How can we establish the biological factors that contribute to variation in assortative mating (based on a quantitative or qualitative trait)? Key assumptions: Assortative (or disassortative) mating for a particular trait can produce sexual isolation between ecotypes or incipient species. The individual contribution to population assortative mating for a quantitative trait can be estimated by means of the ri statistic, which is an additive decomposition of the Pearson correlation coefficient. The mating pair contribution to population sexual isolation can be estimated by the PSI coefficient. These statistics can be used to quantify the variability in assortative mating/sexual isolation in a particular population. Search method: It was recently proposed that both the ri statistic and the PSI coefficient could be used as dependent variables in a multiple regression approach to determine which of a set of independent variables explains the greatest variation in the dependent variable. We describe both statistics and undertake simulations to compare the efficiency of each statistic to infer assortative mating when it is caused a priori by a mate choice decision based on a quantitative or a qualitative trait. Conclusions: The ri statistic outperforms the PSI coefficient when trying to infer the causes of both assortative mating and sexual isolation. The applicability of both methods to other cases is discussed.