Sequential Phases of Judgment and the Value Representation of Product Alternatives

ABSTRACT - Consumer decisions are presumably driven by stable preferences or attitudes (i.e., value representations), but recent advances call this assumption into question. Because research from a variety of literatures suggests that such values change when processing is thorough and deliberative, this study posits a three-phase process by which judgment will influence attribute importance and feature desirability differently than choice. Results replicated previous research regarding decision process, and documented bi-directional and polarized change in subjects' value representations of alternatives under judgment, but not under choice.

Citation:

James R. Bailey and Robert S. Billings (1994) ,"Sequential Phases of Judgment and the Value Representation of Product Alternatives", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 437-441.

Advances in Consumer Research Volume 21, 1994      Pages 437-441

SEQUENTIAL PHASES OF JUDGMENT AND THE VALUE REPRESENTATION OF PRODUCT ALTERNATIVES

James R. Bailey, Rutgers University

Robert S. Billings, Ohio State University

ABSTRACT -

Consumer decisions are presumably driven by stable preferences or attitudes (i.e., value representations), but recent advances call this assumption into question. Because research from a variety of literatures suggests that such values change when processing is thorough and deliberative, this study posits a three-phase process by which judgment will influence attribute importance and feature desirability differently than choice. Results replicated previous research regarding decision process, and documented bi-directional and polarized change in subjects' value representations of alternatives under judgment, but not under choice.

INTRODUCTION

Consumers presumably possess an underlying constellation of attitudes or preferences that guide judgment of and choice between product alternatives that includes at least two key evaluative components: the importance attached to attributes (e.g., the price of an automobile) and the desirability attached to features an attribute can assume (e.g., the specific price of $15,000). Both components are implicated in models of decision making. For example, conjunctive and disjunctive models assume feature desirability is compared to a predetermined standard. The lexicographic model posits that alternatives are first compared on the most important attribute. Similarly, the elimination-by-aspects model holds that attributes are weighted according to their importance to determine the order of examination, and then specific features are compared to predetermined minimum criteria. Compensatory models, whether additive or multiplicative, assert that feature desirability is compared within alternatives.

All of these models assume that attribute importance and feature desirability (referred to herein as values or value representations) are activated from memory and applied to alternatives during processing. However, because traditional decision research has emphasized information integration not representation (see Hastie 1991), we know little about the factors that shape such values. The emerging perspective is that they are not stable, but rather are constructed by, among other things, task factors of the decision problem (see Payne, Bettman and Johnson 1992 for a review). One task factor that has been implicated in the malleability of values is response mode.

Response mode effects are strong and pervasive in decision making. One well-replicated effect involves preference reversals as a function of bidding on versus stating a preference for gambles. Another involves differential weighting of predominant attributes as a function preference assessment (i.e., matching or choice; see Tversky, Sattath and Slovic 1988 for a review of both effects). In addition to outcome, response mode also affects decision process. For example, Schkade and Johnson (1989) demonstrated that response mode influenced the time taken to render decisions, the pattern of information search, and the extent to which alternatives are directly compared.

Although numerous response modes are possible, in multiattribute decisions like those faced most often by consumers, the modes of judgment (i.e., the explicit evaluation of alternatives) versus choice (i.e., the selection of one alternative from many) have received the most attention. Using an information display board, Billings and Scherer (1988) showed that judgment led to more thorough, less variable, and more interdimensional information acquisition patterns than choice. Controlling for information acquisition and time, Bailey, Billings and Strube (1989) found that judgment led to superior recall of product alternatives when compared to choice.

Several arguments suggest that precisely because of these differences in processing, value representations change under judgment, but not under choice. The following section reviews these arguments by examining the judgmental process in relation to research on attitude change, impression formation, and information integration.

SEQUENTIAL PHASES OF JUDGMENT

What is of primary interest is the nature and degree of value change instigated by the processes involved with judgment versus choice. Understanding this change is predicated on recognizing that judgment is not uniform, but rather entails a sequence of phases, each of which alters values accordingly. Based on literature in this tradition, the sequence can be delineated into three interrelated phases-activation, evaluation, and confirmation-each with its own unique effect.

Activation

Logically, the first cognitive act of decision makers when they become aware of a domain of objects to be judged is to activate the relevant constellation of values. However, this is a memory-based process and as such is not comprehensive (Pennington and Hastie, 1986). It stands to reason that the nature of decision tasks would influence recall differentially.

Biased recall. Direct support for the position that the activation of values is less than complete comes from research on the attitute-behavior relationship. Wilson, Dunn, Kraft, and Lisle (1989) reasoned and demonstrated that the lack of attitude-behavior consistency found in some research is due to attitude change caused by processing reasons as opposed to feelings. Specifically, the more thorough evaluation induced by focusing on reasons led to recalling a biased or incomplete set which was adopted as the new attitude, thereby accounting for the discrepancy between attitudes and behavior. The resulting change was shown to be bidirectional, as some change was positive, some negative.

Because the thorough processing cited by Wilson et al. (1989) parallels that evoked by judgment, it is reasonable to expect biased recall under judgment. However, choice also requires that existing values be activated, and is therefore susceptible to biased recall as well. Nevertheless, there are other potential sources of value change in the judgment process, especially in the evaluation phase.

Evaluation

The partial activation of values should be updated in the next phase, evaluation, as decision makers encounter information that reminds them of their original representation. Simultaneously, though, further alterations would occur as trade-offs between attributes and features are considered and previously unknown information is integrated.

Compensatory strategies. One of the central differences between judgment and choice is that the former triggers greater use of compensatory strategies than the latter. Recognizing and employing trade-offs between features may cause decision makers to reevaluate the desirability attached to these features or the importance attached to the relevant attributes. For example, if a decision maker allows the low price of an automobile to compensate for poor gas mileage, the importance of price may increase, whereas the importance of gas mileage may decrease. Similarly, poor gas mileage may become less undesirable, and low price may become more desirable. Overall, the resulting change would be bidirectional, as some trade-offs would lead to positive change and others to negative change.

On the other hand, choice leads to noncompensatory deliberation that proceeds in service of identifying the alternative most harmonious with the initial representation (Tversky 1972). Hence, trade-offs are not emphasized and value representations are less likely to be reevaluated.

Impression formation. Related to a compensation-based explanation is one that concerns the manner in which information is processed under judgment versus choice. Judgment requires the formation of overall evaluations of each alternative, where information regarding specific attributes must be combined into a coherent whole. This process is not fundamentally different than that described in research on impression formation. Fiske and Neuberg (1990) have proposed a continuum model where the primary process is category-based, which is easy to execute and relies on general distinctions. Category impressions will persist unless the perceiver is motivated to reevaluate them. If so motivated, a more thorough, attribute-based process is engaged, where existing impressions will change as they become more individuating. Once again, the resulting change would not be uniformly more positive or negative, but rather some of both (i.e., bidirectional), depending on the individual's perception of the information.

The extension to choice and judgment in a decision situation is straightforward. Choice is more of a category-based process in that it activates but does not reevaluate existing representations, whereas judgment is more attribute-based because existing values are activated and further reevaluated, and new information is integrated.

Confirmation

The third phase-confirmation-occurs after the judgment of all alternatives are formed, and the decision maker reviews the value representations employed in the process. The motivation to assess whether judgmental designations reflect underlying value representations arises both from consistency strivings (cf. Festinger 1959) and a desire to avoid post-decisional regret (Janis and Mann, 1977).

Self-generated cognitions. An extensive research program by Tesser (see 1978 for a review) has found that instructing subjects to reflect on their attitudes causes polarization; positive attitudes become more positive, negative attitudes become more negative. The mechanisms implicated in the polarization process include: (a) the generation of beliefs that are consistent with existing cognitions and attitudes, (b) the failure to consider cognitions inconsistent with existing attitudes, and (c) the reinterpretation of ambiguous cognitions about the attitude.

The polarization dynamic is part cognitive and part motivational in that it reaffirms the decision and relieves the decision maker of any anxiety. This strongly resembles the well-documented bolstering effect in choice contexts (Janis and Mann 1976). However, under choice bolstering occurs predominately between chosen and non-chosen alternatives. In contrast, under judgment each alternative is evaluated, and therefore polarization should occur for all alternatives

SUMMARY AND HYPOTHESES

Research from a variety of disciplines support the notion that deliberative processing has a destabilizing effect on values, attitudes, beliefs, or preferences. Therefore, this study predicted that judgment will lead to more thorough processing than does choice and a concomitant change in value representations. Consistent with previous research on decision process (measured by information search on an information display board; IDB), three hypotheses follow:

Hypothesis 1a. Judgment will lead to more information searched than choice.

Hypothesis 1b. Judgment will lead to less variable patterns of information search than choice.

Hypothesis 1c. Judgment will lead to more interdimensional patterns of information search than choice.

This portion of the study served mainly to replicate Billings and Scherer (1988) with consumer-oriented decision stimuli, and is necessary to document the processes purported to influence value representations.

Change in value representations were measured for information displayed on the IDB with pretest and posttest questionnaires. Two hypotheses follow:

Hypothesis 2a. Judgment will lead to more bidirectional change in value representations than choice.

Hypothesis 2b. Judgment will lead to more polarized change in value representations than choice.

METHODS

Subjects and Procedure

Subjects were 33 female and 31 male undergraduates who received extra course credit for participating in a consumer study on automobiles. All subjects were screened for having been involved in an automobile purchase, and completed the pretest questionnaire regarding attribute importance and feature desirability for automobile alternatives.

Upon arriving at the laboratory one week later, subjects were introduced to a practice IDB, given explanations, and allowed to practice. They were then introduced to the test IDB and instructions were repeated. Experimenters then provided two statements designed to manipulate response mode. Choice was manipulated by instructing subjects to choose one of the eight automobile alternatives. Judgment was manipulated by instructing subjects to judge each of the alternatives on a 7-point scale ranging from extremely positive (1) to extremely negative (7). No time constraints were placed on the subjects, and they were told to examine as much of the information as they liked. These instructions were identical to the ones used by Billings and Scherer (1988). Following the completion of the task, the posttest questionnaire was administered, and subjects were debriefed.

Materials

The IDB consisted of eight used automobile alternatives listed vertically down the left side, and six attributes listed horizontally across the top of the board. The attributes and the features they could take on were as follows: (a) make: Chrysler, Volkswagen, Ford, Chevrolet, Toyota, Honda, Nissan, Subaru; (b) type: sports, compact, mid-sized, full-sized; (c) mileage: 0-1,000, 1,000-10,000, 10,000-25,000, 25,000-50,000; (d) price: $2,500-5,000, $5,000-7,5000, $7,500-10,000, $10,000-above; (e) features: stereo system, digital display instruments, air conditioning, sun-roof; and (f) color: red, blue, black, brown, green, white, orange, silver. Each feature under type, mileage, price and features was presented twice to cover all alternatives.

TABLE 1

MEANS AND STANDARD DEVIATIONS FOR DECISION PROCESS VARIABLES

TABLE 2

MEANS AND STANDARD DEVIATIONS FOR VALUE CHANGE VARIABLES

The pretest and posttest questionnaires were identical and consisted of 38 items (6 attributes and 32 features) responded to on 7-point scales. Attributes were rated on importance, with the endpoints being extremely unimportant (1) and extremely important (7). Features were rated on desirability using a scale from extremely undesirable (1) to extremely desirable (7). Filler items were present so that the questionnaire would appear comprehensive and realistic.

RESULTS

Decision Process

Analysis of variance indicated significant effects for response mode manipulations on amount of information search, F(1, 62)=24.97, p<.001, and variability of information search, F(1, 62)=11.97, p<.001, but not search pattern, F(1, 62)=.04. Inspection of Table 1 indicates that compared to choice, judgment led to more information being disclosed and a less variable search. Though not significant, the search pattern variable was in the expected direction.

A second analysis examined differences in search pattern by treating the first and second halves of the total moves as a repeated measures design. This within-trial analysis revealed no difference by condition (F[1, 62]=.30), but across conditions, the search pattern for the first half was significantly more intradimensional (M= -.398) than the second half (M= .123; F[1, 62]=101.34, p<.001).

Attribute Importance and Feature Desirability Change

Bidirectional. Bidirectional change was measured by computing the average absolute amount of change in attribute importance and adding it to the average absolute amount of change in feature desirability. Analyses indicated significant effects for response mode, F(1, 62)=22.78, p<.001. As predicted subjects in the judgment condition altered their value representations more than did subjects in the choice condition (see Table 2).

Polarization. Polarization was measured by the amount of change that resulted when subjects shifted importance and desirability ratings to a more extreme position (either more negative or more positive) from pretest to posttest. Analysis showed a significant effect for response mode (F[1, 62]=5.15, p< .05), where polarization occurred more in the judgment condition than in the choice condition.

Additional Analysis

The central logic of this paper is that value representations are influenced by thoroughness of information processing. If thoroughness of processing and information search are equivalent, however, then amount of information search should influence attribute importance and feature desirability regardless of response mode condition. Conversely, if the two are not equivalent, then amount of information search should have no impact beyond that of response mode. Finally, information search may relate to depth of processing differently in the two response mode conditions, suggesting an interaction.

To examine these questions, a regression analysis was conducted including amount of information, response mode, and the interaction, and is interpreted in the same manner as a traditional ANOVA. The F-ratios for the effect of response mode on these indexes were, of course, identical to the ones reported previously. Analysis indicated no effect for amount of information (F[2, 61]=.75) or the interaction (F[3, 60]=.66) for polarization. Further, there was no effect of amount of information on bidirectional change (F[2, 61]=.50). However, the effect of the interaction on bidirectional change was only marginally nonsignificant (F[3, 60]=3.09, p=.08). This suggests that amount of information search was of little consequence under choice, whereas under judgment it is positively related to bidirectional change.

DISCUSSION

The central hypotheses of this paper were that response mode affects the value representation of decision alternatives as well as decision process. This study replicated previous results regarding decision process, and further demonstrated that the values attached to product alternatives change under judgment but not under choice.

These decision process results replicate Billings and Scherer (1988) regarding amount and variability of information searched, but not search pattern. This lack of effect was probably due to the larger subject pool employed by Billings and Scherer, where 104 subjects completed eight boards each. However, the current experiment did find that regardless of response mode condition, subjects shifted from intra to interdimensional search patterns. By using consumer products as decision stimuli, this study provides convergent validity for the effect of response mode on decision process.

The most important finding of this paper relates to the change of value representations under judgment but not under choice. The three-phase model of judgment proposed to account for this change is in need of further investigation, but it does possess certain empirical and theoretical advantages. First, because the model is phased, it is testable. For example, judgment and choice both require activation of existing values which results in biased recall and consequently bidirectional value change. However, because the evaluation implicated in judgment is much more pronounced than that of choice, after the decision making process is initiated the two should diverge. Second, the confirmation phase acknowledges the conflictual nature of judgment. Choice between alternatives has long been recognized as inherently conflictual (Janis and Mann 1977), but research on judgmental conflict has focused on resolution within alternatives through compensatory strategies. The current model addresses within alternative conflict (evaluation), but more importantly acknowledges between alternative conflict (confirmation) in judgment. Judgment requires explicit evaluations of alternatives, thereby stating that some alternatives are better than others. The polarization that accompanies the confirmation phase, then, resembles bolstering in that it reaffirms the representations used to render judgments. Whether intended or not, judgments are recommendations for choice and are not devoid of conflict. Third and most important, the model addresses concerns over the static nature of research on accessing, evaluating and resolving information in judgment (Einhorn and Hogarth 1981). This interpretation assumes that decision makers enter decision situations with a constellation of relevant values, and postulates a dynamic process by which decision makers access those values, consider information, and resolve their judgments. In this way, the three-phase model incorporates and compliments research in the information integration tradition.

A related finding is that bidirectional change is associated with increased information search under judgment but not under choice. Although not statistically significant, the general pattern is worth interpreting. This finding is consistent with a multi-phase model of judgment, as more information search means more evaluation, which equates to more change. In contrast, more information search under choice does not mean more evaluation. Rather, it may simply mean that more information was required to identify an appropriate alternative, or to rule out rival ones (Tversky 1972). Hence, even when judgment and choice are similar in terms of information search, they do not influence value representations similarly. Judgment and choice appear to express fundamentally different purposes to the decision maker, which in turn are manifested in the consideration and value representation of decision alternatives.

Implications and Future Directions

Marketing efforts have been largely concerned with changing consumer attitudes or preferences through persuasion and that once changed, these values will extrapolate to consumer decisions. However, the current data suggest that the relationship between values and the actual decision is moderated by the manner in which the decision is made. Specifically, predecisional preferences should predict decisions under conditions where the underlying value representations remain stable (e.g., choice), but the relationship may diminish when conditions promote reevaluation (e.g., judgment).

Various research questions emerge from these findings. If the thorough decision process evoked by judgment accounts for change in value representations, do other factors that evoke thorough decision process also result in value change? To the extent that factors like importance, personal relevance or accountability encourage judgment-like processing, value change should occur (although the type of change should vary according to type of processing evoked). In general, any circumstances that influence the purpose of decision making may also influence the structure of value representation.

Of course, this effort is not without limitations. For example, the model implies choice evokes deliberation similar to yet less-involved than that of judgment. However, in this study choice served mainly as a comparison condition. Further, the three phases-activation, evaluation, and confirmation-while intuitively appealing, are inferred only by outcome measures. Future research may want to employ thought-listing or verbal protocol techniques for corroborative evidence.

CONCLUSION

As decision makers, consumers bring complex structures of values to any purchase decision, and are moved by individual proclivity and task factors to process the relevant information in any number of ways. Thus far, research has assumed that consumers are untouched by the process used to arrive at decisions. This paper contends that consumers are influenced by the path they take to arrive at decisions, and one manifestation of that influence is a change in value representations.

REFERENCES

Bailey, James. R., Robert S. Billings and Michael J. Strube (1989), "The effects of response mode on the organization of memory," paper presented at the 97th meeting of the American Psychological Association, New Orleans, LA.

Billings, Robert S. and Lisa L. Scherer (1988), "The effects of response mode and importance on decision making strategies: Judgment versus choice." Organizational Behavior and Human Decision Processes, 41, 1-19.

Einhorn, Hillel. J. and Robin M. Hogarth (1981), "Behavioral decision theory: Processes of judgment and choice," Annual Review of Psychology, 32, 52-88.

Festinger, Leon (1957), A theory of cognitive dissonance, Standford: Standford University Press.

Fiske, Susan. T. and S. L. Neuberg (1990), "A continuum of impression formation, from category-based to individuating processes: Influences of information and motivation on attention and interpretation," in Advances in Experimental Social Psychology, Vol. 23, ed. Russel Fazio, New York: Academic Press.

Hastie, Reid (1991), "A review from a high place: The field of judgment and decision making as revealed in its current textbooks," Psychological Science, 2, 135-138.

Hastie, Reid and Park, Bernadette (1986), "The relationship between memory and judgment depends on whether the judgment task is memory-based or on-line," Psychological Review, 93, 258-268.

Janis, Irving. L. and Leon Mann (1977), Decision Making: A psychological analysis of conflict, choice and commitment, New York: Free Press

Payne, John. W., James R. Bettman and Eric J. Johnson (1992), "Behavioral decision research: A constructive processing perspective," Annual Review of Psychology, 43, 87-131.

Schkade, David. A. and Eric J. Johnson (1989), "Cognitive processes in preference reversals," Organizational Behavior and Human Decision Processes, 44, 203-231.

Tesser, Abraham (1978), "Self-generated attitude change", in Advances in Experimental Social Psychology, Vol. 11, ed. L. Berkowitz, New York: Academic Press.

Tversky, Amos (1972), "Elimination by aspects: A theory of choice," Psychological Review, 79, 281-299.

Tversky, Amos, Shmuel. Sattath and Paul Slovic (1988), "Contingent weighting in judgment and choice," Psychological Review, 95, 371-384.

Wilson, Timothy D., Dana. S. Dunn, Dolores. Kraft and Douglas. J. Lisle (1989), "Introspection, attitude change, and attitude-behavior consistency: The disruptive effects of explaining why we feel the way we do," in Advances in Experimental Social Psychology, Vol. 22, ed. L. Berkowitz, New York: Academic Press.

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