What could explain the result is self-confidence. The data comes from a source that needs to be double-blind but wasn't. What would be interesting would be to

What could explain the result is self-confidence. The data comes from a source that needs to be double-blind but wasn't. What would be interesting would be to do have one party write two separate emails talking about a different subject. Then randomly switch the name/gender of one of the emails. Then have a second party "grade" the emails. With the "author's name" clearly marked on the email. I'd like to see the results.