Scott D. Swain, Danny Weathers, and Ronald W. Niedrich
Executive Summary
Researchers often include both reversed and nonreversed items in multi-item Likert scales in an effort to minimize the consequences of respondent inattention and/or acquiescence. However, some have expressed concerns about this technique in light of evidence linking item reversal to problems such as unexpected factor structures, diminished scale reliabilities, and confounds in cross-cultural research. These problems may result from a phenomenon the authors call “misresponse,” which occurs when a respondent selects an option in opposition to his or her corresponding beliefs.
The initial goal of this research was to assess the prevalence of misresponse to reversed Likert items in survey data because simulation studies indicate that misresponse rates as low as 10% can compromise basic analyses. The authors develop a procedure for identifying misresponse and, using this procedure, analyze data from previously published studies. In these studies, which span a variety of research settings and respondent groups, misresponse to reversed items exceeded problematic levels, averaging nearly 20%.
The second goal of the research was to conduct a theoretical and empirical assessment of three potential sources of misresponse. First, the authors consider the conventional wisdom that misresponse occurs when inattentive respondents commit insufficient resources to the response task. They develop an account of inattention by integrating the notion of contingent capture with the status quo response heuristic. Second, they consider the phenomenon of acquiescence, which is characterized by a person’s tendency to exhibit uncritical agreement with assertions. The authors link acquiescence to misresponse through the dual-stage model of belief. Third, evidence in psycholinguistic research suggests that information in negated form is more difficult to use in postcomprehension tasks than information in affirmative form. To account for these differences in item verification difficulty, the authors introduce the constituent comparison model. To tease apart the effects of inattention, acquiescence, and item verification difficulty on misresponse, they manipulate both polarity and item truth value (i.e., whether the item is true or false for the respondent) in a series of four experiments.
Across these experiments, the authors observe substantial levels of misresponse, despite low levels of respondent inattention and acquiescence. In contrast, the data indicate that misresponse varies systematically with item verification difficulty. Specifically, as the constituent comparison model predicts, the authors find that true affirmative items are the least difficult to verify (as indicated by misresponse rates and response latencies), followed in increasing order of difficulty by false affirmative, false negated, and true negated items. Extending these findings, they also investigate the impact of different types of negation (affixal, particle, and implicit) and need for cognition on misresponse. Overall, these findings suggest that item reversal, a technique designed to remove bias due to inattentive or acquiescent respondents, can induce substantial misresponse in respondents who are attentive and nonacquiescent.
This research holds several implications. First, if researchers want to use scales that contain reversed items, reversal through means other than negation might be beneficial. For example, they might be able to use affirmative items that are antonymic to the construct being measured. The authors provide guidance for constructing such items. Second, the high levels of misresponse to Likert items indicate that alternative item formats may prove useful. Further research should investigate the potential of interrogative scaling. Finally, the research develops methods for identifying and classifying misresponse according to its underlying source, an important prerequisite to appropriate analytical adjustments.
Biography
Scott D. Swain is Assistant Professor of Marketing in the School of Management at Boston University. His research interests focus on the intersection of marketing research methodology and consumer judgment and decision making. His work has appeared in Journal of the Academy of Marketing Science, as well as several trade publications and refereed conferences. Scott earned an MBA and a PhD in Business Administration at the University of South Carolina, a BS in Electrical Engineering at Clemson University, and a BS in Physics at Francis Marion University.
Danny Weathers is Assistant Professor of Marketing in the E.J. Ourso College of Business at Louisiana State University. His research interests include methodological and measurement issues, pricing, and online marketing, and his research has appeared in various marketing and statistics journals, including International Journal of Research in Marketing, Journal of Business Research, and Journal of Applied Statistical Science. Weathers has a BS in Math from Clemson University, an MS in Statistics from the University of South Carolina, and a PhD in Business Administration from the University of South Carolina.
Ronald W. Niedrich is Kearny-Jolly Endowed Professor of Marketing in the E.J. Ourso College of Business at Louisiana State University. His research focuses on consumer judgment and decision making and appears in various marketing journals, including Journal of Consumer Research, Journal of the Academy of Marketing Science, Journal of Retailing, Journal of Advertising, and Journal of Business Research. Niedrich has a BS in Engineering from Clemson University, an MBA from the College of William and Mary, and a PhD in Business Administration from the University of South Carolina.
Journal of Marketing Research, Vol. XLV, No. 1
View Table of Contents.