The Moderating Role of Decision Task Goals in Attribute Weight Convergence
Korea Unversity Business School
Consumer choice can differ greatly depending on relative importance he or she places on different attribute dimensions (e.g., price or quality). Also, managers rely on consumer’s expressed attribute weight when developing market segmentation and product configuration. Due to such practical importance, issues regarding attribute weight have been studied extensively and a wide range of measurements have been proposed in past research. Some are direct measures (i.e., weights are measured using self-reported rating or self-reported point allocation), whereas others are in direct measures (i.e., weights are statistically derived from the evaluation of product options using different evaluation tasks such as rating, ranking and choice). However, existing research shows only weak to moderate convergences among various direct and indirect measurement techniques. The lack of inter-measurement convergence can be troubling to managers who make important strategic decisions based on different approaches often used interchangeably (Horsky, Nelson, & Posavac, 2004; Kramer, 2007).
In this study we uncover the potential determinant of inter-measurement divergence, thereby help predict the conditions under which attribute weight convergence can improve. Specifically, we posit that processing goals underlying different measurement tasks play important roles. Decision tasks differ in how much a decision maker feels that he or she needs to differentiate an alternative from the other alternatives during an evaluation. In some tasks (e.g., choice), the need for differentiation is high since one has to select one option while giving up on others. In contrast, other tasks (e.g., rating) evoke individual evaluation goal and hence are associated with minimal need for differentiation. The distinction between differentiation and individual evaluation goal is important because it influences consumer’s attribute weighting rule. When evaluation task calls for differentiation processing goal, people tend to assign disproportionately great weights to the most important dimension. On the other hand, tasks characterized by individual evaluation goal are less susceptible to such bias. This difference, in turn, leads to the reduced overall correspondence among the measurements evoking different processing goals.
Our hypotheses are tested in two studies employing a multi-attribute procedure. In the studies, indirect attribute weights are derived from participants’ preferences among the sixteen hypothetical product profiles based on various decision tasks (choice, ranking, rating, inclusion). Direct attribute weights are measured by self-reporting and point allocation. Study 1, in which various compared, shows that weight assignment is a function of type of decision tasks employed in evaluation context. When the product evaluation is made based on the tasks characterized by high differentiation processing goal (choice and ranking), disproportionately greater weight is given to what the evaluator perceives as the most important attribute. This, in turn, leads to the overall divergence with the weights inferred by tasks classified as evoking individual evaluation goal (rating and inclusion). Further, study 2 extends the results to the convergence between direct and indirect measurements. The results show the improved convergence among the measures sharing similar processing goals. Specifically, the maximum inter-measurement convergence was found when both direct and indirect measurements evoke differentiation goal (i.e., point allocation · choice / point allocation-ranking) or individual evaluation goal (i.e., self-repot rating · rating / self-repot rating · inclusion).
The demonstration of convergence between direct and indirect weight assessments has tremendous practical importance for managers. Direct measures are often used in consumer studies to predict weights that are derived indirectly using multi-attribute decision tasks in actual shopping environments. Our findings suggest that a good understanding of the types of decision tasks that consumers use most when evaluating their products in the actual market place, and then incorporating this knowledge when designing self-reported evaluation measures in their market research can greatly enhance the value of their study. For instance, when the shopping environment presents a decision mode that prompts the goal of distinguishing one option from others (e.g., Expedia.com), the use of point allocation or other non-zero sum evaluation methods would enhance the prediction validity of the market research. Conversely, if the purchasing environment presents an evaluation task that activates the goal of individual evaluation (e.g., Pricaline.com), direct importance ratings would yield better predictions.