RESUMO
Animal personality, consistent interindividual differences in behavior through time, has been intensively studied across animal taxa and particularly in nonhuman primates. Two different methods have been used to study personality: questionnaires filled out by trusted raters, following the research tradition in human personality psychology, and behavioral observations or testing, based on the behavioral ecology research tradition. Systematic research of cross-method validity has, however, brought equivocal results. Here we report a systematic method comparison with strict validation criteria in common marmosets (Callithrix jacchus). We compared questionnaire data, observational behavioral data, and experimentally assessed behavioral data independently collected in two different research laboratories and across two years. Notably, we constructed a priori predictions on the degree and direction of correlations between the questionnaire-derived constructs and behaviorally derived constructs/variables. Convergent validity of the constructs/variables was present but was not found in both laboratories for any of the constructs. Discriminant validity was found to some degree for all constructs in both laboratories, with important exceptions. It seems that marmoset personality structure obtained with descriptor ratings does not consistently capture the predicted behavioral variables. Our finding that these two methods do not quite reach common ground thus urges caution in the choice of the suitable methods to study personality in nonhuman primates. We discuss the directions to which animal personality research should go so that it can get the "best of both worlds." (PsycINFO Database Record (c) 2020 APA, all rights reserved).