Do self-reports and informant-ratings measure the same personality constructs?
Personality researchers often supplement or substitute self-reports with ratings from knowledgeable informants, at least implicitly assuming that the same constructs are measured regardless of the source of ratings. However, measurement invariance (MI) of personality constructs across these rating types has rarely been empirically tested. Here, this was done for the Five-Factor Model domains and their 30 facets (N = 3,253). Four domains and 26 facets showed the level of invariance (metric MI) required for comparing the relative standings of individuals across self-reports and informant-ratings, which is what researchers mostly do. However, 28 of the 35 scales failed to achieve the level of invariance (scalar MI) recommended when comparing mean scores. Self-informant pairs who contributed to higher MI also tended to display higher cross-rater agreement. In conclusion, self-reports and informant-ratings appear to measure reasonably similar constructs for most research purposes, but poor MI may contribute to imperfect cross-rater agreement.