Results from a study using the balanced placebo design to assess alcoh
ol's effects on anxiety raise serious doubts about the utility of the
design even at moderate blood alcohol concentrations. Despite being in
formed that they were not drinking alcohol, 44% of the subjects who we
re administered alcohol reported consuming at least some alcohol. More
over, subjects' scores on the Marlowe-Crowne Social Desirability Index
(MCSD) were associated with deception success, suggesting that indivi
dual differences in reporting bias influence the manipulation. In cont
rast. only 6% of subjects who were told that they had consumed alcohol
but were given tonic water were not deceived. Among subjects in this
placebo condition, scores on the MCSD were not associated with success
of the deception. These data suggest that at a moderate dose of alcoh
ol, drink deception in the antiplacebo condition is much more difficul
t to execute than in the placebo condition, and that deception in the
former condition may be confounded with experimenter demand.