arguing 'credibility' of a datum, while he was explaining 'categories' of er, data.