Life-Is-like-a-Random-Walk Model of Class Identification

Wednesday, 18 July 2018: 09:00
Oral Presentation
Atsushi ISHIDA, Osaka University of Economics, Japan
Social psychological aspect of social inequality and social stratification, such as people's cognition, attitudes and emotions in unequal or stratified society, can be seen as antecedent conditions of people's rational choices or actions which would be aggregated and lead to macro (un)change in the society. Thus, it is as important to rational choice approaches to social inequality as other aspect.

Among social psychological features, I will focus on class identification in this paper. Class identification is the extent to which people identifies themselves as members of a certain social class or stratum, and it has been one of main subjects in the social psychological study of social stratification. In this paper, I will introduce a new analytical framework for class identification by applying a mix method of simple mathematical modeling and Bayesian statistical modeling instead of conventional frequentist statistical analysis.

First, I will construct a simple mathematical model which can explain one of the major tendencies of class identification, that is middle concentration tendency where majority of people tend to regard themselves as middle. This is the life-is-like-a-random-walk model where it is assumed that succession of same Bernoulli m-trials with success probability p determines one's subjective class identification.

Second, I will estimate parameters of the model from empirical data by applying a Bayesian statistical model. The Bayesian modeling enables us to construct more flexible model which directly reflects the mathematical model and is able to explain generative mechanism of observed distribution. The distribution of latent success probability p and number of trial m are estimated by MCMC estimation and differences in distributions of p and m among different social economic categories are analyzed by hierarchical models. Japanese cross-sectional survey data from SSM and SSP surveys will be analyzed by the Bayesian model, and interpretation of results will be discussed.