Testing the tightening bond hypothesis: An application of new models for analysing trends in multiway tables

Abstract

The tightening bond hypothesis postulates that in modern economies individuals are increasingly allocated to occupations on meritocratic principles. This hypothesis is tested on a set of nine contingency tables of class of job by qualification level for new male recruits to jobs, obtained from the annual Labour Force Survey in Great Britain for the period 1983 1991. A number of statistical models for examining the changes in the association of a pair of variables over the levels of a third variable are described and evaluated. These include the log-multiplicative layer effect model, multiple-group association RC(M) models and ‘levels’ models. All the models fitted indicate a slight loosening of the bond between qualifications and class of job so that the tightening bond hypothesis is rejected with confidence. The effects of the complex sample design used in the surveys are also evaluated; design effects for the parameters measuring changes in the association between class of job and qualification over years are negligible.