law to complete the IPEDS package of components. Therefore, overall unit and item response rates are quite high for all components for these institutions. Data collection procedures, including extensive email and telephone follow-up, also contribute to the high response rates. Imputation is performed to adjust for both partial and total nonresponse to a survey. Because response rates are so high, error due to imputation is considered small.
Unit nonresponse. Because Title IV institutions are the primary focus of IPEDS and they are required to respond, overall response rates for Title IV institutions and administrative units are high. For example, the overall response rate for eligible institutions in the winter 2013–14 collection was 100.0 percent. Since the implementation of the web collection, Title IV institutional response rates for the various IPEDS surveys have ranged from about 89 percent to over 99 percent.
By sector, the response rates are highest for public 4 year or higher institutions and lowest for private for–profit institutions, especially less–than–2–year institutions.The 1994 Academic Libraries Survey and the FY 95 Finance public–use data files are limited to IHEs because the response rates for postsecondary institutions not accredited at the collegiate level were quite low (specifically: 74 percent in the Finance collection and less than 50 percent in the Academic Libraries Survey).
Item nonresponse. Most participating institutions provide complete responses for all items. Telephone and email follow-up are used to obtain critical missing items.
Measurement Error. NCES strives to minimize measurement error in the IPEDS data by using various quality control and editing procedures. New questionnaire forms or items are field tested and/or reviewed by experts prior to use. To minimize reporting errors in the Finance component, NCES uses national standards for reporting finance statistics. Wherever possible, definitions and formats in the Finance component are consistent with those in the following publications: College and University Business Administration; Administrative Services, Financial Accounting and Reporting Manual for Higher Education; Audits of Colleges and Universities; and HEGIS Financial Reporting Guide.
The classification of students appears to be the main source of error in the Enrollment component. Institutions have had problems in correctly classifying first-time freshmen, other first-time students, and unclassified students for both full-time and part-time categories. These problems occur most often at 2-year institutions (both public and private) and private 4-year institutions. In the 1977–78 HEGIS validation studies, misclassification led to an estimated overcount of 11,000 full-time students and an undercount of 19,000 part-time students. Although the ratio of error to the grand total was quite small (less than 1 percent), the percentage of errors was as high as 5 percent at student detail levels and even higher at certain aggregation levels.
The definitions and instructions for compiling the IPEDS data have been designed to minimize comparability problems. However, survey changes necessarily occur over the years, resulting in some issues of comparability. Also, postsecondary education institutions vary widely, and hence, comparisons of data provided by individual institutions may be misleading. Specific issues related to the comparability of the IPEDS data are described below.
Imputation. Imputed data are on file for institutions with partial or total nonresponse. Caution should be exercised when comparing institutions for which data have been imputed, since these data are intended for computing national totals and not intended to be an accurate portrayal of an institution’s data. Users should also be cautious when making year-to-year enrollment comparisons by state. In some cases, state enrollment counts vary between years as a result of imputation rather than actual changes in the reported enrollment data. To avoid misinterpretation, users should always check the response status codes of individual institutions to determine if a large proportion of data was imputed.
Classification of Institutions. Since 1996, the subset of IPEDS institutions that are eligible to participate in Title IV federal financial student aid has been validated by matching the IPEDS universe with the PEPS file maintained by OPE. Prior to 1996, institutions were identified as aid-eligible from the list of IHEs and responses to the Institutional Characteristics component.
Fields of Study. In analyzing Completions data by field of study, users must remember that the data are reported at the institution level and represent programs, but not schools, colleges, or divisions within institutions. For example, some institutions might have a few computer and information science programs organized and taught within a business school. However, for IPEDS reporting purposes, the degrees are classified and counted within the computer and information science discipline.
Reporting Periods.The IPEDS survey is separated into 12 components, which correspond to three seasonal reporting periods. The Institutional Characteristics, Completions, and 12-month Enrollment surveys are administered in the fall. The Student Financial Aid, Graduation Rates, 200% Graduation Rates, Outcome Measures, and Admissions components are collected in the winter. The Fall Enrollment, Finance, Academic Libraries, and Human Resources components are administered in the spring.
Survey Changes. Over the years, the IPEDS survey forms have undergone revisions that may have an impact on data comparability. Users should consider the following:
History of Classification of Instructional Programs. The purpose of the Classification of Instructional Programs (CIP) is to provide a taxonomic scheme that supports the accurate tracking, assessment, and reporting of fields of study and program-completions activity. NCES has utilized a number of versions of CIP throughout the life of IPEDS, as well as its predecessor, HEGIS.
In 1970, NCES published A Taxonomy of Instructional Programs in Higher Education that was to be used beginning with the HEGIS surveys of 1971–72. This taxonomy was divided into two main sections: one dealt with conventional academic subdivisions of knowledge and training; the other dealt with technologies and occupational specialties related to curricula leading to associate’s degrees and other awards below the baccalaureate. Both sections used 4–digit numerical codes to represent the fields.
In 1981, NCES published A Classification of Instructional Programs. In addition to new programs that evolved or gained new significance since 1970, there were weaknesses in the way instructional programs were classified and disaggregated. The new CIP instituted the current 6–digit code, which allowed obtaining data by 2–digit or 4–digit groups of fields more easily than the older scheme. The new CIP also included program definitions or descriptions that the 1970 version lacked, as well as other improvements.
In 1985, another revision to the CIP was released, although this was more of an update to the 1980 CIP than a radical change. There were 116 fields deleted, either due to duplication or because programs no longer existed to the degree needed for national reporting. Forty fields were added based on write-in entries on surveys returned. In addition, there were a few revisions of codes or names of fields. This CIP was used during the final years of HEGIS and continued into IPEDS.
A more extensive revision of CIP was released in 1990, which included programs at the secondary and adult education levels. Within the postsecondary level, there were several major restructures. Fields previously included in Business and Management (06) and Business (Administrative Support) (07) were integrated into a new Business Management and Administrative Support (52). Similarly, fields previously in Allied Health (17) and Health Sciences (18) were integrated into Health Professions and Related Sciences (51). Again there were deletions and additions, although many were actually combining two former fields into one, or vice versa. The 1990 CIP was first used in IPEDS in 1991–92.
A further revision resulted in publishing Classification of Instructional Programs: 2000 Edition in 2002. This CIP was adopted as the standard field of study taxonomy by Statistics Canada, based on the comprehensiveness and detail of the CIP and the potential for enhanced comparability with U.S. education data. Again, there were several major reorganizations. Fields previously reported in Agricultural Sciences (02) were divided between Agriculture, Agriculture Operations and Related Sciences (01) and Biological and Biomedical Sciences (26). Fields previously reported in Sales and Marketing Operations/Marketing and Distribution (08) were incorporated into Business, Management, Marketing, and Related Services (52). History became a separate 2–digit CIP (54) moved from Social Sciences and History (45). In addition, there were a large number of new fields added. The CIP–2000 was first used in IPEDS in 2002–03.
The web–based version of CIP was implemented during 2008–09, and is accessible online at http://nces.ed.gov/ipeds/cipcode/. The web–based CIP incorporates tools for browsing, searching, and crosswalking. Several additional changes were implemented in conjunction with this online version: 50 new 4–digit codes and 300 new 6–digit codes were added, and several series were reorganized (English Language and Literature/Letters (23), Psychology (42), Nursing (51.16), and Residency Programs (60)). One series was also deleted (Technology Education/Industrial Arts (21)), and several examples of instructional programs were added to assist users in selecting the appropriate field. These revisions have been utilized in IPEDS since 2010–11.
Comparisons with HEGIS. Caution must be exercised in making cross–year comparisons of institutional data collected in IPEDS with data collectedin HEGIS. IPEDS surveys request separate reporting by all institutions and their branches as long as each entity offers at least one complete program of study. Under HEGIS, only separately accredited branches of an institution were surveyed as separate entities; branches that were not separately accredited were combined with the appropriate entity for the purposes of data collection and reporting. Therefore, an institution may have several entities in IPEDS, where only one existed in HEGIS.
Comparison with the Survey of Earned Doctorates. Like the IPEDS Completions survey, the Survey of Earned Doctorates (SED; for more information, see the SED chapter) also collects data on doctoral degrees, but the information is provided by doctorate recipients rather than by institutions. The number of doctorates reported in the Completions component is slightly higher than in SED. This difference is largely attributable to the inclusion of nonresearch doctorates (primarily in theology and education) in the Completions component. The discrepancies in counts have been generally consistent since 1960, with ratios of the IPEDS-to-SED counts ranging from 1.01 to 1.06. Differences in the number of doctorates within a given field may be greater than the overall difference because a respondent to SED may classify his or her specialty differently from the way in which the institution reports the field in the Completions survey.