Abstract
This article is concerned with determining the bias in Lord's (1974) procedure for estimating the ability parameter in item response theory model. Lord's procedure is one of the methods of dealing with omitted responses for dichotomously scored multiple choice test data. This study highlights the bias inherent in Lord's procedure and how to reduce it. Theoretical justification and a simulation study are presented for the estimates obtained by the bias reduction method of this study. The results of the simulation study show that the bias reduction method has a higher efficiency than Lord's procedure in mean squared error. This study also takes notice of the appropriateness of Lord's formula of asymptotic variance of estimates of ability parameter.
