Why?
Science is all about reproducibility. Without a strong grasp of statistics, you do not know whether you've actually discovered an effect, or whether you just got (un)lucky with your data. And far from being a theoretical problem, numerous scientific disciplines, especially those in biology and medicine, are currently dealing with a "reproducibility crisis", as they find that prominent results don't recur when replication experiments are run.
Far from there being too much math in science, I think there's currently too little math. If researchers knew more about the limits of the statistical tests they were using, rather than focusing on the magic p=0.05 level, we'd have much higher quality science, with a much higher level of confidence that the results being reported were real results, rather than statistical abberations.