Well-designed spatial assessments can incorporate multiple sources of complexity that reflect important aspects of spatial reasoning. When these aspects are systematically included in spatial reasoning items, researchers can use psychometric models to examine the impact of each aspect on item difficulty. These methods can then help the researchers to understand the nature and development of spatial reasoning and can also inform the development of new items to better reflect the construct. This study investigated sources of item difficulty for object assembly (OA), a format for the assessment of spatial reasoning, by specifying nine item characteristics that were predicted to contribute to item difficulty. We used data from two focal samples including high-ability students in grades 3 to 7 and undergraduate students who responded to 15 newly developed OA items. Results from the linear logistic test model (LLTM) indicated that eight of the nine identified item characteristics significantly contributed to item difficulty. This suggests that an LLTM approach is useful in examining the contributions of various aspects of spatial reasoning to item difficulty and informing item development for spatial reasoning assessments.
Keywords: explanatory response models; item response theory; linear logistic test models; spatial reasoning; visuospatial ability.