Article

Creating Equated Test Items: Strategies for Controlled Difficulty Variation

---

Abstract

To create new items with controlled item difficulty, two strategies were tested: an item model approach and the application of an ordinal construct theory. An item model approach was applied to Number Series (NS) items, and an ordinal construct theory was applied to Paper Fold and Cut (PFC) items. A matrix design was used for NS (four forms, n = 3 315 test takers), and a common item design for PFC (three forms, n = 2 003 test takers). Measurement properties were analyzed according to the Rasch model, and parallelism and equating between test forms were evaluated using the Leunbach model. After removing some less functional items, the tests showed acceptable fit to the Rasch model within each form. In the NS tests, easier items were better aligned with the model items and in the PFC tests, harder items were better predicted by the ordinal construct theory (PFC). All NS and PFC tests met criteria for parallelism, allowing equating across forms. The findings provide initial support for the item design strategies used, yet, there are still unexplained or unidentified factors that need to be incorporated during the design phase. However, it offers valuable guidance for developing item pools with controlled difficulty variation. 

Keywords: Cognition, Item difficulty, Item generation, Logical test, Parallelism, Rasch Measurement Theory, Test equating

How to Cite:

Melin, J., (2026) “Creating Equated Test Items: Strategies for Controlled Difficulty Variation”, Practical Assessment, Research, and Evaluation 31(1): 18. doi: https://doi.org/10.7275/pare.3568