The design of communication campaigns to alter health behaviors often begins with the identification of behavioral beliefs assumed to be causal antecedents of behavioral intentions. The assumption beliefs are causal derives from various theories of belief and intention/behavior and from statistical patterns of correlation. In cases of high-risk/cost campaigns, presuming causal order should require additional evidence. One approach is the parallel encouragement design which involves "randomly encouraging" levels of the mediator to establish its causal linkage to the outcome. This study proposes and tests a novel method of randomly encouraging beliefs as the mediator of messages on intention. Results show that semantic priming altered misbeliefs about Natural American Spirit cigarettes which in turn influenced intentions, suggesting its utility as an encouragement method to establish causal mediation of beliefs in message effects models. Results for countercampaign messages and broader theoretical and practical implications are discussed.