Abstract:
|
Computerized adaptive tests are becoming increasingly popular due to their increasing flexibility and an advancement of modern computer technology. Person-fit assessment refers to the determination of whether the response pattern of an examinee is aberrant (or an outlier) possibly due to fraudulent behavior. Person-fit assessment for computerized adaptive tests is an under-developed research area. This paper suggests several person-fit statistics based on tests for a change point (e.g., Andrews, 1993; Chen & Gupta, 2012; Csorgo & Horvath, 1997; Hawkins, Qiu, & Kang, 2003). The Type I error rate and power of the statistics are examined in detailed simulation studies. The suggested statistics are also compared with three existing person-fit statistics (Armstrong & Shi, 2009; van Krimpen-Stoop & Meijer, 2000, 2001) using receiver operating characteristics curves. A real data demonstration follows. The suggested person-fit statistics appear promising for assessment of person fit for computerized adaptive tests.
|