The Multidimensional Perfectionism Cognitions Inventory–English (MPCI–E): Reliability, Validity, and Relationships With Positive and Negative Affect

Abstract

The Multidimensional Perfectionism Cognitions Inventory (MPCI; Kobori & Tanno, 2004) is a promising new instrument developed in Japan to assess perfectionism cognitions regarding personal standards, pursuit of perfection, and concern over mistakes. In this study, we examined reliability and validity of the English version of the MPCI, the MPCI–E (Kobori, 2006), in a sample of 371 native English speakers. A confirmatory factor analysis confirmed the MPCI–E's 3-factorial oblique structure. Moreover, correlations with measures of dispositional perfectionism and past-week positive and negative affect provided initial evidence of the MPCI–E's convergent and differential validity. Finally, hierarchical multiple regressions indicated that the MPCI–E showed incremental validity in explaining variance in positive and negative affect above variance explained by dispositional perfectionism. Overall, the findings provide initial evidence for the reliability and validity of the MPCI–E as a multidimensional measure of perfectionism cognitions that has the potential to further the understanding of positive and negative cognitions in perfectionism.

Acknowledgments

We thank Kathleen Otto and two anonymous reviewers for helpful comments and suggestions on an earlier version of this article. This research was supported by a Promising Researcher Initiative grant from the University of Kent to J. Stoeber.

Notes

^a All factor correlations are significant at p < .001.

* p < .001.

¹In the United Kingdom, parental consent is only required for participants under the age of 16 years (British Psychological Society, 2005).

* p < .05.

** p < .01.

*** p < .001.

²In SPSS syntax, COMPUTE scale score = MEAN.x(items), with x = k – 1 and k = number of items in the scale.

³Although EQS for Windows 6.1 provides estimation methods to estimate missing values, these methods require normal distribution of variables (see Normality section). To be able to compute robust statistics, EQS needs a complete set of raw data (Bentler & Wu, 2004; Byrne, 2006).

⁴Because self-oriented perfectionism scores showed a higher reliability (Cronbach's alpha) than self-oriented perfectionism scores, we investigated if the differences were due to the differences in reliability of measurement. However, when we computed structural equation models separating measurement model from structural models (Kline, 2005), the results were the same: The MPCI–E scales explained considerably more variance in self-oriented perfectionism (R ² = .662, p < .001; f ² = 1.96) than in socially prescribed perfectionism (R ² = .451, p < .001; f ² = 0.82), with f ² denoting effect size (see Cohen, 1992). Further details can be obtained from J. Stoeber on request.

⁵Before exploring the differences between these correlations, we conducted Meng et al.'s (1992, Formula 5) test of the heterogeneity of a set of correlated correlations. The test was significant for both positive affect, χ²(2, N = 363) = 44.35, and negative affect, χ²(2, N = 363) = 29.51, both ps < .001, indicating that the correlations differed significantly.

* p < .05.

** p < .01.

*** p < .001.

^a N = 198.

^b N = 358.

^c N = 60.