Document Type




Publication Date



Great Lakes Institute of Management

Source Publication

Great Lakes Herald


Studies in the field of Decision Sciences that employ multi-item rating scales to measure latent constructs have predominantly used ANOVA rather than Means and Covariance Structure Analysis (MACS) in order to investigate group mean differences. However, traditional statistics in ANOVA (e.g., t and F) attenuate when dealing with imperfect measures, which in turn potentially leads to incorrect interpretation of results in the form of accepting the false null hypothesis and/or underestimating the true effect size. To address this issue, we describe in this paper a new but simple procedure to disattenuate the ANOVA-based statistics for measurement error. Using previously published studies, we provide an illustration for practically implementing this procedure that has not been used in prior literature. A major implication of our work is that scholars in decision sciences can now report correct estimates of test statistic and enhanced effect size when examining between-group mean differences, thereby leading to a richer and more appropriate interpretation of findings in contemporary research.


Published version. Great Lakes Herald, Vol. 12, No. 1 (March 2018): 72-91. Publisher link. © 2018 Great Lakes Institute of Management. Used with permission.

durvasula_13315acc.docx (122 kB)
ADA Accessible Version

Included in

Marketing Commons