Incentive salience measures based on three IAT modifications
According to recent neurobiological models, food choices are influenced by two separate reward systems: motivational wanting (incentive salience of the reward) and affective liking (hedonic pleasure associated with the reward). Both are assumed to have conscious and unconscious components. Applying such promising conceptual frameworks within consumer research would not only be helpful for understanding human appetite but also has implications for predicting consumer behaviour. Since the affective liking system has strong similarities to contemporary attitude theories, implicit and explicit measures of evaluation could be applied. However, no comparable procedures have been developed for the motivational wanting component; generally accepted “low-tech” measures are therefore still lacking! Thus, the aim of this study was to develop and test implicit measures of wanting that can be used as dependent variables in consumer and sensory research. Modified versions of three IAT paradigms were developed, including the conventional implicit association test (IAT) and two recent modifications, the single-block IAT (SB-IAT) and the recoding-free IAT (IAT-RF). All three tests were designed to measure the relative strength of the associations between two opposing motivational tendencies (approach, avoidance) and two target products (liquorice, wine gum), each represented by a series of verbal and visual stimuli. Reaction time was recorded as the dependent variable. The difference between the three paradigms was that the test task changed block-wise (IAT), remained constant (SB-IAT), or changed trail-wise (IAT-RF). Using a between-subjects design (N=131), the implicit measures were validated against explicit measures of wanting and liking as well as behaviour (amount of candy eaten). Common difference scores were calculated and a latent-difference bifactor model was fitted to the data to obtain a more stringent psychometric representation. The modified IAT-RF had excellent psychometric characteristics, was highly predictive of behaviour and showed strong convergent validity with explicit measures of wanting. Furthermore, its discriminant validity with respect to explicit measures of liking was extremely satisfactory. The much criticized IAT was also strongly related to explicit measures but weakly related to behaviour. The SB-IAT did not discriminate and was systematically influenced by task order.