Perceptual measures of performance: Fact or fiction?

In contrast with the general management and strategy literature, performance measurement in the extant empirical operations management literature has relied heavily on perceptual measures of operational and financial performance. At the same time, rigorous examinations of the psychometric properties of such performance measurements instruments have not been performed. Especially needed are examinations of systematic informant bias and the related method variance and method bias, which are not captured by conventional methods used in operations management to assess construct reliability and validity. We conduct a multitrait-multimethod analysis of perceptual performance measures to investigate item-specific trait, method and error variance. Data from 164 factories with three informants per factory is analyzed. We find that while random error and systematic bias account for a large portion of item variance, perceptual measures satisfy the requirements of reliability and validity. However, we caution against the use of single-informant studies in future research as substantive conclusions may be affected.