Scenario #2 Investigators based their hypothesis generation about the association between oral contraceptive use and ovarian cancer on the relationship between ovarian cancer and pregnancy. Pregnancy was described as potentially protective against ovarian cancer. The investigators were interested in seeing whether the association between oral contraceptive use and ovarian cancer differed for women of different parity (number of babies delivered). To investigate this, they compared four different studies. Table 2 presents results from 4 studies (these data are fictitious) for the crude OR and the stratum-specific associations. For the purposes of this exercise, do not worry about statistical significance. Also assume that parity is not on the causal pathway between oral contraceptive use and ovarian cancer. Table 2. Comparison of crude and stratum-specific estimates in four different studies Study Crude OR Parity = 0 Parity 2 1 Escargots et al 1.25 0.87 0.86 Baguette et al 0.78 0.77 0.76 Champagne et al 0.95 1.97 1.98 Bearnaise et al 0.56 1.98 0.23 Based on the data shown in Table 2, complete the following table to indicate whether there is confounding or effect modification for each study. Study Effect Modification Confounding Escargots et al 8. a. Yes 9. a. Yes b. No b. No Baguette et al 10. a. Yes 11. a. Yes b. No b. No Champagne et al 12. a. Yes 13. a. Yes b. No b. No Bearnaise et al 14. a. Yes 15. a. Yes b. No b. No