I'd read the back and forth on this controversy when it happened, and in general I agree with that article -- as it applies to fields such as medicine. It applies less to fields like CS (although there's data that proofs are frequently wrong).
A common mistake is to do things such as mining for correlations. A stronger grounding in statistics would definitely help most scientists (and ppl in general -- one reason I support the push to replace calculus with statistics in HS math).
Although oddly, even if most finding are false they typically create the basis to do an actual real follow-up study, with actual hypothesis' which are much more likely to give real results. IOW, these false findings still accelerate the rate that science occurs. Although better science upfront would be more efficient still.
A common mistake is to do things such as mining for correlations. A stronger grounding in statistics would definitely help most scientists (and ppl in general -- one reason I support the push to replace calculus with statistics in HS math).
Although oddly, even if most finding are false they typically create the basis to do an actual real follow-up study, with actual hypothesis' which are much more likely to give real results. IOW, these false findings still accelerate the rate that science occurs. Although better science upfront would be more efficient still.