A preference-based measure of responses to warning labels was used to
study alternative alcohol warning labels that differed in: (a) length,
(b) presence of qualifier words, (c) alternative content, and (d) spe
cific beverage. The specific risks mentioned were more important than
the overall label length, qualifier words reduced avoidance responses,
various alternative warnings elicited more avoidance responses than t
he existing alcohol warning label, and more avoidance responses were m
ade to labels on a whiskey bottle than on a beer bottle. Warnings with
the words ''poison,'' ''cancer,'' or ''health problems'' were the mos
t powerful. The results for college students in Study 1 were replicate
d in a high-school sample in Study 2.