Psychometrics Research Lab
Personality inventories have become popular in the employee selection context over the past couple decades, with much of the attention focused on the conscientiousness trait of the Big Five personality model. We are currently developing a circumplex model of conscientiousness subdimensions, and conducting research into cross-contextual (e.g., work, school, home) profiles of personality and how these profiles relate to context-specific outcomes. This research involves developing personality test content, administering surveys, and evaluating data using confirmatory factor analysis and structural equation models.
High-stakes occupational tests such as licensure and certification tests are constant targets of attempts at test fraud, such as stealing test content and other forms of cheating. We are currently conducting research into the statistical detection of test-taker collusion and pre-knowledge of test content, automated generation of new test items and alternate test forms, and comparability of remotely-proctored professional examinations to traditionally-proctored (in person) examinations. This research involves analysis of large data archives with several methods including item response theory, as well as the development of new statistical indices and evaluation of their effectiveness relative to existing indices, using Monte Carlo simulation.
Setting Cutoff Scores
While employee selection tests are often used to rank-order job candidates, in settings like occupational licensing and certification test-takers must exceed a minimum proficiency level, as indicated by a cutoff score, to pass the test. We are currently conducting research to integrate item response theory models with the human judgment involved in setting cutoff scores with the Angoff method, in attempt to overcome some limitations in human judgment by aligning judgments with previously-estimated item characteristics. This research involves a combination of Monte Carlo simulation and analysis of data archives.