The practical rule of thumb for a p-value is that it's nothing more than the rate of expected false-positives. p=.05 means you can expect 5% of false positives

The practical rule of thumb for a p-value is that it's nothing more than the rate of expected false-positives. p=.05 means you can expect 5% of false positives. If more than 5% of your studies are failing to replicate, then something else is involved, but it's not necessarily fraudulent. Post-hoc hypotheses are usually the biggest culprit, but it's insidious to the point that researchers often don't realize they're doing it.