I'm worried we're overstating the results of the study. The paper can be downloaded for free from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4453958 and if you go look at Section 4 ("results"), figures 1 and 2, you'll note that:
1. The 95% confidence interval error bars almost all overlap each other...
2. ... and this is even after they truncated the y-axis to make the differences more visible (i.e. appear bigger).
I'm worried we're overstating the results of the study. The paper can be downloaded for free from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4453958 and if you go look at Section 4 ("results"), figures 1 and 2, you'll note that:
1. The 95% confidence interval error bars almost all overlap each other...
2. ... and this is even after they truncated the y-axis to make the differences more visible (i.e. appear bigger).