Jos Dessens

Estimating the Concomitant-Variable Latent-Class Model With the EM Algorithm:

Latent class analysis assumes the existence of a categorical latent variable that explains the relations between a set of categorical manifest variables. Simultaneous latent class analysis deals with sets of multiway contingency tables simultaneously. In this way an explanatory categorical grouping variable is related to latent class results. In this article we discuss a tool called the concomitant-variable latent-class model, which generalizes this work to continuous explanatory variables. An EM estimation procedure to estimate the model is worked out in detail, and the model is applied to an example on juvenile delinquency.

Patterns and trends in occupational attainment of first jobs in the Netherlands, 1930-1995: ordinary least squares regression versus conditional multinomial logistic regression

Summary. This paper brings together the virtues of linear regression models for status attainment models formulated by second-generation social mobility researchers and the strengths of log-linear models formulated by third-generation researchers, into fourth-generation social mobility models, by using conditional multinomial logistic regression (CMLR). These CMLR models are capable of capturing the discrete and multidimensional nature of social mobility patterns (a characteristic of third-generation output) while reducing the number of parameters leading to parsimonious models (a characteristic of second-generation output). Using data from eight pooled surveys in the Netherlands, an extended Blau–Duncan status attainment model is formulated and analysed.The corresponding CMLR model is formulated incorporating general and specific inheritance effects.The final CMLR model gives a relatively parsimonious description of Dutch mobility patterns, similar to the extended Blau–Duncan model, at the same time offering the possibility of including specific effects where necessary. Effects of gender and education appear to be too complex to be captured by a single parameter.

Income advantages of communist party members before and During the transformation process

ABSTRACT In this study contradictory hypotheses are tested about the changing income advantages of Communist Party (CP) members derived from the Elite Circulation Thesis and the Elite Reproduction Thesis, using cross-sectional datasets from before and during the transformation process. CP members are matched with non-CP members on several important income determinants such as human capital, occupational class, market capital, age, gender, and marital status. Independent-samples t-tests, on differences in mean personal (ln)income reveal that CP members earn more than non-CP members do before and during the transformation process. An ANOVA shows that the income advantages of CP members are most persistent in the Czech Republic and Russia while they get smaller in Slovakia and Hungary. Comparing the four countries suggests that the remaining income advantages of CP members may partly be explained by transformation specific differences between countries.

Using covariates in loglinear models with sampling zeros; a cautionary note

Abstract When analyzing loglinear models, most computer packages will first construct the contingency table and subsequently fit the required loglinear model. Some computer packages allow the incorporation of covariates, such as age and/or income, into the loglinear model. In the SPSS modules LOGLINEAR and GENLOG this option is implemented by calculating for each pattern of the classifying variables the mean value(s) of the covariate(s), while this covariate vector is subsequently added to the design matrix describing the loglinear model. In this article it is shown that the use of covariates in the case of contingency tables containing sampling zeros will lead to incorrect results for the SPSS modules LOGLINEAR and GENLOG. Parameter estimates, deviances and the number of degrees of freedom may be highly incorrect. This is illustrated in the case of the loglinear uniform association model, using an example from a handbook on categorical data analysis. A second, more complex example is discussed where, as a result of this incorrect handling of covariates, wrong results have been published in the literature. It is concluded that (conditional) multinomial logit models should be used. While SPSS currently does not include such models, it is demonstrated how to obtain correct results for the case that covariates are functions of the classifying variables.

Marshall and Gorman's Markov chain model for occupational prestige

Abstract This report scrutinizes the Markov chain model for occupational prestige proposed by K. P. Marshall and B. L. Gorman (1975, Social Science Research 4, 41–64). It is shown that the results of their application are highly unstable.