BAYESIAN IDENTIFICATION OF OUTLIERS IN COMPUTERIZED ADAPTIVE TESTS

We consider the problem of identifying examinees with aberrant respons e patterns in a computerized adaptive test (CAT), The vector y of resp onses of an examinee from a CAT is a multivariate response vector. Mul tivariate observations may be outlying in many different directions, a nd we characterize specific directions as corresponding to outliers wi th different interpretations. We develop a class of outlier statistics to identify different types of outliers based on a control chart-type methodology. The outlier methodology is adaptable to general longitud inal discretes data structures. We consider several procedures to judg e how extreme a particular outlier is. Data from a nationally administ ered CAT examination motivates our development and is used to illustra te the results.