Affective temperaments in clinically-well subjects in Turkey: initial psychometric data on the TEMPS-A

Vahip S., Kesebir S., Alkan W., Yazici C., Akiskal K., Akiskal H.

JOURNAL OF AFFECTIVE DISORDERS, cilt.85, ss.113-125, 2005 (SCI İndekslerine Giren Dergi) identifier identifier identifier

  • Cilt numarası: 85
  • Basım Tarihi: 2005
  • Doi Numarası: 10.1016/j.jad.2003.10.011
  • Sayfa Sayıları: ss.113-125


Background: This is a first attempt to evaluate the reliability and factor structure replicability of the Temperament Evaluation of Memphis, Pisa, Paris and San Diego Autoquestionnaire (TEMPS-A) in its Turkish Version. The questionnaire is a self-report 110-item measure that postulates five affective temperaments-the depressive, cyclothymic, irritable, hyperthymic, and anxious-which embody both strengths and liabilities along affective lines. Methods: The questionnaire was administered to 658 clinically-well subjects in a Turkish university circle. We undertook item analysis and test-retest reliability. We then examined internal consistency through factor analysis with PCA rotation. Results: We found good to excellent test-retest reliability (0.73-0.91), and internal consistency (0.77-0.85). We deleted 10 items with factor loading < 0.20 for their own subscales, resulting in a questionnaire with 99 items. Despite considerable overlap between depressive and cognitive anxiety traits, a distinct "nervous"-anxious factor emerged as well, and the hypothesized (original English) 5-factor structure of the TEMPS-A was supported. Cut-offs for each temperament were based on z-scores higher than +2S.D. Dominant irritable (3.7%). nervous-anxious (3.7%) and depressive (3.1%) temperaments were the most common in this population, whereas dominant cyclothymic (1.7%) and hyperthymic (1.2%) temperaments were relatively uncommon. These temperaments tended to lose their intensity with age. As expected, women scored significantly higher on the nervous-anxious, and men on the hyperthymic temperaments. Limitations: The sample was composed of younger subjects with higher education than the General population of Turkey. Although the distribution of the scores for each of the temperaments deviated somewhat from normal Curves, for heuristic reasons we did attempt to provide prevalence rates based on z-scores. Conclusion: In this preliminary version of the TEMPS-A, we have retained 100 (of the original 110) traits loading > 0.20. Some deleted items referred to sleep, others appeared socially desirability traits in the Turkish culture endorsed by many subjects. Nonetheless, item analyses within each factor revealed traits indicative of personal assets (specific to each temperament) along with those which might represent vulnerability to affective illness. This is in line with the hypothesized original theoretical framework of the senior authors. Even in this "first pass," in its Turkish version the TEMPS-A is a reliable and valid instrument. Further refinement of the instrument will require the study of a nationally representative sample in Turkey. (c) 2003 Elsevier B.V. All rights reserved.