Executive Summary
Normatively, people should be sensitive to changes in the range of the attribute levels but not to changes in the number of attribute levels within a specific range. However, several studies from the conjoint literature show that derived importance weights increase with increases in the number of attribute levels, with the range held constant.
In the search for methods to control for the number-of-levels effect, several accounts of the effect have been offered. First, it has been proposed that the number-of-levels effect is related to the data collection method, the measurement scale for the dependent variable, and the parameter estimation procedure. Although tests of these methodological accounts have allowed researchers to reduce the size of the number-of-levels effect, no methodological adjustment has completely eliminated the bias. Second, researchers have acknowledged that the effect could be a consequence of attentional processes. More specifically, it has been proposed that respondents may assign more weight to attributes with more levels because novel attribute levels draw attention or because more attribute levels result in a level changing more frequently across profiles. These accounts rely on nonrelational directed attention because the number-of-levels effect for one attribute is predicted to be independent of the number of levels of the other attributes (e.g., moving from a 2 x 2 design to a 4 x 4 design will increase the absolute importance of both attributes). Several studies have included one or more tests of nonrelational directed attention, but they have failed to find support for the hypothesis.
Methodological factors and psychological accounts based on nonrelational directed attention cannot fully account for the number-of-levels effect. The authors propose that another source of the effect is attentional contrast—attention directed toward relatively more novel attribute levels in sequential judgments, reflected in attribute importance. For example, a 4 x 2 design will create a number-of-levels effect relative to a 2 x 2 design because each level of Attribute 1 appears half as often as each level of Attribute 2.
Three experiments, each using three designs, examine whether attentional contrast contributes to the number-of-levels effect. The authors investigate these designs using two different procedures. The first procedure used a choice-based conjoint task and a hierarchical Bayesian estimation technique. The second procedure used a full profile task with reservation price estimates as the dependent variable and an analysis of variance technique. In Experiment 1, the authors show that the number-of-levels effect occurs only when one attribute has more levels than a second attribute (i.e., a relative novelty effect). In Experiment 2, the authors manipulate the relative novelty of one attribute’s levels and produce a number-of-levels effect when both attributes have the same (absolute) number of levels. In Experiment 3, the authors use a design in which the levels associated with an attribute having fewer levels are made relatively more novel than levels associated with an attribute having more levels and obtain the number-of-levels effect on the attribute having fewer (absolute) levels. The results of the three studies show that the relative novelty of attribute levels contributes to the number-of-levels effect.
Biography
Els De Wilde is Assistant Professor of Marketing in the Faculdade de Economia at the Universidade Nova de Lisboa (Portugal), where she teaches Consumer Behavior and International Marketing. She holds a Master’s degree in Psychology (1999) from the KULeuven (Belgium) and a PhD in Marketing (2005) from the University of Florida. Her research interests are in the areas of consumer behavior and consumer judgment and decision making. Previously, she held positions as postdoctoral researcher at the KULeuven and as Assistant Professor of Marketing at HEC Montréal (Canada). She has published in Memory and Journal of Consumer Research.
Alan D.J. Cooke is Associate Professor of Marketing and Graduate Coordinator of the Marketing Department in the Warrington College of Business Administration at the University of Florida, where he has taught since 1997. He received his PhD in Cognitive Psychology from the University of California, Berkeley, and his BS in Engineering Psychology from Tufts University. His research focuses on consumer decision making, particularly the role of context and comparison processes on preference. His research has been published in academic journals, such as Journal of Consumer Research, Journal of Marketing Research, and Journal of Experimental Psychology. He has taught in the undergraduate, traditional, and executive MBA and PhD programs on various topics.
Chris Janiszewski is the Faricy Professor of Marketing in the Warrington College of Business Administration at the University of Florida. His current research focuses on perception, learning theory, and context effects as applied to price perception, consumer responses to advertising, and consumer purchase behavior. He has published in Journal of Consumer Research, Journal of Marketing Research, and Journal of Marketing and is a member of the editorial boards of Journal of Consumer Research, Journal of Consumer Psychology, Journal of Marketing, and Journal of Marketing Research.
Journal of Marketing Research, Volume 45, Number 4, August 2008
View Table of Contents.