How do physicists approach statistical significance?

In particle physics, significance is set at 5 sigma—a p value of 3 × 10–7 or 1 in 3·5 million (if the result is not true, this is the probability that the data would have been as extreme as they are).

Compare that with educational research: we typically set it at 1.96 sigma. Imagine how many studies would never be published if we raised the threshold.