The replication crisis is not largely due to the math/stats used in studies.

It comes from trying to make inference from noisy and often confounded experiments from limited data.