Where the evidence comes from

It comes from doing the thing at a size you can survive, repeatedly. Speaking in a seminar before speaking at a conference. Sending one difficult email before a difficult meeting. Each one is a data point, and the point is not that it goes well — it is that it happened and you are still here.

This is why it is domain-specific rather than a general personality setting. Somebody assured in a lab can be lost in an interview, and the reason is not character. It is that they have a record in one place and none at all in the other.