Bayesian Multidimensional Models Fit to Comparative Judgment Data on Art Educator Perceptions of Art Ability

• In Event: AERA Poster Session Wednesday 11:45am
  In Poster Session: Poster Session I -- AERA Division D Section 1: Measurement, Psychometrics, and Assessment

Abstract

I fit Bayesian, multidimensional models (Martin et al., 2011; Yu & Quinn, 2022) to data from an adaptive comparative judgement (ACJ) assessment (Pollitt, 2012) of drawing ability. Art educators evaluated student drawings for the Clark Drawing Abilities Test (CDAT; Clark, 1989). I hypothesized that drawing ability may be comprised of two independent constructs: (1) creative elaboration or fluency and (2) skill and knowledge with conventions of pictorial representation. Two-dimensional models fitted to the data supported this assumption, predicting 68-percent of raters’ choices correctly in simulations while controlling for diverse rater perspectives. With better predictiveness than unidimensional models, the two-dimensional models also were shown to ameliorate a persistent problem with subjective performance assessment which often exclude raters with diverse perspectives.

Author

• Steven Heil, Indiana University