Computerized adaptive attitude testing using the partial credit model
Sun-Geun Baek, Mark R. Wilson · 1993
This study was about computerized adaptive testing (CAT) using the partial credit model (PCM) for attitude measurement. CAT is a procedure for administering tests which are tailored individually for each examinee. The appeal of CAT as an alternative to traditional paper-and-pencil measurement instruments is that a person's trait level may be measured quite efficiently and accurately with relatively few items. For this study, data collected from Korean graduate students at U.C. Berkeley using the culture shock questionnaire, as an attitude measurement scale, and some simulated data were analyzed. Results showed that (a) there was equivalence between a computerized version and a paper-and-pencil version of an attitude scale--the culture shock questionnaire, (b) most students enjoyed taking the computerized version of the questionnaire and preferred it to the paper-and-pencil test version, (c) CAT procedures for attitude testing using the PCM with real items and real examinees performed very well, (d) there was no difference between the CAT procedures estimating the examinee's starting attitude level and the CAT procedures without such procedures, and (e) there was almost no difference in the person parameter estimation by the CAT using the PCM according to the calibration sample size for item step difficulties. In addition, the unbalanced content problem for the content validity of the CAT, and the person fit statistics within CAT procedure to diagnose individual differences in response style, were investigated. Testing is a very important instrument for optimum decision-making in education as well as in society. As computerized adaptive attitude testing expands, it should allow the development of a new paradigm for efficient and useful attitude measurement.