Data Envelopment Analysis (DEA) in the Psychological Sciences

Using methods like linear regression or latent variable models, researchers are often interested in maximizing explained variance and identifying the importance of specific variables within their models. These models are useful for understanding general ideas and trends, but often give limited insight into the individuals within said models. Data envelopment analysis (DEA), is a method with roots in organizational management that make such insights possible. Unlike models mentioned above, DEA does not explain variance. Instead, it explains how efficiently an individual utilizes their inputs to produce outputs, and identifies which input is not being utilized optimally. This paper provides readers with a brief history and past usages of DEA from organizational management, public health, and educational administration fields, while also describing the underlying math and processes behind said model. This paper then extends the usage of this method into the psychology field using two separate studies. First, using data from the Project KIDS dataset, DEA is demonstrated using a simple view of reading framework identifying individual efficiency levels in using reading-based skills to achieve reading comprehension, determining which skills are being underutilized, and classifying and comparing new subsets of readers. Three new subsets of readers were identified using this method, with direct implications leading to more targeted interventions. Second, DEA was used to measure individuals’ efficiency in regulating aggressive behavior given specific personality traits or related skills. This study found that despite comparable levels of component skills and personality traits, significant differences were found in efficiency to regulate aggressive behavior on the basis of gender and feelings of provocation.