Adaptive estimation when the unidimensionality assumption

This study examined some effects of using a unidi-mensional IRT model when the assumption of unidi-mensionality was violated. Adaptive and nonadaptive tests were formed from two-dimensional item sets. The tests were administered to simulated examinee populations with different correlations of the two un-derlying abilities. Scores from the adaptive tests tended to be related to one or the other ability rather than to a composite. Similar but less disparate results were obtained with IRT scoring of nonadaptive tests, whereas the conventional standardized number-correct score was equally related to both abilities. Differences in item selection from the adaptive administration and in item parameter estimation were also examined and related to differences in ability estimation. Index terms: ability estimation, adaptive testing, item pa-rameter estimation, item response theory, multidimen-sionality. In computerized adaptive testing (caT), the ability (0) of each examinee is estimated from individually tailored tests. Currently, most CAT procedures are based on unidimensional item response theory (IRT) models which incorporate a single 0 to account for test performance in a particular domain. In reality, many tests measure multidimensional domains. It is not clear what the effects on 0 estimation will be if such tests are administered adaptively. The unidimensionality requirement would be less critical if multidimensional IRT models could be used in practical testing. Much work has been done to develop multidimensional IRT models (e.g., Bloxo