Effect modification was defined as the effect of an exposure varying with a third variable. Effect modification by age occurs, for example, if the relative risk increases with rising age. The fact that the stratum-specific relative risks are constant means therefore that there is no effect modification when the multiplicative scale is used. The multivariate logistic regression model, for example, is an example of a multiplicative model. A consequence of this is that a logistic model without interaction term has no effect modification with multiplicativity as criterion but usually with additivity as criterion. The occurrence of synergism is evaluated by comparing the actual incidence rate for the "doubly exposed" with that which would be expected in the absence of synergism. The presence of synergism can thus also be evaluated on the basis of relative risks obtained.