Confirmatory factor analysis and measurement invariance by gender, age and levels of psychological distress of the short TEMPS-A

Background The Temperament Evaluation of Memphis, Pisa, Paris and San Diego - Autoquestionnaire (TEMPS-A) is a widely used self-reported tool aimed at measuring the affective temperaments that define the bipolar spectrum, with cyclothymic, depressive, irritable, hyperthymic, and anxious subscales. Confirmatory factor analysis (CFA) was rarely used to confirm the expected five-factor model. Measurement invariance was never tested. Methods Cross-sectional, survey design involving 649 Italian college students (males: 47%). The short 39-item TEMPS-A and the 12-item General Health Questionnaire (GHQ-12) were used as measures of the affective temperaments and of psychological distress, respectively. CFA was applied to the TEMPS-A. Measurement invariance by gender, age and levels of psychological distress on the GHQ-12 was calculated with the establishment of subsequent equivalence constraints in the model parameters across groups. Results The expected five-factor model had the best fit for all CFA indexes. Configural, metric and scalar invariance of the five-factor model of the TEMPS-A was proved across gender, age and levels of psychological distress of the participants. The hyperthymic temperament subscale has low or no links with the other affective temperament subscales, which were interrelated with medium to large effect sizes. Limitations College students might be not representative of the general population. No information on the clinical status of the students was available beyond self-report data. Conclusion The study proved the measurement invariance of the (short) TEMPS-A, which is a pre-requisite to compare groups or individuals in cross-sectional and longitudinal surveys. Generalizability cannot be assumed without replication of the findings in clinical samples. © 2013 Elsevier B.V.