NotesFAQContact Us
Search Tips
ERIC Number: ED479923
Record Type: Non-Journal
Publication Date: 2003
Pages: 16
Abstractor: N/A
Reference Count: N/A
Exploring Boundary Conditions on the Invariance Property of Item Parameter Estimates of Item Response Models.
Sadesky, Greg S.
In this study, the robustness of item parameter estimates with respect to the underlying distribution of abilities was explored. Using simulated datasets, item parameters were derived from a large sample representing the population versus samples representing a subset of ability and subsequently compared for possible mismatch. Comparisons were made under these conditions: (1) increasing differences between ability of examinees in the pilot sample and those of the population; and (2) the size of the pilot sample. All datasets were generated using DIFSUM, a computer program that permits the creation of dichotomous test data according to prespecified characteristics. In general, the study shows that item parameter estimates derived from a sample contain more errors when ability differences exist between the sample and its population. Fewer errors result when the size of the sample is increased. This finding highlights some limitations of the item invariance property claimed by item response theorists. Some of the limitations of the study are also discussed. (SLD)
Publication Type: Reports - Research
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A