Exploring the Accuracy of MIRT Scale Linking Procedures for Mixed-format Tests

TL;DR

This study evaluates the accuracy of Stocking-Lord scale linking procedures for mixed-format tests using UIRT and MIRT models under different anchor conditions and format effects, providing practical recommendations.

Contribution

It offers a comparative analysis of scale linking accuracy for UIRT and MIRT models with various anchor scenarios and format effects in mixed-format tests.

Findings

Stocking-Lord procedures vary in accuracy depending on anchor conditions.

MIRT models perform better under certain format effect conditions.

Recommendations for choosing linking procedures based on test design.

Abstract

This study investigates the accuracy of Stocking-Lord scale linking procedures for UIRT and MIRT models with common-item nonequivalent-group design for mixed-format tests under two anchor conditions: MC-Only and MC-CR across three different levels of format effects (FEs). Results provide recommendations on the appropriateness of UIRT and two MIRT models when FEs present under each anchor scenario.