Open AccessEVALUATING SOURCES OF COLLATERAL INFORMATION ON SMALL‐SAMPLE EQUATING
Author(s) -
Kim Sooyeon,
Livingston Samuel A.,
Lewis Charles
Publication year - 2009
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/j.2333-8504.2009.tb02171.x
Subject(s) - equating , resampling , statistics , sample size determination , sample (material) , mathematics , collateral , raw score , bayes' theorem , test score , econometrics , bayesian probability , raw data , standardized test , economics , chemistry , finance , chromatography , rasch model
The purpose of this study was to examine the effectiveness of using collateral information to improve the accuracy of small sample equating. Collateral information from the equating of other tests was incorporated into an empirical Bayes (EB) estimate of the reference‐form score corresponding to each new‐form raw score. The evaluation consisted of resampling procedures using data from large‐group equatings of 10 different test forms, representing 9 different tests. Each large‐group equating provided the data for repeated small‐sample equatings based on resampling. The small‐sample equatings were done by 5 methods, including 2 EB methods that used the equating of the other 9 test forms as collateral information. The small‐sample equating methods were evaluated for agreement with the large‐group equating results. The new‐form sample size in the small‐sample equatings was systematically varied (10, 25, 50, 100, 200) as was the sample size in the equatings used as collateral information (100, 200, all available examinees). The results indicated that the use of collateral information tended to improve the accuracy of the equating when the new‐form sample size was 25 or fewer.