Attentional Contrast during Sequential Judgments: A Source of the Number-of-Levels Effect

Conjoint analysis is used to measure the importance of attribute-level trade-offs. A methodological anomaly is the number-of-levels effect; that is, as the number of intervening attribute levels increases, the derived importance weight of an attribute increases. The authors use three studies to show that attentional processes contribute to the number-of-levels effect. When there is an inequality in the number of levels across attributes, a given profile may include levels of one attribute that are relatively more novel than levels of the accompanying attributes. A process of attentional contrast directs attention toward the relatively novel attribute levels within each profile. Increased attention toward these attribute levels results in a larger derived importance weight for the attributes defined on those levels.