Abstract

Person-fit statistics measure the likelihood of a candidate’s questionnaire responses, given the model used to calculate their scores. An unlikely set of responses would be to get all the difficult questions right and the easy questions wrong. Low person-fit on a personality questionnaire might be due to inattention when answering, an attempt to manipulate the results, psychological traits of the candidate, or an untraited individual. We tested these hypotheses using 100-item Big Five test results from 150,000 participants and 6 different subsamples, including self-monitoring scores, NEO facet scores, impression management and fair-mindedness scores, time taken to complete a questionnaire, test-retest correlations, and self-other correlations. We found that person-fit is stable over 275 days (mean r = .46), and that suspicious individuals, and intellectual and imaginative individuals have a low person-fit. We find scant evidence that low person-fit can improve the validity of a personality questionnaire by removing inattentive respondents.