An epidemiologist in Mumbai designs a cohort study to examine the relationship between oral contraceptive (OC) use and venous thromboembolism (VTE) in women aged 20–45 years. Over 5 years, 10,000 OC users and 10,000 non-users are followed. The crude relative risk (RR) for VTE in OC users is 3.5 (95% CI: 2.8–4.4). However, when the analysis is stratified by the presence of Factor V Leiden mutation, the RR in Factor V Leiden carriers is 8.2 and in non-carriers is 2.1. The crude RR of 3.5 lies between these two stratum-specific RRs. What epidemiological phenomenon is being demonstrated?