UNFOLDING EVENT HISTORIES - SCALING OF NONREPEATING EVENTS

The present paper discusses the use of Coombs' (1950, 1964) unfolding procedure applied to event history data. This procedure is appropriate when several stimuli are rated according to the individuals' preference, and yields scale values for stimuli and individuals, as well as a 'best' order of stimuli. We propose to consider the order of occurrence in time of non-repeatable events (such as first marriage, leaving home, and first job) as an indication of the subjects' preference for these stimuli. In that case unfolding may be applied to the analysis of event history-data.

The statistical principles of unfolding analysis in general are briefly discussed. Subsequently an illustration is given, using data concerning the life course of 216 Dutch women of birth cohort 1961, concerning five events: first steady relationship, leaving home, first child, first job, and first occurrence of living together. It is shown that these events can be ordered at an underlying one-dimensional continuum on which individuals can be scaled. Furthermore, using regression analysis it is shown that the location of the subjects at this continuum correlates significantly with various explanatory variables, showing how unfolding may be used in practical applications.