Complex models and the conjugation fallacy: a caution