A case-control study is proposed to test the hypothesis that recent respiratory tract infection increases the risk of myocardial infarction (MI). For each case, three controls will be selected from among those patients who visit the case's primary care doctor on the day of the case's MI. As the epidemiologist on the review panel, you express concern that such controls may not be appropriate, since respiratory infections are a common reason for visiting a doctor. You explain that, as a result, the observed odds ratio would be (assuming no other source of bias and no confounding):

1. an overestimate of the true increased association (bias away from the null).
2. an underestimate of the true increased association (bias towards the null).