Research spotlight on reliability statistics

Faculty research:  “Meta-Analysis of Coefficient Alpha: A Reliability Generalization Study.”  Published in Journal of Management Studies.

NIU Business author: Assistant Professor of Management Bethany Cockburn.  Additional Author(s): Lindsey M. Greco, Oklahoma State University, Ernest H. O’Boyle, Indiana University, and Zhenyu Yuan, the University of Iowa.

Impact of Article:  Researchers rely on reliability statistics to learn about and communicate the precision of their data. One commonly
used reliability statistic is the coefficient alpha (alpha), which conveys how internally consistent the data is. The benchmark cutoff for alpha values is .70.

In our meta-analytically calculated sample we found that every construct was above the minimum cut-off, although there was significant variability between constructs. Using a meta-analytic technique known as reliability generalization, we cumulated alphas across 36 commonly used individual differences, attitudes, and behaviors from 1675 independent samples (N = 991,588).

Our findings have practical implications for researchers. First, we suggest that the previous benchmark of .70 should be replaced with higher and more construct specific cutoffs. Second, we provide baseline alphas that can be used for research planning and design. Finally, we offer best practices for reliability generalization.


Reprinted from the Dean’s “2018 NIU College of Business Research Report.”  This article is one of several published in journals that are used by Financial Times to compile the FT research rank, which is included in the publication’s global MBA, MBA, and executive MBA and online MBA rankings.


Print Friendly, PDF & Email