I don't think he's faking data. I think it's more a very strong case of bias confirmation, lazy thinking and lazy sourcing.