The effect of item revisions on classical item statistics and preequating outcomes
(English)Manuscript (preprint) (Other academic)
The purpose of this study was to examine the effect of textual item revisions on classical item statistics and the adequacy of preequating outcomes. Three forms of the SweSAT data sufficiency subtest, comprising a total of 66 items, were examined. The items were categorized after type of revision, and the differences in p-values and point-biserial correlations between regular test and pretest were averaged in each category. These averages were subjected to a t-test, and it was found that revisions have a significant effect on both difficulty and discrimination indices. Also, while the preequating method used in this study produced adequate results, the revisions seem to increase the amount of error in preequating outcomes.
achievement testing, textual revisions, item statistics, item pretesting, preequating
IdentifiersURN: urn:nbn:se:umu:diva-25431OAI: oai:DiVA.org:umu-25431DiVA: diva2:231764