Next, to examine the interaction effects between parental divorce/parental relationship discord and offspring PRSALC on alcohol outcomes, product terms between mean-centered parental divorce/ discord and PRSALC were calculated (i.e., parental divorce × PRSALC; parental relationship discord × PRSALC) and added to the previous main effects models. A statistically significant product term would provide evidence for interaction on a multiplicative scale. For example, when the combined effect of parental divorce/discord and PRSALC is larger (or smaller) than the product of the individual effects, this would indicate an interaction on a multiplicative scale. Alternatively, if the combined effect of parental divorce/discord and PRSALC is larger (or smaller) than the sum of the individual effects of parental divorce/discord and PRSALC, this would be consistent with interpretation for interaction on an additive scale (Knol et al., 2007; Li & Chambless, 2007). A relative excess risk due to interaction (RERI) of zero would indicate there is no interaction on the additive scale (i.e., perfect additivity). A RERI >0 indicates positive interaction on an additive scale, whereas a RERI <0 indicates negative interaction on an additive
