An experimental study of the effects or reinforcement, question length, and reinterviews on reporting selected chronic conditions in household interviews