A comparison of two procedures for peer group assignment of institutions of higher education

TR Number

Date

1989

Journal Title

Journal ISSN

Volume Title

Publisher

Virginia Polytechnic Institute and State University

Abstract

The exchange of institutional information for the purpose of comparative analysis is an established practice in the decision-making and planning processes of institutions of higher education. For comparative analysis to be meaningful, the institutions must be sufficiently alike to be comparable. Hence, colleges and universities are placed into peer groups, comprised of institutions categorized on the basis of relevant criteria. Comparative analyses and the decisions resulting from the analyses are only as good as the peer groupings that provide the sources for comparison. Thus, the way in which peer groups are formed is a cornerstone of the comparative analysis process.

The purpose of this study was to compare two grouping procedures, focal proximity, and cluster analysis, in forming institutional peer groups in higher education. The two procedures were examined in terms of their relative reliability, determined by the stability of group membership over a five-year period of time, and in terms of their comparability at a single point in time, measured by the degree of agreement between peer groups formed in 1983 by each of the two procedures. Discriminant analysis was used to examine the relative importance of the variables used as criteria in the grouping procedures.

Both procedures were found to form peer groups with greater than chance stability over the five year period of time. For one sample of institutions, focal proximity was found to be more reliable than cluster analysis in assigning institutions to peer groups. There was no difference between the two methods for a second sample of institutions. For both samples of institutions, agreement between groups formed by focal proximity analysis in 1983 and those formed by cluster analysis at the same point in time was greater than could be expected by chance. The variables that loaded most highly on the first discriminant function of the total canonical structure were PHD (proportion of degrees at the doctoral level) and LRES (research expenditures). The variables that were most important when holding all other variables constant were BA (proportion of degrees at the bachelor level) and MA (proportion of degrees at the master's level).

Description

Keywords

Citation