This article makes a very good point, but I think that statistical error is the tip of the iceberg. A much bigger problem is outright fraud, bought studies by corporations, and a lack of interest in reproducing others' experiments. Incorrect, fraudulent, and/or otherwise misleading studies are printed all the time and are never retracted. The one that are retracted often remain available in online archives, where they are read and cited by new researchers.
TL;DR I think scientists have been put under too much pressure to produce and publish and the work has gotten sloppy.