top of page

CHAPTER 4- ASSUMPTIONS, TESTS AND COMPARATIVE CRITERIA IN QUALITATIVE PREFERENCE MODELS                                                                                                                                                                                   120                                                                                                                                

Cite this chapter

İşçi Güneri, Ö., Durmuş, B., İncekırık, A. (2023). Chapter 4 Assumptions, Tests and Comparative Criteria in Qualitative Preference Models. In M. E. Camargo  (Ed.), Academic Research & Reviews in Social, Human and Administrative Sciences -II- (pp. 120-145). Ankara, Türkiye: Global Academy Publishing House.

Showing a limited preview of this publication:
In cases where the dependent variable contains two or more qualitative preferences, it is not appropriate to use linear regression models when assumptions such as normality, linearity and constant variance are not met. Instead, qualitative preference models are used. In qualitative preference models, the dependent variable (Y) is coded as “0” and “1” in binary preference models. A household's decision to buy a car can be given as a classic example of this situation. When the dependent variable is coded to take the value of “0” for “not to buy” and “1” for “to buy”, binary preference models are used (Davidson and MacKinnon, 1999). Binary preference models are analyzed with linear probability model (DOM), binary logit model, binary probit models, and tobit models. Due to some drawbacks of the DOM, binary logit and probit models are widely used. Tobit model is applied for censored data. Multiple choice models can be viewed as extensions of binary logit and binary probit models. In these models, the dependent variable has at least three or more categories and is either nominal or ordinal (ordered). Therefore, multiple preference models are generally based on ordered and non-ordered preferences.

bottom of page