Sociodemographic Effects on the Test-Retest Reliability of the Big Five Inventory

Abstract

Abstract. Psychometric properties of self-report scales can be affected by sociodemographic differences among respondents. For example, factor structures of established personality scales tend to be distorted in samples with less formal education. Whether test-retest reliabilities are comparably affected is of yet not well known. Therefore, this study examined the measurement precision of a short Big Five instrument in a diverse sample of the German population. A sample of 1,696 (50% women) participants reported on their personality twice within an interval of 10 weeks. The median test-retest reliability for the five traits, rtt = .66, was notably smaller than previously reported coefficients from college students, median rtt = .78. Moderator analyses identified modest effects of age and educational attainment on these reliability estimates, whereas sex showed no differential impact. These results highlight that test-retest reliabilities derived from student samples should not be generalized to sociodemographically diverse groups of respondents.