CSRDA Discussion Paper Series

No. 38 An Ordered Logistic Regression Model for Educational Stratification: Applications to Trend and Comparative Analyses
Sho Fujihara
Sho FujiharaUniversity of Tokyo
Inequality of educational opportunityordered logistic regressioneducational expansion
Goal 1: No PovertyGoal 4: Quality EducationGoal 10: Reduced Inequalities
(1) National Survey of Social Stratification and Social Mobility, 2015 (2015 SSM Survey Management Committee). (2) International Social Survey Programme: Social Inequality III - ISSP 1999 (ISSP Research Group).

This paper describes an ordered logistic regression model for educational stratification. The ordered logistic regression model with proportional constraints provides estimates of the basic patterns associating socio-economic background with educational attainment. Moreover, the parameters for changes in the strength of the associations are very easy to interpret. Using data from the 1955–2015 Social Stratification and Social Mobility surveys conducted in Japan, I applied the model to analyze trends in the inequality of educational opportunity. The estimated parameters depict similar trends obtained from the variances and means of the coefficients from the ordinal ordered logistic regression. This study also uses data from the International Social Survey Programme conducted in 1999 to conduct a comparative analysis of 23 societies. The results show similar but not identical cross-societal differences in the inequality of educational opportunity to those found by Hout (2006).