Comparability of personality trait scores across adulthood (contribution to an invited symposium: Developmental aspects of personality)

Investigating age-differences in the personality traits subsumed under the Five-Factor Model has been a prominent personality research topic for years. Importantly, comparing test scores of people at various points across the life-course is based on the assumption that the scores represent the same latent traits at different age-levels. To date, however, this assumption has been only rarely explicitly tested. In this presentation, data from nearly 2,200 Estonians, aged between 18 and 89 years, was used to test for cross-sectional measurement invariance of the 30 NEO Personality Inventory-3 facets across adulthood. In addition to self-reports, ratings from knowledgeable informants were used. Analyses were carried out at the level of single items, which were treated as ordered- categorical variables. In self-reports, most facet scales met the criterion of metric but not of scalar invariance, implying that, for the majority of the scales, the same item scores corresponded to different latent trait levels at different ages. In other words, items belonging to the same facets tended to exhibit different age-differences and, therefore, aggregate scores could have masked more nuanced developmental patterns. In informant-ratings, however, most facets met the criterion of scalar invariance. This may suggest that personality trait levels are more easily comparable across adulthood if they are based on informant-ratings rather than self-reports. Alternatively, given that age-differences were generally weaker in informant-ratings than in self-reports, this may suggest that informants failed to perceive age-differences to their full degree and complexity. In either case, care is warranted when studying age-differences in personality traits.