Interesting that their take is that the outliers represent measurement errors. Plausible and reasonable. But the fact that it changes the conclusion of the study deserves skepticism.