The structure of perfectionism: a twin study.

Abstract

Perfectionism may be a premorbid risk factor for eating disorders. Evidence of familial transmission suggests features of perfectionism may be genetically determined. This study examines the structure of perfectionism using classical twin design models. Independent (IP) and common (CP) pathway models are used to investigate the extent to which genetic and environmental factors can help to identify and differentiate three behavioral domains of perfectionism as measured by a shortened version of the Multidimensional Perfectionism Scale (MPS) [Frost et al. (1990). Cognit. Ther. Res. 14: 449-468]. Three of the original subscales were included: Personal standards (PS), Doubts about actions (DA), Concern over mistakes (CM). We studied a sample of 1022 paired and unpaired female twins from the Mid-Atlantic Twin Registry. MZ correlations were consistently higher than DZ twin correlations for all three composite subscales. The multivariate independent pathway model provided a better fit to the twin correlations then did the more parsimonious common pathway model suggesting the pattern of familial resemblance for the three subscales is not well characterized by a unidimensional perfectionism factor. CM phenotypic variance was completely accounted for by common heritability influences in both the IP and CP models. Based on the IP model results, there was evidence that PS and CM but not DA shared some common genetic effects, with DA and CM sharing some common environmental factors. These multivariate twin modeling results support conceptualizations of perfectionism as a multidimensional construct. The biometric structural results for the three subscales examined here suggest CM is the core feature of Perfectionism with DA and PS serving as indicators of CM. Although not the best fitting model, the common pathway model estimated this behavioral domain to be isomorphic with the construct of perfectionism. The better fitting independent pathway model provided evidence of non-trivial differences in the pattern of heritability for CM, DA, and PS.