| A study was conducted which compared the process of item selection and scale evaluation of Classical Test Theory (CTT) and Item Response Theory (IRT) for a brand loyalty scale. Difficulty and discrimination item parameters were estimated using CTT and IRT graded response techniques. Classical Test Theory and Item Response Theory scales of 3-, 9-, 15-, 21-, and 30-items were compared in terms of reliability (Cronbach's alpha), variance, information, and the property of invariance. Results of these tests revealed little difference between CTT and IRT in terms of the scale evaluation criteria and the property of invariance.; A second phase of the study was conducted to validate the CTT and IRT revised brand loyalty scales. Measures of consumers' purchasing behaviors and product involvement were used as validation measures of the brand loyalty scales. Subjects completed the CTT and IRT brand loyalty scales and Zaichowsky's Personal Involvement Inventory (1985), which is a measure of consumers' involvement with a product category. A week after completing the scales, subjects completed a purchase diary which asked subjects to report the brands and number of times each brand was purchased for three product categories during the previous week. Neither brand loyalty scale was validated with the measures of the present study. |
