Measuring academic library efficiency and alignment with institutional resource utilization priorities using data envelopment analysis: An analysis of institutions of higher education in Texas and their libraries
Academic and library administrators are increasingly required to demonstrate efficiency in programs, services, and operations as well as effectiveness. An important component of efficiency measurement is identification of a relevant peer group against which to compare the administrative unit to determine relative efficiency. The two-fold purpose of this study is to identify efficiencies related to teaching and research in academic libraries and institutions of higher education (IHEs) and to determine the usefulness of data envelopment analysis (DEA) as an efficiency measurement tool for the academic administrator.
Using a population of academic libraries and IHEs in Texas as a case study, variables were identified that represented the teaching- and research- related inputs and outputs for IHEs and academic libraries. Three separate models types were developed for each administrative level in each year of the two-year study. The first focused on teaching efficiency; the second focused on research efficiency; the third combined teaching and research to examine overall efficiency. Separate variables were selected for each administrative level to represent the teaching- and research-related inputs and outputs for the administrative level. Data were gathered for 2007 and 2008 for both academic libraries and IHEs to permit model stability testing. Each model was completed once in each study year. A total of twelve individual models were completed across the two years of study.
In the first phase of the study, variables were selected based on an extensive review of the literature and the researcher's professional judgment, following the process the academic administrator might employ to select variables. All variables for each model were calculated, transformed as needed, and tested for isotonicity using a correlation matrix. Variables were entered into the DEA analysis tool and relative efficiency scores were calculated using input-oriented CCR-CRS and BCC-VRS models. The initial calculations indicated that scale was a factor in efficiency and BCC-VRS was employed to determine final efficiency scores. Discrimination in each model was increased using a backward removal of variables procedure. Each model identified the relative efficiency of the academic libraries and IHEs in the study population.
In the second phase of the study efficiency scores for the population of IHEs and academic libraries were subjected to statistical analysis. Related-samples Wilcoxon signed rank tests and Spearman's rho correlations were performed to test the stability of the model and identify both significant differences in and correlations of scores at each administrative level across years. The Mann-Whitney U test was used to identify relationships in efficiency scores across administrative levels in each year of the study. An institutional control filter was applied to all analyses to determine the extent to which institutional control influenced its efficiency score. Additional analyses were conducted using Spearman's rho and Holms sequential Bonferroni to determine the influence of institution size and Carnegie degree classification on efficiency scores.
Results of the DEA process and analysis identified the method's strengths and pitfalls. The process highlighted the influence of population size and homogeneity, of data availability and imbalances, and of the method used to increase model discrimination. Difficulties arising from these influences were addressed and final efficiency scores were presented. Statistical analyses identified general trends in efficiency at both the academic library and the IHE levels of administration and suggest that size, classification, and control may influence efficiency.
While results of the study indicate that the complexity of DEA may limit its usefulness to academic administrators, the study provides a foundation from which additional efficiency analysis tools may be developed in the context of an overall assessment plan. Academic and library administrators wishing to pursue DEA analysis will find this study useful as they identify variables, processes, populations, data, and relevant DEA models. The study also provides a foundation for future research in IHE and library efficiency analysis and highlights research opportunities in data collection and preparation, variable selection, population identification, and discrimination methods.