A significant three-way interaction indicated that higher impulsivity was associated with stronger stimulant response (Table 2). These results were found while holding other variables constant: family history, gender and test day (Table 2). To illustrate this effect, plots of stimulant response by dose and timepoint among the most and least impulsive quartiles are given in Figure 1. After reviewing plots of subjective response by dose over time among the four impulsivity quartiles, our next step in interpreting the three-way interaction was to fit additional nonparametric models to assess changes between the timepoints when clamped ethanol infusion occurred (between the +10 and +30 timepoints and between the +30 and +60 timepoints) and across the timepoints following the end of clamped ethanol infusion (from the +110 through the +230 timepoint). In a model to compare the +10 with the +30 timepoint, the three-way interaction was again significant (num df = 1.93, ATS = 7.37, p = 0.001). While differences were not observable by impulsivity level or dose condition at the +10 timepoint, the increase in stimulation between the +10 and +30 timepoints
