Search results

Data on housing careers in Cardiff are analysed within the statistical modelling framework developed in a companion paper. Several simplifying assumptions were necessary to overcome complex statistical and computational problems but the results obtained appear to be robust. However, these results include some features which were unexpected and difficult to interpret.

It is argued that the notion of housing careers has an important potential role in integrating the diverse theoretical perspectives which characterise research into residential mobility and tenure choice. By the synthesis of concepts from a range of theoretical and empirical traditions, a statistical modelling framework is developed that treats housing careers as trajectories from a dynamic optimisation process. This framework is operationalised in a companion paper that reports an empirical study of housing careers in Cardiff.

In this paper, aspects of entrepreneurship in Ireland—North and South—based upon a sample of 4300 male work histories are analysed. Complementary log-log models are employed to investigate the factors underlying the first business startup by an individual. A between-sector analysis of entrepreneurial behaviour is presented; the preceding employment experience of entrepreneurs is analysed; and their subsequent employment spell is examined. Results show that religious denomination is not associated with self-employment in any sectors, but that Anglicans are more likely to set up employer businesses outwith the construction industry. There are also spatial differences in entrepreneurial expression, with the highest levels of formation of employer business occurring throughout Northern Ireland and the Dublin region of the Republic.

It is explained that statistical problems over omitted variables and endogenous factors cast doubt upon the wisdom of attempting inference about many forms of human behaviour from cross-sectional data. Simulations are used to confirm that cross-sectional analyses can produce seriously misleading results. The problems are avoided by using longitudinal methods.

Male unemployment in Northern Ireland from 1926 to 1973 is examined by using aggregate time series and individual life-history data. County-specific and sector-specific unemployment rates are derived from various published statistics by using an entropy-maximising procedure. These aggregate data show a systematic relative improvement in male unemployment in the predominantly Protestant eastern counties and a corresponding deterioration in the more Catholic western counties. A model for investigating the factors affecting the duration of unemployment of individuals is presented. It is applied to the lifetime unemployment histories of a sample of men in Northern Ireland who were aged between eighteen and sixty-five during August 1973. With controls for other factors, it was found that the probability of reemployment was generally lower for Catholics than Methodists and members of the Church of Ireland, but it was not markedly lower than for Presbyterians. These differentials are also modified by several other interacting factors.

A two-state Markov model of individual voting transitions is developed which allows voters to differ with respect to their voting propensities and loyalties. Such a model allows the estimation of various process parameters without the specification errors common in other models. Additional parameters are introduced to allow for the effects of a third party, considerably extending the scope of application. The model is applied to the electoral voting histories for 1964, 1966, and 1970 of 671 individuals from sixty-seven English constituencies.

In this paper the authors address some of the inferential problems posed by longitudinal data on the discrete choice behaviour of a collection of individuals. In particular, an integrated framework is developed which enables heterogeneity and nonstationarity to be explicitly included in stochastic models of binary choice behaviour. The emphasis is upon minimum assumptions about the nature and determinants of heterogeneity and nonstationarity, as any uncontrolled variations may result in the identification of spurious adaptive behaviour. Moreover, all the models are readily calibrated and tested using widely available computer software. Some assessment is made of the flexibility of the modelling framework with respect to its potential for handling attrition in panel membership, more extensive choice sets, and exogeneous variables.