A diagnostic classification model of TIMSS 2023 grade four mathematics

Studies have analyzed large-scale assessments using diagnostic classification models; however, these studies tend to analyze only a small subset of the data; rarely report the distinctiveness of diagnostic scores; and inaccurately estimate standard errors. This study addresses these oversights by modelling complete data for TIMSS 2023 grade four mathematics to produce plausible values for analysis. A version of the TIMSS assessment framework was used as an attribute model, meaning that seven content and three cognitive attributes were modelled. The diagnostic classification model shows reasonable global- and person-fit, but item-fit has some issues. The diagnostic scores have good accuracy and consistency. Achievement gaps for gender, socioeconomic status, liking mathematics, and calculator use are presented to illustrate the results. Overall, the study demonstrates both weaknesses and benefits that come from using diagnostic classification models to analyze assessments which are not designed for this purpose.