The Estimation Stability of Logistics Model Parameters Under Different Scenarios

Wulansari, Andhita Dessy (2016) The Estimation Stability of Logistics Model Parameters Under Different Scenarios. [Seminar and Workshop] (Unpublished)

Preview

Text Artikel Seminar Thailand The Estimation Stability of Logistics Model Parameters Under Different Scenarios.pdf Download (398kB) | Preview

Abstract

This research identifies how the contribution of IRT model, sample size N and number of item/test length n to the stability estimation of item parameter and examinee parameter ϴ. The data used in this study is generated by WINGEN, and the parameter estimation uses BILOG-MG. The trend pattern is observed based on the coefficiency correlation between the actual parameter and the estimated parameter. The findings of this research are: (1) the greater the N, the stabler the item parameter; (2) the more n, the stabler the estimation of examinee parameter Ɵ; (3) one-parameter logistic model is the stablest model in estimating item b parameter, two-parameter logistic model is the stablest model in estimating parameter item a and three-parameter logistic model is the stablest model in estimating item parameter c, whereas two-parameter logistic model is the stablest model in estimating examinee parameter (4) the stability of the item parameter is more affected by the sample size N than the number of item/test length n, while the stability of examinee parameter Ɵ is more influenced by the number of item/test length n than sample size N; (5) In one-parameter logistic model, two-parameter logistic model, and three-parameter logistic model, the stability of item parameter is more influenced by sample size N than the number of item/test length n while the stability of examinee parameter Ɵ is more influenced by the number of item/test length n than sample size N, where one-parameter logistic model is the best model in estimating item parameter and two-parameter logistic model is the best model in estimating examinee parameter, and three-parameter logistic model is the most unstable model in estimating both item and examinee parameter.

Actions (login required)

View Item