Suppose that a researcher is trying to determine whether jogging is good for a person's health. She examines this question in two ways. In method A, she looks to see whether joggers live longer than nonjoggers. In method B, she looks to see whether jogging reduces cholesterol in the bloodstream and lowers blood pressure; then she asks whether lower cholesterol and blood pressure prolong life. Which of these two methods will produce reduced-form evidence and which will produce structural model evidence?