Search
Program Calendar
Browse By Day
Browse By Time
Browse By Person
Browse By Session Type
Personal Schedule
Sign In
Deadlines
Policies
Program Updates
Accessible Presentation
FAQs
Search Tips
Annual Meeting App
Analysis of intergenerational social mobility with respect to educational attainment and class position has been dominated for several decades by log-linear models, most prominently the so-called Unidiff a.k.a. Log Mutiplicative Layer Effect model. In this paper we argue that these models are not ideal when the goal is to evaluate how the general relevance of social origin changed in a society over time or differs between countries, because these models neglect important aspects of the marginal distribution. We propose an alternative methodology based on the Mutual Information Index (M-Index), that solves this problem and, at the same time, equips researchers with a much more flexible framework for modelling effects of social origin. Our arguments are illustrated by two brief empirical applications.