ABSTRACT
In Chapter 8, we discussed the details of quantification of multiplechoice data. The object was to determine category (option) weights so as to maximize the correlation ratio or Cronbach’s internal consistency reliability coefficient α. This optimization was carried out in terms of the entire set of items. In this context, the category weights of one item are likely to change if another item was omitted or a few other items are added to the data set. This may not always be what we want. Consider the following questions: 1. Can we not carry out quantification in such a way that all the
characteristics of one item are uniquely determined, independently of what other items are included in the data set?
