ABSTRACT
The previous chapters have demonstrated how the basic features of Rasch model l ing can be used to deal with simple right-wrong, or dichotomous, data. A n extension of these principles allows us to extend the idea of Rasch modell ing to polytomous (erroneously called polychotomous) data. One form of this approach to collecting data that has been around for a long time is the principle of the Likert scale (Likert, 1932), which usually is used to collect attitude data. Likert scales share a number of common features, regardless of which attitudes they assess, and with possible responses usually expressed in a format such as: SD (strongly disagree), D (disagree), N (neutral), A (agree), and SA (strongly agree). Similarly, each Likert scale item is provided with a stem (or statement of attitude) and the respondent is required to mark a response on the disagree-agree cont inuum, indicating the extent to which the statement in the stem is endorsed. A n uneven number of response options might provide a "neutral" response at the mid-point while an even number of response options might force the respondent into choosing either a positive or negative response.
