John R. Donoghue is a principal research scientist in the Research & Development division at ETS. He received an M.A. in general/experimental psychology from California State University, Northridge in 1987 and a Ph.D. in quantitative psychology from the University of Southern California in 1990.
His research spans a range of issues in psychometrics, particularly investigating the properties of item response theory models, measures of item parameter drift and differential item functioning, and test models that assume that item data provide only ordinal information about the construct of interest. His second area of research interest is the connection to the emerging field for data science and combining these techniques with more traditional psychometric methods. A third area of research is the investigation of constructed-response scoring, encompassing the design and monitoring of scoring and statistical models to detect and adjust for rater effects such as differences in rater leniency/severity. A fourth major area of research has been cluster analysis. His recent work has examined the properties and estimation of finite mixtures of normal distributions.
His work over the past several years has also examined the production of high-quality software implementation of complex statistical models. A major thrust of this work is employing and disseminating industry best practices in the development of complex psychometric/statistical software systems. He holds several professional certifications in data science techniques, Java, and C# programming and has been involved in numerous software development projects. These projects encompass a range of activities from modernizing complex legacy statistical applications to developing high quality, new systems that estimate novel psychometric and statistical models.