Catch a Tiger By His Toe: Ambivalence in Decision Making in the 2000 Presidential Election

ABSTRACT - Consumer decision making can be hindered by ambivalenceBthe presence of conflicting positive and negative evaluations of an object. We integrate and extend extant research on ambivalence in psychology and consumer behavior to examine the nature and effects of ambivalence on aspects of decision making. A framework that embeds ambivalence management in judgment processes that serve as inputs to choice processes is developed as the basis for a set of hypotheses. The results of a study of voting behavior in the 2000 presidential election indicate that ambivalence toward objects in a choice task exerts predictable, significant effects on evaluative judgments. In addition, reducing ambivalence is conducted in a systematic manner, thus creating opportunities for marketing activity.

Citation:

Robert D. Jewell, Eloise Coupey, and Mark T. Jones (2002) ,"Catch a Tiger By His Toe: Ambivalence in Decision Making in the 2000 Presidential Election", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 333-338.

Advances in Consumer Research Volume 29, 2002     Pages 333-338

CATCH A TIGER BY HIS TOE: AMBIVALENCE IN DECISION MAKING IN THE 2000 PRESIDENTIAL ELECTION

Robert D. Jewell, Virginia Tech

Eloise Coupey, Virginia Tech

Mark T. Jones, Virginia Tech

ABSTRACT -

Consumer decision making can be hindered by ambivalenceBthe presence of conflicting positive and negative evaluations of an object. We integrate and extend extant research on ambivalence in psychology and consumer behavior to examine the nature and effects of ambivalence on aspects of decision making. A framework that embeds ambivalence management in judgment processes that serve as inputs to choice processes is developed as the basis for a set of hypotheses. The results of a study of voting behavior in the 2000 presidential election indicate that ambivalence toward objects in a choice task exerts predictable, significant effects on evaluative judgments. In addition, reducing ambivalence is conducted in a systematic manner, thus creating opportunities for marketing activity.

Despite a relatively low voter turnout for the 2000 presidential election (51.2%: www.infoplease.com), reports of long waits in seemingly short lines were abundant. Equally numerous were anecdotes of voters standing in the booth, with minds not yet made up, trying to decide for which candidate they should vote. The election, so close in outcome that it was not resolved until over five weeks later, may reflect a difficulty to clearly state a stronger preference for one candidate over the other, rather than a nearly even split between staunch Gore supporters and equally avid Bush proponents.

Such difficulty in making a decision may reflect ambivalence. For instance, someone may be able to express a consistent set of positive beliefs about a candidate’s platform, but still feel uncomfortable about the candidate’s personality. This situation creates a state of conflict that may affect decision behavior. For marketers, ambivalence may result in deferred choices and residual feelings of doubt, or diminished satisfaction post-purchase. In addition, the effort to resolve ambivalence may affect aspects of decision making, such as information search, in systematic ways that can provide marketers with insights about effective forms of information provision that affect persuasion.

In this paper, we describe research designed to examine characteristics of ambivalence and its effect on decision making. We bgin with a brief review of the literature on ambivalence, and we describe alternative conceptualizations and operationalizations of the construct. Based on this review, we propose a conceptual model of ambivalence management for situations in which a decision is made with a deadline, such as an election. We develop a set of hypotheses and discuss the results of a longitudinal study designed to test them. The paper ends with a discussion of the implications and possible extensions of this research for theory and practice in marketing.

CONCEPTUALIZING AMBIVALENCE: A REVIEW

The construct of ambivalence has been defined in different ways: as a function of the discipline in which the construct is examined (e.g., psychology versus sociology (c.f. Otnes, Lowrey, and Shrum 1997)); as a function of the application for a particular piece of research (e.g., emotional reactions to behavior (e.g., Celsi, Rose, and Leigh 1993)); and as a function of the operational form of the construct (e.g., as a component of an attitude, or as a separate construct (c.f., Kaplan 1972)). A common thread through the different definitions, however, is the idea that ambivalence stems from the presence of conflict. This conflict occurs when positively and negatively evaluated aspects of an object coexist and are deemed important in the creation of the overall evaluation of the object.

Approaches from Psychology Research: A Focus on Attitudes

Ambivalence has received greatest attention as a focus of attitude research. For instance, research to understand and predict the relationship that exists between attitudes and behaviors has focused on the predictive ability of bipolar measures that summarize sets of beliefs about the attitude-object (e.g., a presidential candidate). This approach assumes that an attitude can be described on a single dimension (c.f., Cacioppo, et al., 1997). The popularity of a unidimensional view of attitude expression is reflected in the widespread use of semantic differential scales that require respondents to decide the extent to which they have a positive or negative evaluation of the attitude object (Thurstone 1928). Kaplan (1972) notes that this measurement technique can obscure the presence of situations in which attitudes are not held with certainty. The midpoint on a unidimensional scale may mean that a person is indifferent toward to the attitude object, but it can also mean the person is ambivalent: some aspects of the object are viewed positively, while others are viewed negatively.

Research that builds on the work by Kaplan (1972) and Moore (1973) reflects an increasing acceptance of the view that attitudes do not always result from object evaluations in which positive and negative evaluative processes are reciprocal, and can be readily integrated into a single, overall attitude. Efforts to understand and explain attitudinal ambivalence have examined conflict in terms of separately functioning systems or processes. For instance, ambivalence has been studied as conflict within and between affctive and cognitive systems (e.g., Abelson, Kinder, Peters and Fiske, 1982; Lavine, Thomsen, Zanna, and Borgida (1998)). In addition, ambivalence has been studied in terms of positive vs. negative evaluative processes (Priester and Petty, 1996), and different motivational systems (Cacioppo, Gardner, and Berntson 1997).

Efforts to measure ambivalence reflect the core idea that conflict is not well-captured by a unidimensional scale. In contrast to compensatory depictions of conflict management, (e.g., a weighted adding strategy (Hogarth 1987); a multi-attribute attitude model (Fishbein and Ajzen 1975)), recent work explicitly disentangles perceptions of positive and negative elements of an evaluation. Thompson, Zanna, and Griffin (1995) describe several approaches to ambivalence measurement, including comparison of positive and negative components (e.g., Katz and Hass 1988), ratio-based methods that reflect the intensity of the conflicting components, and methods that consider the similarity and the intensity of the conflicting components. The authors argue that ambivalence can stem from conflict both within and between the components of an attitude. That is, conflict can result from dissonance between positive and negative emotions, or positive and negative beliefs, or from a combination of positive (negative) emotions and negative (positive) beliefs.

Approaches from Consumer Research: A Focus on Conflict

In contrast to the attention focused on ambivalence in attitude research in psychology, ambivalence has received far less scrutiny in the consumer research. The work that does exist, however, emphasizes the effect of conflict as a characteristic of decisions that are difficult because the decision consequences matter, and because the conflict is not readily eliminated.

One approach to studying ambivalence in the consumer literature is focused on qualitative efforts to understand the causes and effects of mixed emotions. The emphasis on emotion is consistent with work by Lavine, et al. (1998), who present results that indicate a primary role of affect on attitude and behavior when affective-cognitive structures are in conflict. For example, Otnes, et al. (1997) examine how consumers cope with ambivalence that stems from psychological sources, such as a mismatch between expectations and reality, and sociological sources, such as conflict between personal desires and cultural customs. Other studies have examined ambivalence that occurs because of the need to enact consumption behaviors (Sherry, McGrath, and Levy 1993), and as the result of consumption behaviors (Celsi, et al. 1993).

A second approach to consumer ambivalence uses a quantitative research approach to examine consumer’s choice strategies for coping with conflict when negative emotional content is present in a decision situation. Hogarth (1987) predicts a shift from compensatory processing to non-compensatory processing for choice when conflict is engendered by the presence of negative emotional content in a decision. Luce, Bettman, and Payne (1998) and Luce (1998) provide empirical support for Hogarth’s prediction, and demonstrate changes in the choice share of options that enable subjects to avoid making conflict-confronting tradeoffs. As yet unaddressed in the consumer literature, however, is the issue of how consumers process information with a goal of confronting and resolving ambivalence when tradeoffs must be made in order to reach a decision that reflects a preference for one option over another. As a first step toward this goal, in the following section we present a framework for describing the role of ambivalnce in decision making.

A CONCEPTUAL MODEL OF AMBIVALENCE: A FOCUS ON DECISION MAKING

The formation of an attitude is one element of a larger process of decision making. We treat the development of an attitude toward an object, such as a product or a president, as a judgment process that results in an evaluation of the object’s desirability. Hogarth (1987) notes that judgmental heuristics are rarely discrete judgments C or ends in themselves. Instead, judgmental processes are typically part of a larger, continuous process of decision making, in which judgments serve as inputs to actions, like making a choice between two alternatives. As such, the judgments, such as attitudes, that serve as antecedents to choices should be viewed within the context of the decision situation. Under this view, the dynamic characteristics of many types of continuous judgment and choice environments should be incorporated into understanding how judgments are formed, and how they are expressed in the choice stage of the decision process. For example, one person’s attitude toward Ronald Reagan as an actor might be very different from his or her attitude toward Ronald Reagan as president. The change in the situation (i.e., Reagan runs for president) requires the person to incorporate situational feedback from the changing context into his evaluation of Reagan.

Our characterization of decision behavior as judgments and choices is compatible with other conceptualization of choice processes. For instance, our judgment stage can be viewed as a predecision phase (c.f., Payne, Braunstein, and Carroll (1978); Coupey, Bodur, and Brinberg (1998), followed by a decision phase in which alternatives are compared and one is chosen, and then a post-decision phase, in which the decision process and its outcome is evaluated. In addition, a judgment-choice view of decision behavior is also compatible with an information processing model of decision making that explicitly delineates the roles of information processing in each stage of the model. For instance, an often-invoked model of decision making in consumer research (c.f., Mowen and Minor 1998, page 349) includes the stages of pre-purchase (characterized by information search and acquisition), purchase (characterized by alternative evaluation and selection, as by lexicographic or elimination-by-aspects heuristics), and post-purchase (characterized by learning processes and dissonance reduction).

Using a decision making perspective, we develop a model in which ambivalence is a characteristic or property of an attitude. That is, rather than view ambivalence as a construct separate from the attitude, like an intention that can mediate the attitude-behavior relationship or an affective state (e.g., feeling ambivalent), we conceptualize ambivalence as a component of an attitude that reflects the presence of positive and negative feelings or beliefs that are deemed important by the decision maker. This view underscores the idea that ambivalence cannot exist independently of the attitude; an attitude is itself either ambivalent or univalent.

In the judgment phase of the decision process, an attitude is formed with respect to the attitude object, given the characteristics of the situation. The process results in an overall evaluative judgment that serves as input to the next phase, choice. An important characteristic of the judgment phase is that the formation of the judgment, such as the evaluation of an alternative, is completed within-alternative. That is, the goal of the udgment process is to return an overall evaluation of the option. While this judgment may incorporate knowledge of acceptable values and characteristics, given exposure to and experience with other options, the end result is a wholistic, independent evaluation of the option. A second characteristic of ambivalence in the judgment phase is that its does not motivate the decision maker to attempt to manage it (e.g., by resolving it or avoiding it). A person can maintain sets of conflicting beliefs and emotions for extended periods of time. The need to cope with the ambivalence is a function of the imminence of the need to make a choice between alternatives for which conflict is present within either or both of the alternatives.

In the choice phase, the decision process shifts to a between-alternatives orientation. In this phase of the overall decision process, the task of the decision maker is to manage indecision. The act of making a choice implicitly indicates that indecision has been reduced to a point at which one alternative is unambiguously preferred to the remainder of the alternatives, or that the alternatives are equally preferred and the decision maker is indifferent. Indifference in a choice between two alternatives suggests that beliefs and feelings held toward the alternatives are not strongly weighted, or intense.

The primary components of the framework, and their characteristics, are presented in Figure 1.

Framework Propositions

Our framework suggests three propositions. First (P1), the reduction of indecision suggests that as a decision maker reaches the time horizon for making a decision, the evaluations of the alternatives will tend to polarize, relative to the initial judgments of the alternatives. Second (P2), the processes to make a choice will depend on the outcome of processes to formulate overall evaluative judgments of each alternative. Thus, between the stages of being able to express an overall judgment of each alternative B independently of the other alternative B and the final expression of preference between or among alternatives (i.e., choice), ambivalence must be reduced. Further (P3), the framework suggests that there is a relationship between ambivalence reduction and evaluative judgment, such that the extent to which consumers change their judgments of alternatives over time is a function of the amount of initial ambivalence. That is, more initial ambivalence predicts greater movement in the evaluative judgments of the alternatives.

FIGURE 1

A PROCESS-BASED FRAMEWORK FOR EXAMINING AMBIVALENCE IN DECISION MAKING

AN EMPIRICAL EXAMINATION OF AMBIVALENCE EFFECTS ON DECISION MAKING

We conducted a study to examine the effects of ambivalence on aspects of decision making. To assess the veracity of our propositions, we used a decision situation with the following characteristics. First, the decision had to be important to the subject population. This was desirable to reduce the possibility that the targets of evaluations induced only indifference and satisficing behaviors. Second, th decision had to be embedded in a situation that would enable measurement of judgments and related ambivalence, over time. Third, the situation had to have a defined end, so that subjects were a) required to make a choice, and b) required to confront the ambivalence. Fourth, the decision situation had to involve targets for evaluation that were likely to engender ambivalence. To satisfy these criteria, we collected data from subjects in a seven-week period spanning the 2000 presidential election.

Sample and Procedure

Two hundred and thirty undergraduates from a major southeastern university participated in the longitudinal study for course extra credit. The study consisted of a pen and paper questionnaire that took, on average, approximately twenty minutes to complete. Data were obtained six weeks prior to the election and one week prior to the election.

Assessing Situational Validity

The ability to test hypotheses regarding ambivalence means that the choice situation must be one that engenders ambivalence. As a back-check on the appropriateness of the voting choice task, we examined scatter-plots of the correlations between subjects’ ratings of presidential candidates and their levels of expressed ambivalence. Bromer (1998) argues that ambivalent attitudes are less accessible than non-ambivalent attitudes, and that this lesser accessibility results in a diminished biasing effect of the attitude on aspects of information processing. Extended to the present study, Bromer’s rationale suggests that when ambivalence is lower, evaluative judgments about a target should tend to be extreme: held with greater certainty. In contrast, when ambivalence is higher, evaluative judgments will tend to be expressed with less certainty. This reasoning suggests that the presence of an inverted U-shaped function indicates the presence of ambivalence. Figure 2 illustrates the expected effect. The x axis reflects the ratings for Bush at Time 1, and the y axis reflects that subjective ambivalence for Bush at Time 1. This result confirms that the choice task is suitable for testing the hypotheses.

Study Hypotheses

Four hypotheses provided a situation-specific examination of our research propositions. P1 suggested that judgments over time, but prior to choice, will tend to polarize. We can recast the proposition as the operational hypothesis that:

H1: The absolute value of the difference in overall ratings of Bush and Gore will be significantly greater at Time 2 than at Time 1. That is, |B(rate)₁ BG(rate)₁| < |B(rate)₂ BG(rate)₂|.

P2 suggested that the polarization effect on judgments proposed in P1 is not simply due to the effect of time. That is, the effect is enhanced by the reduction of ambivalence.

Thus,

H2a: Subjective ambivalence for the selected candidate will be significantly smaller at Time 2 than at Time 1.

Because there is no reason, a priori, to believe that the polarization process in judgment would incorporate ambivalence reduction differently for the unselected option, we predict

H2b: Subjective ambivalence for the unselected candidate will be significantly smaller at Time 2 than at Time 1.

Our third proposition described the relationship proposed to exist between evaluative judgments and ambivalence. Recall that P3 suggested that the extent to which evaluative judgments are polarized over time depended on the initial amount of ambivalence for the selected alternative. Recast as a formal hypothesis, we predict

H3: The extent to which ratings of Bush and Gore change over time is a function of the level of initial ambivalence for the candidates. That is, the greater the level of initial ambivalence for both candidates, then the greater the amount of observed change in ratings.

Our final hypothesis is exploratory in nature. It addresses the manner in which ambivalence is managed in the decision. That is, we anticipate that reduction in ambivalence is a systematic process used to reinforce differentiation between alternatives. This reduction can be accomplished by increasing the evaluation for the initially preferred option.

FIGURE 2

AN ILLUSTRATIVE REPRESENTATION OF THE RELATION BETWEEN AMBIVALENCE AND JUDGMENT EXTREMITY

RESULTS

Data obtained from 230 subjects were analyzed to test the hypotheses and to conduct several exploratory analyses to examine aspects of ambivalence in decision making. Due to subject attrition and missing data, the total sample was reduced to 169 subjects.

Dependent Measures. Three main dependent measures were used. Following the lead of Thompson, et al. (1995) and Priester and Petty (1996), subjective ambivalence was measured with three scales. Subjects were asked to indicate on a 9-point scale the extent to which their perceptions of each candidate were a) conflicted (affective), b) indecisive (conative), and c) mixed (cognitive). Cronbach’s alpha indicated a correlation of .77 among the scales. The scale scores were averaged for each subject to create a summary index of ambivalence at Time 1 and Time 2.

The evaluative judgment was obtained from thermometer scales, on which subjects’ were asked to rate each candidate from 0 to 100. Voting behavior was derived from subjects’ statements of voting intention just prior to the election, and used as a proxy for the actual, unobservable behavior. Strength of intention was assessed for each candidate, using ratings of likelihood from B3 to +3. An intention of +2 or +3 was counted as a vote for that candidate, provided the intention for the other candidate was a B2 or -3.

Hypotheses Tests. To test H1, we used a repeated-measures ANOVA to examine the nature of changes over time in overall evaluative judgments of the candidates. The absolute values of the difference in judgments for each candidate from Time 1 to Time 2 did increase, as expected. The mean difference score at Time 1 was 39.12, compared with 42.7 at Time 2. This difference was significant (F_{(1, 168)} = 5.23, p < .02).

The next set of analyses addressed the nature of subjective ambivalence during the course of the decision process. Providing support for the prediction in H2 that ambivalence for the selected candidate and for the unselected candidate would diminish over time, a within-subjects ANOVA for each candidate indicated a significant decrease in ambivalence. For the selected candidate, subjective ambivalence decreased from 3.73 to 3.18 (F_{(1, 168)} = 19.18, p < .0001). For the unselected candidate, subjective ambivalence decreased from 4.11 to 3.66 (F_{(1, 168)} = 10.98, p < .001).

H3 proposed an influence of the amount of initial ambivalence on the extent to which judgments became polarized over time. We tested H3 with a stepwise regression. This approach enabled us to determine which factors contributed significantly to the amount of difference in the overall evaluations of the candidates from Time 1 to Time 2. The regression analysis assessed the relative contribution of ambivalence for the selected and unselected candidates, and the interaction effect of these candidates in predicting the mean value of the ratio of the absolute values of the differences between candidates at Time 1 and Time 2. Using a criterial probability of .05 for including a factor in the regression, the analysis indicated that the combination of initial ambivalence for the selected and unselected candidates accounted for the largest amount of variance in the overall ratings (beta = .277, p < .0001). This result suggests that the amount of initial ambivalence toward both alternatives affects evaluative judgments of the alternatives, rather the amount of ambivalence for one or the other alternative. This result supports the prediction of H3.

Exploratory Analyses. Our fourth hypothesis addressed the nature of ambivalence reduction. Research in confirmatory hypothesis testing shows that decision makers tend to seek out and nclude in decision representations information that provides justification for initially favorable alternatives (Klayman and Ha 1987). This effect suggests that the selected candidate should tend to exhibit a positive change in ratings. Descriptive statistics show a mean shift in ratings of 5 for the selected candidate, but 0 for unselected. That is, people shift ratings in the positive direction B on average C for selected, but that the direction is not clear for the unselected candidate. Interestingly, however, the amount of shift indicates that the processes for reducing ambivalence require additional analysis. The absolute value of the change in ratings for selected versus unselected indicate less consistency in ratings for the unselected candidate (mean absolute value of change for selected = 9.6, for unselected = 13.5). Further examination of the data reveals that the change in rating for the selected candidate is a function of the initial amount of ambivalence toward that candidate. This effect was tested and confirmed with a stepwise regression that assessed the impact of initial ambivalence and changes in ambivalence, for the selected candidate and the unselected candidate, with change in ratings for the selected candidate as the dependent measure. The regression indicated that initial felt ambivalence for the selected candidate exerted the sole significant influence on the change in rating (beta = .152, p < .05). That is, all that matters is where you start, not what you do with the ambivalence.

In contrast, for the unselected candidate, the amount of change in the ratings is a function of the amount of change in the perceptions of ambivalence, with change in ambivalence for both the selected and the unselected candidate having significant impact. This effect was illustrated with a stepwise regression of the initial ambivalence and ambivalence changes for both candidates, using the change in ratings for the unselected candidate as the dependent measure. The betas for ambivalence changes for both the selected and the unselected candidates were significant (selected beta = .26, p < .001; unselected beta = .15, p < .04).

SUMMARY AND CONCLUSIONS

Key Insights

This study provides insight into the nature and effect of ambivalence in decision making. The study results indicate that when a deadline for choice is imposed, evaluations of alternatives become increasingly polarized as the deadline nears. In addition, the data indicate that the evaluative polarization is not simply due to the nearing deadline. Conflict between the components of an evaluative judgment also influences the judgment, such that higher initial levels of ambivalence result in a greater increase in the spread of the final evaluations of alternatives. The extent to which subjects reduce ambivalence over time was shown to be significant for both alternatives.

In contrast to previous research, the use of a longitudinal study enabled the examination of patterns of ambivalence management over time, as well as the effect of these tactics on evaluative judgments. In addition, the voting situation provided the opportunity to assess the amount and influence of ambivalence in a real-life decision, thus enhancing generalizability of the results beyond what might be possible with a laboratory expeiment in which a hypothetical situation necessitated B and complicated B the creation of ambivalence.

Study Limitations

The same characteristics that make the voting decision a good one for studying ambivalence also serve as possible limitations to the conclusions we can draw. For instance, the complex nature of the decision means that it is impossible to know that all of the factors that contribute to ambivalence and its resolution can be anticipated and incorporated into the study design and analysis. To manage the complexities inherent in the decision situation, we used simple measures of subjective ambivalence that rely on subjects’ abilities to recognize and report their perceptions of conflict, indecision, and mixed feelings. Although subjective ambivalence has been shown to be an acceptable proxy for other, constructed measures of ambivalence that integrate positive and negative components (Priester and Petty 1996), we cannot know whether the simpler measures of subjective ambivalence mask or alter the appearance of patterns for managing ambivalence.

Future Directions

Further research should be conducted to examine the relative impact of conflict within and between affect and cognition in the formation of evaluative judgments as inputs for choice. For instance, is ambivalence that stems from affective conflict managed differently from ambivalence that stems from cognitive conflict, or from conflict between affect and cognition? Another issue to explore is that of when ambivalence is reduced. That is, is ambivalence reduced just before choice? Or after choice, too? The relation between pre-choice and post-choice ambivalence also suggests the need to develop conceptual and empirical explanations of whether, and how, ambivalence and dissonance are related.

Developing our understanding of ambivalence and how people manage it in decision processes can be effected through controlled laboratory experiments, in which factors presumed to engender ambivalence are manipulated, and process and outcome measures are obtained to observe coping strategies and their effect on choices. In contrast to previous work on conflict avoidance and strategy selection (e.g., Luce 1998), such research could be focused on situation in which conflict must be confronted and overall evaluative judgments must be formed.

Marketing Implications of Ambivalence

Research on ambivalence has been directed toward efforts to explain the nature of ambivalence, and to understand the effect of ambivalence on attitudes, intentions and behaviors. From the perspective of marketing application, however, the present research suggess several avenues along which ambivalence creates opportunities for practitioners to incorporate ambivalence and consumers’ strategies for coping with ambivalence into strategic marketing efforts. For instance, our research indicates that marketers can use ambivalence to create desired perceptions of difference in the goodness and badness of choice alternatives. This finding suggests that marketers can work proactively to engender and influence the experience of ambivalence from the consumer’s point-of-view.

Two applications that illustrate the opportunity to leverage ambivalence are product introductions and comparative advertising. By increasing perceptions of ambivalence about extant brands, a marketer can position a new product as an alternative for choice that is differentiated from existing brands by consistency between the components of an overall evaluative judgment. This procedure implies that the marketer should provide information, as about an attribute of the new brand, that contrasts with characteristics of existing brands, but which is consistent with the new brand.

A similar application of ambivalence is afforded by comparative advertising. By creating affective or cognitive conflict about a competing brand, a marketer may be able to influence the formation of evaluative judgments in a context that pits one brand directly against another brand. Because ambivalence and its management appear to be a less conscious process than a directed product comparison, introducing ambivalence as a means of generating a shift in evaluation may result in decreased levels of retroactive interference and psychological reactance.

REFERENCES

Abelson, Robert P., Donald R. Kinder, Mark D. Peters, and Susan T. Fiske (1982), "Affective and Semantic Components in Political Person Perception," Journal of Personality and Social Psychology, 42, 619-630.

Bettman, James R., Mary Frances Luce, and John W. Payne (1998), "Constructive Consumer Choice Processes," Journal of Consumer Research, 24, 187-217.

Bromer, Philip (1998), "Ambivalent Attitudes and Information Processing," Swiss Journal of Psychology, 57 (4), 225-234.

Cacioppo, John T., Wendi L. Gardner, and Gary G. Berntson (1997), "Beyond Bipolar Conceptualizations and Mesures: The Case of Attitudes and Evaluative Space," Personality and Social Psychology Review, 1, 3-25.

Celsi, Richard L., Randall L. Rose, and Thomas W. Leigh (1993), "An Exploration of High-Risk Leisure Consumption through Skydiving," Journal of Consumer Research, 20, 1-23.

Coupey, Eloise, Onur Bodur, and David Brinberg (1998), "Predecision Processes in Consumer Choice: Effects of Prior Knowledge on Aspects of Decision Structuring," Advances in Consumer Research, XXV, 226-232.

Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading, MA: Addison-Wesley.

Hogarth, Robin M. (1987), Judgement and Choice, 2^nd Edition, New York: Wiley.

Jonas, Klaus, Michael Diehl, and Philip Bromer (1997), "Effects of Attitudinal Ambivalence on Information Processing and Attitude-Intention Consistency," Journal of Experimental Social Psychology, 33, 190-210.

Kaplan, Kalman J. (1972), "On the Ambivalence-Indifference Problem in Attitude Theory and Measurement: A Suggested Modification of the Semantic Differential Technique," Psychological Bulletin, 77, 361-372.

Katz, I., and Glenn, R. Hass (1988), "Racial Ambivalence and American Value Conflict: Correlational and Priming Studies of Dual Cognitive Structures," Journal of Personality and Social Psychology, 55, 893-905.

Klayman, Joshua, and Young-Won Ha (1987), "Confirmation, Disconfirmation, and Information in Hypothesis Testing," Psychological Review, 94, 211-228.

Lavine, Howard, Cynthia J. Thomsen, Mark P. Zanna, and Eugene Borgida (1998), "On the Primacy of Affect in the Determination of Atiutdes and Behavior: The Moderating Role of Affective-Cognitive Ambivalence," Journal of Experimental Social Psychology, 34, 398-421.

Luce, Mary Frances (1998), "Choosing to Avoid: Coping with Negatively Emotion-Laden Consumer Decisions," Journal of Consumer Research, 24, 409-432.

Moore, Michael (1973), "Amnbivalence in Attitude Measurement," Educational and Psychological Measurement, 33, 481-483.

Mowen, John C., and Michael Minor (1998), Consumer Behavior, 5^th Edition, Upper Saddle River, NJ: Prentice-Hall, Inc.

Otnes, Cele, Tina M. Lowrey, and L. J. Shrum (1997), "Toward an Understanding of Consumer Ambivalence," Journal of Consumer Research, 24, 80-93.

Payne, John W., Myron L. Braunstein, and John S. Carroll (1978), "Exploring Predecisional Behavior: An Alternative Approach to Decision Research," Organizational Behavior and Human Performance, 22, 17-44.

Priester, Joseph R., and Richared E. Petty (1996), "The Gradual Threshold Model of Ambivalence: Relating the Positive and Negative Bases of Attitudes to Subjective Ambivalence," Journal of Personality and Social Psychology, 71, 431-449.

Sherry, John F., Jr., Mary Ann McGrath, Sidney J. Levy (1993), "The Dark Side of the Gift," Journal of Business Research, 28, 225-244.

Thompson, Megan M., Mark P. Zanna, and Dale W. Griffin (1995), "Let’s Not Be Indifferent About (Attitudinal) Ambivalence," In R. E. Petty and J. A. Krosnick, (eds.), Attitude Strength: Antecedents and Consequences, 361-386, Hillsdale: Erlbaum.

Thurstone, L.L. (1928), "Attitudes Can Be Measured," American Journal of Sociology, 33, 529-554.

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