If the findings are replicated, thats fine, you can begin to trust.

But the findings are often not replicated.