Does Deferring Items Change Their Psychometric Characteristics?

Abstract

In this study, we investigated possible context effects when students chose to defer items and answer those items later during a computerized test. In 4 primary school reading tests, 126 items were studied. Logistic regression analyses identified 4 items across 4 grade levels as statistically significant. However, follow-up analyses indicated that the effect sizes for these items were in the small category, and no consistent characteristic was identified to support a causative factor for these items. In addition, the few items flagged at each grade level and their small effect sizes would argue against any effect on students' total scores. Item discrimination values calculated including and not including students who deferred each item showed no significant differences. These results not only extended the study of item context to the primary school grades and computerized tests but also supported deferring as a desirable option for computerized tests. Further research should study longer tests and more content areas.