Kelli Samonte, a third year ERM Ph.D. student, recently received the Outstanding Poster Award at the 2014 annual meeting of the Psychometric Society, held in Madison, Wisconsin. Her poster, titled “Using Factor Mixture Models to Model Dimensionality Differences Across Proficiency Levels”, described a latent variable modeling approach to operationalizing and measuring growth in proficiency. Often gains in proficiency are modeled as increases in attributes, whereby the attributes are treated as consistent across all levels of proficiency. Some researchers, however, suggest that information is reorganized (e.g., becomes automatic) and constructs change as individuals increase in proficiency. Kelli’s study examined whether confirmatory mixture multidimensional item response theory (MIRT) models could be used to classify individuals into the correct proficiency level. To this end, Kelli conducted a simulation study in which the relationship between the factors and the group size was varied to examine how the models would perform in different situations. The preliminary results suggest that the models perform best (i.e., correctly classifying individuals and adequately recovering parameters) in conditions with low correlations between factors.
Congratulations Kelli on this wonderful accomplishment!!