On the other hand, any time a hypothesis appears significant, the first reaction should be to verify that all the data going into the calculation is correct, rather than just assume it is. In my day-to-day industry experience, significant results come far more often from incorrect data than an actual discovery.