Estimation of Relative Risk Under M to R Matching

Estimation of the relative risk from a rare disease is carried out using M:R matching. Both conditional and unconditional likelihood methods are used, leading in each case to the same estimate; a non-iterative estimate is always available under M:R matching, M ≦2. The method also enables the testing of homogeneity even when information on the matching variate is unavailable and this is an advantage over logistic regression methods.