leading questions response bias and omitted variables