22 The sample size of the groups is uneven. To ensure that sample size is not driving the results, I create three different random
samples of firms that rely on financial measures that total 37 firms each. I then rerun the analysis three times using each of
the random samples paired with the 37 firms that use non-financial measures. I find that the statistical inferences regarding the
significance of INTER are unchanged. In two of the three samples, I find that the statistical inferences regarding the significance
of DIFF and HCAP are unchanged. Thus, I conclude that uneven group sample size is not driving the results.