New Models for Comparing Mobility Tables: A Log-Multiplicative Approach

This paper proposes new models for comparing mobility tables that constrain cross-table variation in the origin-destination association as the log-multiplicative product of a common association pattern and a table-specific parameter. Like the method proposed by Yamaguchi (1987, ASR), the new models provide one-parameter tests and thus facilitate analyses of the difference in "vertical mobility" between two mobility tables. A distinct advantage of the new models over Yamaguchi's models is that the new models are powerful in their specification of the origin and destination association. Various models (including Hauser's levels model and Goodman's RC model) for the association patterns between origin and destination are allowed and estimated simultaneously with the table-specific comparison parameters. All that is required is a common pattern for the origin-destination association shared by the tables compared. Properties of the new models are demonstrated using three data sets previously analyzed in comparative mobility research: Yamaguchi's data (1987) for the United States, Great Britain, and Japan; Erikson, Goldthorpe, and Portocarero's (1982) data for England, France, and Sweden; and Grusky and Hauser's (1984) data for sixteen countries.