ABSTRACT
The multidisciplinary Zeitgeist of American social science has witnessed numerous efforts to integrate psychological and sociological conceptions of human life-span development. The conceptual sweep of these efforts is often impressive, but seldom do these incorporate a commitment to verify existing hypotheses regarding various aspects of constancy and change in the empirical record. Taking the investigation of human stability as theoretically and empirically problematic, this article proposes a framework for integrating what is known about patterns of stability over the life span. This framework focuses explicitly on the introduction of the concept of molar stability, the persistence of a behavior or behavioral orientation as expressed in age-homogenous rates of change over specified periods of time, as a means of organizing empirical information on human constancy and change. The approach taken to the estimation of molar stability solves three problems that have plagued past researches on the 136question of lifespan trajectories of stability. First, using a latent variable model, the approach insists on the unconfounding of measurement errors and true change. Second, this model can be used to assess differences in stability between occasions of measurement in longitudinal studies of the same individuals, or within age groups of reinterview studies of shorter duration in order to ascertain life-span trajectories of stability. The model is extremely useful in conjuction with a synthetic-cohort approach, as stability estimates can be generated across several groups of cohorts varying in age, with an eye toward estimating different trajectories of human stability within panels that have considerable heterogeneity in age. Third, this approach allows us a method by which one can not only compare estimates of stability across cohorts differing in age, but also compare molar stability estimates across concepts, or content domains of personality, as well as across different studies using different remeasurement designs. Using this approach, six different prototypic models of human stability are introduced — the persistence, lifelong openness, increasing persistence, impressionable years, midlife stability, and decreasing persistence models — and their descriptive applicability to several relevant domains is considered. Longitudinal assessments of intelligence and personality traits reveal relatively high levels of stability from early adulthood to old age, whereas extant evidence on the stability of identities, self-image, and attitudes appear to follow either the impressionable years or mid-life stability models, in which constancies in behavioral orientations are substantially lowest in young adulthood, but reach a peak in midlife, and from there either persist or decline in stability through the mature years.
