The relationship between item parameters and item fit statistics for the three-parameter logistic model

• Main Authors: Yi-chen Ho, 何翊禎
• Other Authors: Bor-Yaun Twu
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/61147358936002645544

碩士 === 國立臺南大學 === 測驗統計研究所碩士班 === 97 === The purpose of this study was to examine the relationship between the item parameters andthe item fit statistics. The performance of three item fit statistics ( G−χ2 , χ 2* and S−χ 2 ) in two sample sizes (500 and 1000 examinees) and two fitting conditions (all items generated by 3PL model vs. all but 20% of items generated by 3PL model) were examined. The results showed that both test length and sample size affected the performance of item fit statistics whether the data was generated all by using 3PL model or not. With regard to the relationship between item fit statistics and item parameters, there were not significantly relationship between the item fit statistics and item discrimination parameter in all conditions. The relationship patterns between the item fit statistics and the difficulty parameters were not consistent across different combination of the conditions. Only the relationship pattern between the item fit statistics and the guessing parameters were consistent among different conditions. The higher the guessing parameters were, the higher the percentage of item misfitting were.