We are grateful to Hidetoshi Saito for his careful and reasoned critique of theVARBRUL procedure in his article, “Dependence and Interaction in Frequency DataAnalysis in SLA Research” (this issue). Saito reanalyzes Young's (1988, 1991)study of -s plural variation in the English interlanguage of native speakers of Chinese.He raises two criticisms of the statistical analyses in the original work: (a) data from allparticipants were lumped together, resulting in an analysis that ignores possible variation acrossparticipants; and (b) interaction between independent variables was not investigated, whereasSaito finds that an interaction term contributes to a statistical model that fits the data better thanthe original analysis.