Item Characteristic Curve Parameters: Effects of Sample Size on Linear Equating.
Abstract
In order to establish large item pools, it is necessary to try out hundreds of items for possible inclusion. Because it is impractical to administer several hundred items to several thousand people, the invariance properties of Item Characteristic Curve (ICC) parameters are used to advantage to equate test items. A simulation study was conducted in order to investigate the effects of errors in estimated ICC parameters on equated parameters. Four samples of simulated examinees, ranging from 250 to 2,000 subjects, were administered a different test containing the 20 anchor items, plus 60 additional items on the same scale. The effectiveness of the equating procedures was studied through analysis of summed absolute deviations from the known item parameters of the estimated item parameters and the equated item parameters for the various combinations of 'anchor' item calibration sample size and equating same size. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1980
- Accession Number
- ADA082341
Entities
People
- Harald E. Jensen
- Malcolm James Ree
Organizations
- Air Force Research Laboratory