An empirical study of the applicability of the rasch latent trait model to mental ability test: A case of BILOG analysis

Rasch (1960) was the first psychometrician who conceptualized the plausibilities for objectivity that resided in the logistic response model with one item parameter and one person parameter in the realm of measurement methodology, and made it applied to the analysis of mental test data. Lord (1980) preferred to label it as item response theory. Distinctive features of the Rasch Model covered (i) item parameter estimates were independent of the group of examinees used from the population of examinees for wom the test was designed (ii) examinee ability estimates were independent of the particular choice of test items used from the population of items which were calibrated and (iii) precision of ability estimates could be found. These features could overcome many well-documented shortcomings and drawbacks of the ways in which educational and psychological tests were traditionally constructed, evaluated and tested (Hambelton & van der Linden, 1982). Gallini (1983) who employed unconditional maximum likelihood procedure, UCON of the computer programme BICAL (1969) conducted a Rasch analysis and he was able to show a good fit of the Rasch Model to his mental test data. It was the purpose of the present study to investigate the extent to which Rasch Model could be applied appropriately to mental test data. BILOG 3 computer programme (An Item Analysis and Test Scoring with Binary LOGgistic Models Computer Programme, Mislevy & Bock, 1990) was employed to undergo relevant computations including: (i) marginal maximum likelihood (MML) estimation of item parameters (ii) test of fit for individual items (iii) standard errors for all item parameter estimates and scale score of respondents (iv) maximum likelihood (ML) estimation of scale scores, and (v) test and item information analyses. Result indicated that nearly all items had perfect fit to the model, whereas only a few items could be identified as poorer fit items. Conclusion was that there existed substantial evidence that the applicability of the Rasch Model in mental ability measurement be justified, and the utilization of BILOG 3 in Rasch analysis could further be substantiated. Besides, in view of the strength of the Rasch Model over the classical test theory in educational assessment and the easy-portability and user-friendliness of BILOG 3, it was advocated that in order to increase the teaching effectiveness of ordinary classroom teachers and pre-service teacher education students, they should be adequately trained capable to perform Rasch analysis in their routine assessment duties. Copyright © 2000 The Hong Kong Institute of Education.