Working Papers and Publications

 

The Selection of Peer States for Performance Benchmarking in Higher Education

This working paper informs the selection of peer states for performance benchmarking by estimating the degree of similarity among states along three dimensions: population characteristics, capital advantage, and market conditions. 

 

Measuring the Efficiency of Postsecondary Institutions: A Regression-based Approach to Adjusting for Differences in Cost Structures

The question of whether institutions manage resources efficiently has not been widely examined. Some have resorted to using relatively simple performance indicators, including the total educational expenditures per academic credential. However, such raw indicators may unfairly characterize an institution’s degree of efficiency, for educational expenditures can vary tremendously by the types of credentials produced (e.g., certificates, bachelor’s degrees, doctoral degrees) and the disciplines represented. Accordingly, an unadjusted efficiency indicator may not reflect differences in resource management but rather legitimate variation in cost structures. This study proposes an alternative efficiency indicator, wherein a common regression method is used to estimate whether educational expenditures are lower or higher than expected, given the institution’s degree production profile, faculty attributes, and differences in the cost of living.

 

The Reliability and Validity of Using Regression Residuals to Measure Institutional Effectiveness in Promoting Degree Completion

Despite the widespread use of graduation rates in accountability systems, it is doubtful that relevant dimensions of institutional effectiveness are being adequately assessed. To be sure, numerous factors that frequently lie beyond institutional control strongly influence degree completion, such as the socioeconomic status and academic preparedness of incoming students. Raw graduation rates may thus better reflect advantages of circumstance than strengths of policy and practice. Accordingly, some have advocated for the use of an adjusted completion indicator that accounts for potentially confounding inputs. A typical approach utilizes residuals from a regression model, thereby providing a value-added score that reflects the difference between actual and predicted graduation rates. It remains unclear, however, whether the residual approach ultimately produces an indicator of institutional effectiveness with desirable psychometric properties. The present study begins to bridge this gap by examining the test-retest reliability and both the convergent and discriminant validity of residuals as measures of institutional effectiveness for two- and four-year institutions.