How to Ransack Social Mobility Tables and Other Kinds of Cross-Classification Tables

Abstract

This article presents (1) methods for examining every part of a social mobility table in search of the various possible relationships between an individual's status category and his father's status category that are congruent with the data in the table; (2) methods for comparing two (or more) social mobility tables by examining the corresponding parts of each of the tables in search of the differences between the corresponding relationships in the tables; and (3) conceptual tools that can be used by the research worker to assist and stimulate him to conceive of a wide variety of possible relationships between an individual's status category and his father's status category, which could then be checked with the data. These tools and methods can also be applied to other kinds of cross-classification tables to assist in the conception and analysis of the various possible relationships between the column classification and the row classification of each table, and in studying the differences between the corresponding relationships in two (or more) such tables. The techniques presented here for studying the possible relationships between two qualitative variables (the column classification and the row classification) were obtained, in part, by adapting, for qualitative variables, some of the multiple-comparison ideas developed earlier in the analysis of variance context. To illustrate the application of the conceptual tools and methods proposed here, we reanalyze data on intergeneration social mobility in Britain and in Denmark. We find, for example, with the introduction of a new concept of "intrinsic status inheritance and status disinheritance," that there is a statistically significant amount of intrinsic status disinheritance in the middle status category in Britain and in Denmark, as well as statistically significant amounts of intrinsic status inheritance in the upper and lower status categories. (This result concerning status disinheritance is an apparent contradiction of all earlier findings based on the more usual methods of analysis.) Furthermore, there is more intrinsic status inheritance in the upper status category than in the lower, and the difference is statistically significant in both countries. Also, in contrast to all earlier findings, we find that the relationship between a subject's status and his father's status in the British study differs from the corresponding relationship in the Danish study in tatistically significant ways which we describe in some detail.