On the notion of importance in research, and the misuse of relative importance analysis.
To inform policy, recent guidelines emphasize that researchers should produce findings that are important (Aguinis & Vandenberg, 2014; Bettis, Ethiraj, Gambardella, Helfat, & Mitchell, 2016). Statistical techniques promising ways of assessing the relative importance of different regressors thus seem attractive. Consequently, it is not surprising to see that a technique to rank-order regressors (Budescu, 1993) has (a) become very popular in applied domains like management research, and (b) also triggered the development of alternative techniques that are less computationally demanding. However, do these tools deliver on their promises? And, do applied researchers use these tools appropriately?
This presentation will review the two most common relative importance analysis techniques and how they are used in research, discuss the notion of importance, present the results of a simulation that put these methods to the test, and conclude with some practical guidelines.