The standard Rasch facets model was developed to account for facets data, such as student essays graded by raters. The standard facets model accounts for only rater severity. In practice, raters may exhibit different levels of tendency of using middle or extreme scores in their ratings, referring to centrality/extremity response style. To achieve better measurement quality in facets data, it is desirable to consider rater’s severity and centrality/extremity jointly. A new facets model is thus developed to represent rater’s centrality/extremity by adding a weight parameter on thresholds for each rater to the standard facets model. Simulation results show that the parameters of the new facets model could be well-recovered; failing to consider centrality/extremity would deflate the ability difference between ratees. Two empirical examples are provided for illustration, along with the implication of the new model. Copyright © 2017 International Meeting of the Psychometric Society.
|Publication status||Published - Jul 2017|