Research Article
The Effects of Sample Size and Logistic Models on Item Parameter Estimation
@INPROCEEDINGS{10.4108/eai.28-9-2019.2291082, author={Suwarto Suwarto and Eko Putro Widoyoko and Budi Setiawan}, title={The Effects of Sample Size and Logistic Models on Item Parameter Estimation}, proceedings={Proceedings of the 2nd International Conference on Education, ICE 2019, 27-28 September 2019, Universitas Muhammadiyah Purworejo, Indonesia}, publisher={EAI}, proceedings_a={ICE}, year={2020}, month={2}, keywords={sample size logistic models item parameters}, doi={10.4108/eai.28-9-2019.2291082} }- Suwarto Suwarto
Eko Putro Widoyoko
Budi Setiawan
Year: 2020
The Effects of Sample Size and Logistic Models on Item Parameter Estimation
ICE
EAI
DOI: 10.4108/eai.28-9-2019.2291082
Abstract
This research aims to ascertain: (1) the effects of sample size (N) to determine item parameters; (2) the effects of logistic models (1PL, 2PL, and 3PL) on item parameter estimation. The research was conducted with a computer simulation, which was performed in the following steps: by using sample sizes of N=200, N=500, N=1000, each with ten replications. In order to generate the dichotomous data, the DGEN program was used and then these data were performed in the BILOG 3 program. From the DGEN program the “true” item parameters could be ascertained, and from the BILOG 3 program, the estimated item parameters could be ascertained. The criteria used to determine the stability of item parameters were: (1) Seeking the correlation between “true” parameter and the average item parameters to find the highest correlation; and (2) Seeking the MSD average with the smallest variance in each item parameter, which was the best one. The research results show that (1) the order of the effects of sample size from the best to the worst in estimating item parameters were: N=1000, N=500; N=200, (2) the order of logistic models in item parameter estimation were as follows: (a) the order of the logistic models to estimate the difficulty parameter were: IPL (the best), 2PL (medium), 3PL (low); (b) the order of the logistic models to estimate the discrimination parameter were 2PL and then 3PL; (c) the logistic model to estimate the “guessing parameter” was 3PL.