SOEPpapers on Multidisciplinary Panel Data Research 372
Rating variables indicate the extent to which a quality is present, or absent, in a unit of observation. In this paper, we discuss a class of non-linear regression models for rating dependent variables and their estimation by parametric and semiparametric methods. An application to life satisfaction illustrates the main differences between the Rating Scale Model and ordinary least squares.
rating variables non-linear least squares quasi-maximum likelihood semiparametric least squares subjective well-being