Nonparametric analyses were the most appropriate option given the non-normal subjective response data. However, due to concern about ambiguity in the interpretation of rank order scores and loss of information from converting impulsivity score into an ordinal variable, we confirmed results of the final nonparametric models using linear mixed effects (LME) models with random subject effects and other predictors treated as fixed effects. Results from the LME models were very similar to the ATS results and thus are not reported here (results available from the first author).
