OB Research Seminar - Sirio Lonati, NEOMA Business School

The Impact Threshold of a Confounding Variable (ITCV) is an increasingly popular technique in management studies. A widespread approach involves comparing the ITCV with control variable correlations to assess the presence of omitted variable bias. In this talk, I will show why this practice should be avoided. First, the ITCV and control variable correlations represent fundamentally different mathematical quantities that cannot be directly compared. Second, a fair comparison requires a rescaled version of the ITCV. Third, even the interpretation of this rescaled ITCV becomes complex in the presence of multiple omitted variables, numerous control variables, and correlations between the omitted and control variables. These results suggest abandoning simple rules of thumb based on the comparisons between the ITCV and control variable correlations.