Special Session Summary New Insights Into Measuring Attitudes and Intentions: Implications For Survey Design, Data Analysis, and Consumer Behavior

Citation:

Barbara Bickart (1999) ,"Special Session Summary New Insights Into Measuring Attitudes and Intentions: Implications For Survey Design, Data Analysis, and Consumer Behavior", in E - European Advances in Consumer Research Volume 4, eds. Bernard Dubois, Tina M. Lowrey, and L. J. Shrum, Marc Vanhuele, Provo, UT : Association for Consumer Research, Pages: 191-192.

European Advances in Consumer Research Volume 4, 1999      Pages 191-192

SPECIAL SESSION SUMMARY

NEW INSIGHTS INTO MEASURING ATTITUDES AND INTENTIONS: IMPLICATIONS FOR SURVEY DESIGN, DATA ANALYSIS, AND CONSUMER BEHAVIOR

Barbara Bickart, Rutgers UniversityBCamden, U.S.A.

SESSION OVERVIEW

Marketers rely extensively on attitude and intentions measured via surveys, both for making management decisions and for theoretical research on consumer judgment and decision processes. In this session, we explored some new insights into measuring attitudes and intentions. Specifically, the papers in this session examined how questionnaire design and individual difference factors affect responses to attitude questions and subsequent behavior and the implications of these effects for survey design and data analysis. In addition, we touch on the more general implications of this research for the formation and structure of consumer attitudes and the relationship between attitudes and behavior.

We began the session by examining how two characteristics of a surveyCquestion order and question framingCaffect reported attitudes and opinions. In the first paper, titled "Hong Kong 1997 in Context", Priya Raghubir and Gita Johar examine these issues in the context of attitudes toward the change in sovereignty in Hong Kong for one’s own and other countries. They examine part-part order effects and find attitude polarization such that positive attitudes regarding the effect of the transition on the future of Hong Kong (China) become more extreme if they are elicited after similar responses regarding a related entity, China (Hong Kong). They also find evidence of carryover effects from responses to specific questions on subsequently posed general questions. Finally, they examine the effect of framing the transition issue as a "reunification", "handover" or "takover" on attitudes toward the effect of Hong Kong’s transition on Macau and Tawan. The negative "takeover" frame elicits less favorable responses when respondents are motivated to hold attitudes perceived as valid (i.e., for attitudes toward one’s own country), but not otherwise. These findings have important implications given that issues such as the future reunification of Taiwan with the People’s Republic of China may be based on public opinion collected through a referendum.

The second paper, titled "How Response Styles Weaken Correlations from Rating Scale Surveys", by Eric Greenleaf, Barbara Bickart, and Eric Yorktson, further explores the effects of individual differences on attitude measurement. This paper shows how response styles weaken correlations between two attitude measures. The authors examine the source and size of this weakening in two large studies of consumers’ attitudes, interests, and opinions, conducted in Australia and the United States. The results indicate that weakening due to response styles is often as large as, or larger than, weakening from scale discreteness, so that actual weakening in survey data is at least twice as large as previously suspected. Further, they provide evidence that response style weakening represents measurement error and not differences in true correlations. In another study, the authors show how response style weakening can distort comparisons made between individuals with different response styles. In particular, comparisons of the relative weight given to attributes or beliefs in predicting attitudes can vary significantly depending on whether or not the correlations are adjusted for this weakening. The authors recommend a method to adjust for response style weakening when comparing correlations across groups and individuals.

The final paper, titled "The Mere-Measurement Effect: Why Does Measuring Purchase Intentions Change Actual Purchase Behavior?", by Vicki Morwitz and Gavan Fitzsimons, examines how responding to a purchase intent question affects subsequent behavior. The results of seven experiments provide both further support for the occurrence of the mere-measurement effect and a more clear understanding of the cognitive mechanism through which it operates. The authors find that when asked to provide purchase intentions, consumers are more likely to choose brands for which they hold positive and accessible attitudes, and are less likely to choose brands for which they hold negative and accessible attitudes, compared to a control group of consumers who are not asked an intentions question. These results provide support for the hypothesis that the mere-measurement effect operates through an increase in attitude accessibility.

The papers in this session focus on major issues in attitude measurement. These issues include questionnaire design, dealing with individual response styles, and the effects of responding to intention questions on the accessibility of related attitudes and subsequent behavior. All three papers point to the constructive nature of the attitude measurement process and describe some of the outcomes of this process. Further, the papers provide some insight into how the error associated with attitude measures might be minimized, either through questionnaire construction, question wording, or data analysis. We feel that the field of attitude measurement could benefit from a clearer understanding of attitude formation and structure and vice versa. We hope the papers presented in this session and the subsequent discussion have helped to further this dialogue.

 

HONG KONG 1997 IN CONTEXT

Priya Raghubir, University of California, Berkeley, U.S.A.

Gita Johar, Columbia University, U.S.A.

On July 1, 1997, Hong Kong ceased to be a British colony and became the first Special Administrative Region of the People’s Republic of China (PRC). We report the results of three field experiments, conducted in Hong Kong, Macau, and Taiwan (n=1250) from June to August 1997, that examine attitudes toward the impact of Hong Kong’s transition as a function of question order and framing We invoke two aspects of attitudes: (1) whether they are stored in a memory network, and (2) whether their validity is important to the respondent, to predict when order and framing effects are most likely to occur. Experiment 1 examines part-part order effects and reveals attitude polarization such that positive attitudes regarding the effect of the transition on the future of Hong Kong (China) become more extreme if they are elicited after similar responses regarding a related entity, China (Hong Kong). Experiment 2 focuses on part-whole order effects and finds evidence of carryover effects from responses to specific questions on subsequently posed general questions. Experiment 3 examines the effect of framing the transition issue as a "reunification" "handover" or "takeover" on attitudes toward the effect of Hong Kong’s transition on Macau and Taiwan. The negative "takeover" frame elicits less favorable responses when respondents are motivated to hold attitudes perceived as valid (i.e., for attitudes toward one’s own country), but not otherwise. We conclude by suggesting that reliance on survey data in making policy decisions could result in the implementation of decisions that are, objectively speaking, unpopular when the survey instrument evokes response biases.

 

HOW RESPONSE STYLES WEAKEN CORRELATIONS FROM RATING SCALE SURVEYS

Eric Greenleaf, New York University, U.S.A.

Barbara Bickart, Rutgers UniversityCCamden, U.S.A.

Eric Yorkston, New York University, U.S.A.

Market researchers need to understand data disortions created by survey measures and how to compensate for them. This paper examines how response styles weaken true correlations between marketing variables measured with rating scales. This adds to already recognized correlation weakening from scale discreteness. We examine both kinds of weakening in two surveys completed by large and representative samples of the adult populations of Australia and the United States. We find that correlation weakening caused by response styles is often as large as, or larger than, weakening from scale discreteness, so that actual weakening in survey data is at least twice as large as previously suspected. We also verify that response style weakening represents measurement error and not differences in true correlations. We show that one important consequence of this weakening is that it reduces statistical power in survey research. Another consequence is that it causes distortions when researchers compare individuals with different response styles. We recommend a method to adjust for response style weakening when comparing correlation-based models across such individuals. We describe how a similar distortion can affect analyses that compare groups with different response styles, such as in demographic segmentation.

 

THE MERE-MEASUREMENT EFFECT: WHY DOES MEASURING PURCHASE INTENTIONS CHANGE ACTUAL PURCHASE BEHAVIOR?

Vicki G. Morwitz, New York University, U.S.A.

Gavan J. Fitzsimons, University of Pennsylvania, U.S.A.

What is the outcome of asking a consumer to report his or her intention to perform a specific action? Traditional wisdom dictates that respondents retrieve pre-existing intentions and that these intentions should be good predictors of their subsequent behavior. However, recent research has shown that the actual process may be more complicated (Fitzsimons and Morwitz 1996; Kalwani and Silk 1982; Morrison 1979; Morwitz, Johnson, and Schmittlein 1993). Several studies have shown that respondents may not retrieve a pre-existing intention but rather may construct a response to an intentions question only after the question is asked (Feldman and Lynch 1988; Simmons, Bickart, and Lynch 1993). Furthermore, a number of studies have demonstrated that the act of forming and reporting a response to an intentions question can alter respondents’ subsequent behavior (Fitzsimons and Morwitz 1996; Greenwald, Carno!t, Beach, nd Young 1987; Morwitz et al. 1993; Sherman 1980).

Several different alternative explanations have been proposed to explain why this "mere-measurement effect" occurs. However, these explanations have not been tested to date. The purpose of this paper is to test several competing explanations for why measuring intentions changes behavior.

The results of seven experiments provide both further support for the occurrence of the mere-measurement effect and a more clear understanding of the cognitive mechanism through which it operates. We find that when asked to provide purchase intentions, consumers are more likely to choose brands for which they hold positive and accessible attitudes, and are less likely to choose brands for which they hold negative and accessible attitudes, compared to a control group of consumers who are not asked an intentions question. We argue that these results provide support for the hypothesis that the mere-measurement effect operates through an increase in attitude accessibility.

----------------------------------------