Belief importance in expectancy-value models of attitudes

In this study, 312 respondents were asked to indicate their attitude toward smoking and their smoking behavior. Attitudes were assessed by a direct attitude measure (4 items) and a series of 15 belief statements about the possible consequences of smoking. Next, respondents were asked to select the 3 consequences they found most important. Attitude scores derived from these important beliefs were more predictive of behavior than those based on all 15 beliefs, and far superior to attitude scores based on the remaining 12 beliefs. Differences between the 2 groups (smokers vs. nonsmokers) were more pronounced when attitude scores were based on important beliefs, and disappeared for scores based on nonselected, less important beliefs. It is argued that incorporating belief importance in expectancy-value models helps to assess the structure of attitudes and could improve our insight in the determinants of behavioral preference.