For each participant, the values of the 6 candidate variables during cognitive testing for both alcohol and placebo sessions were submitted to a multivariate divergence analysis, a simple type of discriminant analysis that searched for the subset of variables producing the largest divergence between alcohol and placebo data [28], [29]. The divergence analysis considered all possible subsets of up to 3 of the 6 variables and chose the subset that, considered together as a group, best recognized alcohol. This resulted in an equation which produced a score that could be used to classify data samples from an individual as being either from alcohol or placebo/pre-drinking conditions.
