Let’s Fix Diversity Measurement for Continuous Variables: Failings of Current Practice and a New Coverage and Evenness Index (CEI)

Research into team diversity is flourishing, yet often inconclusive. While diversity measures need to reflect the construct under investigation, this often goes wrong for continuous variables such as age or tenure diversity. Reviewing 95 papers that report relationships between age/tenure diversity and team performance, we find a fundamental mismatch: while most hypotheses appear to be concerned with diversity as variety (maximized with many different groups), most measures reflect separation (maximized with two polarized groups). The most widely used measures—the coefficient of variation and standard deviation—reflect separation, while existing variety measures require discretizing the continuous variable, losing distance information. Here, we propose a measure that captures variety on a continuous variable using all available information: the Coverage & Evenness Index (CEI). Compared to established measures, we show that it produces an intuitive ordering of teams and increases statistical power, as shown through a simulation. By adopting the CEI, diversity researchers can reduce false negatives, improve comparability across studies, and generate findings that more faithfully test their hypotheses. We provide R code for calculating the CEI in the divMetrics package.