Placebo efficacy was monitored during three separate experiments conce
rned with the effects of ethanol (0.8 ml/kg body weight) on some aspec
t of performance: at regular intervals during experimental sessions, b
lood alcohol concentration was estimated using a breathalyser and subj
ects completed an intoxication rating. The simple placebo beverages an
d manipulations employed were similar to those typically described in
the literature, and each experiment involved a repeated measures desig
n. Across the three experiments, maximum mean (median) ratings during
placebo sessions, expressed as a percentage of those during alcohol se
ssions of equivalent period, ranged between 10% (0%) and 69% (72%), an
d the number of subjects in each experiment for whom a placebo effect
was considered to have been negligible ranged between 8% and 73%. For
each treatment, intoxication ratings were higher during treatment peri
ods which occurred first than during those which occurred second. Howe
ver, this transfer effect was twice as large for placebo treatments as
for alcohol treatments and, whereas in respect of the latter the effe
ct was statistically non-significant, it was highly significant in res
pect of the former. The occurrence of such asymmetrical transfer of pl
acebo efficacy suggests that repeated measures designs should be used
with caution in drug studies in which a convincing placebo treatment i
s necessary and difficult to devise.