Preferences inferred from choice are more likely to favor the alternative t
hat is superior with respect to the prominent (most important or salient) a
ttribute than are preferences inferred from matching (direct tradeoff) judg
ments. This prominence effect violates standard models of rational choice a
nd complicates the task of measuring preferences. In this article, we propo
se a new task-goal hypothesis regarding the prominence effect: The prominen
t attribute receives more weight in tasks whose goal is to differentiate am
ong options than in tasks whose goal is to equate options. We use this hypo
thesis to generalize the prominence effect beyond choice and matching to se
veral additional tasks, including the choice-based matching and difference
comparison methods that are widely employed in decision analysis. The resul
ts of three studies provide strong support for the task-goat account of the
prominence effect and cast doubt on competing explanations. We discuss the
implications of these findings for descriptive decision theory and for pre
ference measurement in decision analysis, public policy, and marketing.