What is being assessed and why it matters: the impact of transient error on trait research

J Pers Soc Psychol. 2009 Jul;97(1):186-202. doi: 10.1037/a0015618.

Abstract

Temporal instability can reflect either true psychological change or transient measurement error, and it is important that trait psychologists be able to distinguish one from the other. The authors report results from large retest studies of Big Five, trait affectivity, and personality disorder measures across time frames (2 months and 2 weeks) over which these constructs should show little or no true change. On average, nearly 25% of the variance in the measures was a product of transient error rather than true change; however, the proportion of error varied widely-but consistently-across measures. In addition, a reexamination of long-term longitudinal data demonstrated that ignoring transient error can lead to inaccurate conclusions. Most notably, a substantial portion of the observed instability in the Big Five and trait affectivity is due to transient error; thus, these traits are even more stable than commonly thought. The present data further suggest that previous reports of differential stability between the Big Five and trait affectivity are due, in part, to differential levels of transient error in measures of these constructs.

MeSH terms

  • Adolescent
  • Bias
  • Character*
  • Female
  • Humans
  • Longitudinal Studies
  • Male
  • Personality Assessment / statistics & numerical data*
  • Personality Disorders / diagnosis
  • Personality Disorders / psychology
  • Personality Disorders / therapy
  • Psychometrics / statistics & numerical data
  • Reproducibility of Results
  • Students / psychology
  • Young Adult