Title: On the Relationship between Treatment Effect Heterogeneity and the Variability Ratio Effect Size Statistic

Abstract: Recently, the variability ratio (VR) effect size statistic has been used with increasing frequency in the study of differences in variation of a measured variable between two study populations. More specifically, the VR effect size statistic allows for the detection of treatment effect heterogeneity (TEH) of medical interventions. While a VR that is different from 1 is widely acknowledged to implicate a treatment effect heterogeneity (TEH) the exact relationship between those two quantities has not been discussed in detail thus far.
In this note we derive a precise connection between TEH and VR. In particular, we derive precise upper and lower bounds on the TEH in terms of VR. Moreover, we provide an exemplary simulation for which VR is equal to 1 and there exist TEH.
Our result has implications for the interpretation of VR effect size estimates regarding its connection to treatment effect heterogeneity of (medical) interventions.