Disparity Pursuit Theory: the Role of Expectations in Product Choice

Kurt A. Carlson, Cornell University
ABSTRACT - This paper introduces the concept of disparity as a separation of utility between products in consumer choice. Awareness of disparity in a product domain, or disparity expectations, influence how consumers make product choices. Disparity Pursuit Theory (DPT) asserts that consumers engage in disparity pursuit by attempting achieve coherence between expected and perceived disparity. Consumers can bring about this coherence through information distortion, importance weight manipulation, and updating of disparity expectations. Evidence in support of DPT is presented and future work related to DPT is discussed.
[ to cite ]:
Kurt A. Carlson (2000) ,"Disparity Pursuit Theory: the Role of Expectations in Product Choice", in NA - Advances in Consumer Research Volume 27, eds. Stephen J. Hoch and Robert J. Meyer, Provo, UT : Association for Consumer Research, Pages: 301-306.

Advances in Consumer Research Volume 27, 2000      Pages 301-306

DISPARITY PURSUIT THEORY: THE ROLE OF EXPECTATIONS IN PRODUCT CHOICE

Kurt A. Carlson, Cornell University

[I would like to thank Andrew Ainslie, Alice M. Isen, Jeffrey W., Hales, Lisa R. Klein, Margaret G. Meloy, J. Edward Russo, participants in the Predicisional Distortion Research Group at Cornell University, and two anonymous reveiwers for their helpful comments on an early version of this paper.]

ABSTRACT -

This paper introduces the concept of disparity as a separation of utility between products in consumer choice. Awareness of disparity in a product domain, or disparity expectations, influence how consumers make product choices. Disparity Pursuit Theory (DPT) asserts that consumers engage in disparity pursuit by attempting achieve coherence between expected and perceived disparity. Consumers can bring about this coherence through information distortion, importance weight manipulation, and updating of disparity expectations. Evidence in support of DPT is presented and future work related to DPT is discussed.

INTRODUCTION

Sports fans have opinions about the level of parity that exists in certain sports leagues. A league with low disparity (or high parity) is said to be one that is highly competitive, wherein any team can beat any other team on a given day. n contrast, a league with high disparity likely contains one or two teams that are clearly superior and one or two that are clearly inferior to the rest. [Consider the case of baseball. It is well known that huge disparity often exists among Little League teams. However, as one moves to high school, college, and then professional baseball, disparity among teams decreases. For example, consider that the 1998 New York Yankees, the winningest team in major league baseball history, won only 70% of their games.] Just as sports fans are aware of the disparity among teams in a given league, consumers are aware of the disparity among products in a given domain. In the context of consumer products, disparity refers to the difference in utility one believes the alternatives in a set will provide. A consumer’s awareness of disparity in a domain helps her establish expectations regarding the amount of disparity among products drawn from that domain. These domain-specific disparity expectations are updated during the choice process to reflect the expected disparity among the currently considered alternatives. Disparity expectations influence how consumers evaluate product information and justify their decisions.

The remainder of this paper proceeds as follows. A short section introducing the concepts of disparity and expected disparity relates these concepts to other work. Following this is a discussion of how disparity expectations influence consumer behavior in product choice. I then present a brief explanation of why disparity expectations have been understudied. Next, Disparity Pursuit Theory (DPT) is introduced, and evidence in support of it is provided. The paper concludes with a brief discussion of applications and future work related to DPT.

DISPARITY AND EXPECTED DISPARITY

Disparity is both similar to and different from Svenson’s (1992) notion of differentiation in his Diff-Con theory. There are three important similarities. First, both disparity and differentiation refer a degree of separation between alternatives. Second, both can be created during a choice process using predecisional distortion of information or alteration of importance weights. Third, high levels of either should increase confidence in the chosen alternative.

Disparity also differs from differentiation in three important ways. First, disparity can be used to define the degree of separation among any number of alternatives, whereas differentiation is typically used to describe the separation of one alternative from either a set or from one alternative. Along these lines, disparity exists not only for those alternatives currently under consideration but also for those alternatives in much larger sets. Indeed, disparity exists prior to any particular choice, because it is a property of a domain. Second, disparity in a given domain gives rise to expectations of disparity among alternatives drawn from that domain as part of a specific choice. For example, a consumer who believes high disparity exists in a domain should expect a subset of products drawn from that domain (say a consideration set) to be more disparate than a subset of products drawn from a low disparity domain. Third, a low level of disparity can result in high confidence in the chosen alternative because low disparity alternatives are easily substitutable for each other.

One must be careful to resist the temptation to equate disparity with product differences. Table 1 reveals the distinctions between differences and disparity; it is possible to have two different products that are qualitatively similar. Likewise it is possible to have two products that are characteristically similar but are disparate. The disparity a consumer expects for a given domain will be influenced by previous experience with the domain, knowledge of related domains, and information conveyed to the consumer about the domain. Meanwhile, the disparity a consumer perceives between two products will be a function of disparity expectations, preferences, goals, needs, and the information about those two products.

THE ROLE OF DISPARITY EXPECTATIONS IN PRODUCT CHOICE

Dispariy expectations provide the consumer with information about the choice process that is independent of the alternatives under consideration. For example, disparity expectations give the consumer an indication of the difficulty of the choice that is ahead.

The relation between expected disparity and expected difficulty of choice is expected to be an inverted U. To understand why, consider that the difficulty of choice will be function of two factors: the difficulty of identifying the superior alternative and the consequence of mistakenly selecting the inferior alternative. At extremely low levels of expected disparity, the expected consequences of making an incorrect choice are trivial as the alternatives are expected to be interchangeable. At zero expected disparity, the consumer does not expect to be able to discriminate among the products, thus the consumer will likely ignore the goal of selecting the best alternative, instead saving effort by choosing quickly and somewhat randomly. As expected disparity increases, so do the expected consequences of selecting the inferior alternative and, in turn, consumers expect to expend more effort making their selection. Thus, at low levels, increases in expected disparity lead to increases in expected choice difficulty. However, beyond some level of expected disparity, the consumer expects to be able to detect the superior alternative with relative ease and, despite the growing consequences of failing to do so, the high expected disparity drives down the expected effort required. At some (large) level of expected disparity, the consumer expects that the disparity among alternatives will be so large that the choice will be transparent. Though the consequences of failing to detect the superior alternative are extreme in this instance, the likelihood of doing so is near zero and the choice is again expected to take little effort. [Here, unlike for the low expected disparity case, the choice is expected to be simple due to a large amount of expected disparity rather than due to a tiny amount of disparity with no consequences associated with failure. To make a quick note during a telephone conversation, one will grab either the red or black pen, depending on which is closer.]

TABLE 1

THE DISPARITY-DIFFERENCES DISTINCTION

Now consider what disparity expectations reveal about the role of expertise in decision making. Experts in a given domain should be able to distinguish smaller levels of disparity between alternatives. As a consequence, expected difficulty should rise earlier and more quickly (as a function of expected disparity) for experts than for novices. Similarly, the expected difficulty curve for experts should peak and descend at lower levels of expected disparity for experts than for novices. This implies that expected choice difficulty will only be equal for novices and experts at trivial expected disparity levels [Though it is unlikely that novices and experts possess the same expected disparity for a given domain, differing expected disparity levels simply make it more difficult to identify the trivial conditions in which expected difficulty is equal across the two groups. That is, for the non-trivial expected difficulty levels, it will be very difficult to assess the expected disparity and measure the expected difficulty that it translates into for both novices and experts.] (i.e., zero and large levels) and at one non-trivial expected disparity level when the two expected difficulty curves cross. [Note that since expected difficulty is increasing for novices and falling for experts at this intersection, the two groups have different factors dominating their expected difficulty. How exactly this will influence choices is not yet clear.]

To the extent that experts and novices differ in their expectation of a choice task’s difficulty, we might expect the two groups to employ different decision strategies. That is, for a low (non trivial) level of expected disparity, experts might adopt a compensatory strategy to address a choice that is expected to be difficult, while novices employ a heuristic strategy as they expect the alternatives to be interchangeable and thus easy to choose from.

Disparity expectations also provide clues regarding the difficulty consumers will have with justifying their choice. For example, when consumers expect high disparity, they can expect to find no challenge (from others) if they have selected the "best" product, and extreme challenge if they have selected a product that others deem inferior. Along these lines, disparity expectations can signal the expected cost of making a mistake in the choice process; low disparity between products is usually correlated with a low cost of erring.

WHY NOT DISPARITY EXPECTATIONS SOONER?

The concept of expected disparity may have been overlooked in consumer research to date for any number of reasons. Most notable is the emphasis given to the weighted-adding model (WAD) as a prescription for consumer decision making. Under WAD, consumers use importance weights together with unbiased assessment of the product information, to generate a set of product values. Consumers compare these values and select the alternative with the largest value. In such a model, where the product information is not malleable, there is little role for disparity expectations. That is, such expectations are meaningless to the choice process when attributes, as in WAD, are assessed independent of expectations.

Perhaps a role for expectations could have been fashioned in the post-decisional distortion (i.e., dissonance reduction) literature, but it would certainly have been minor as pre-evaluation expectations are not very likely to survive the information evaluation process for a consideration set. Maybe the best opportunity for expectations to have previously played a role would be in the information search and acquisition literature. However, this literature has been heavily influenced by the economics of information paradigm, wherein expectations are commonly used to represent probabilities over outcomes of information search.

Recent work on predecisional distortion (Russo et al. 1996, 1998) has paved the way for disparity expectations in consumer choice processes. By demonstrating that consumers distort information as a means of supporting the currently leading alternative, Russo and colleagues have opened the door to factors (such as expectations) that influence those evaluations.

DISPARITY PURSUIT THEORY

The driving force behind Disparity Pursuit Theory (DPT) is consumers’ desire to achieve coherence between expected and perceived disparity. [Coherence is used throughout to represent a consumer's aligning of perceived and observed disparity.] That is, consumers engage in disparity pursuit. Consumers bring about this coherence through various means: information distortion, importance weight manipulation, and updating of disparity expectations. DPT differs from Svenson’s Diff-Con theory on several accounts, but the chief differences are the allowance for disparity expectations and that disparity achievement need not be the consumer’s sole goal. That is, a consumer may be content to choose quickly, and with little effort, if no disparity is perceived among alternatives under consideration. In what follows, I address each means through which consumers pursue expected disparity in turn.

Distorted Interpretation of Attributes

If we treat all attributes as subject to interpretation, then it becomes clear how information distortion can be used to align expectations and perceptions of disparity. Specifically, a given consumer’s attribute assessment is drawn from a distribution of possible assessments wherein the integrated meaning of each attribute is dependent on a number of context specific factors. These factors include the nature of the task, the goals of the individual, verifiability of attribute claims, attribute order, and the individual’s disparity expectations. When consumers are presented with information that conflicts with her disparity expectations, it may be easier for them to distort the information (draw from one tail of the assessment distribution) than to assess the information without bias (draw from the center of the distribution) and challenge strongly held expectations. Along these lines, a consumer with strongly held disparity expectations might distort product information to bring their disparity observation into coherence with their expected disparity.

Alteration of Importance Weights

A consumer might also adjust the importance given to a particular piece of information in order to bring disparity perceptions into line with disparity expectations. Consider the case of a consumer faced with assessing information about two alternatives that are expected to be highly disparate. If consumers are presented with two attributes on which the alternatives are very different (Attribute D) and very similar (Attriute S), they might increase the weight given to Attribute D and decrease the weight given Attribute S. Doing so brings their disparity perceptions into coherence with their disparity expectations.

Changing Expectations of Disparity

Though consumer decision making often occurs in the absence of external accountability, consumers typically want to select the best alternative. As such, it would be very difficult to argue that consumers achieve coherence between disparity expectations and observations solely through distortion and importance weight modification. Accordingly, the final, and probably most common, means for a consumer to achieve coherence is through updating of disparity expectations. Though one can construct a passive choice task wherein consumers select between two alternatives that are expected to be disparate, typical choice tasks are much more active in that they allow the consumer to determine what alternatives enter the consideration set. While consumers may begin a choice with disparity expectations defined over a given domain, it is likely that these domain disparity expectations are replaced with alternative-specific disparity expectations when the consideration set is well established. As the consideration set generally consists of those alternatives that have survived one or more eliminations, the expected disparity should be lower for those alternatives remaining under consideration than that for the entire domain from which the alternatives were drawn. Further, as information about the alternatives under consideration is evaluated, these alternative-specific disparity expectations are updated, thereby aligning disparity expectations and observation for the alternatives.

Work Related to DPT

The work that is most closely related to DPT is that which deals with separation among alternatives (Tyska 1998, distinctiveness; Oppewal et al. 1994 and Louviere 1984, hierarchical models; Svenson 1996, differentiation and consolidation theory; Johnson and Payne 1985, effort and accuracy; Boulding, Kalra, and Staelin 1997, bayesian updating; Russo, Medvec, and Meloy 1996, predecisional information distortion). However, this work is neither unified nor does it provide for disparity expectations. In fact, to date there is little work on expectations in consumer choice that is not based on Subjective Expected Utility (SEU) theory (see van Raaij 1991 for a review of expectations related research).

EVIDENCE FOR DISPARITY PURSUIT THEORY

While rigorous testing of DPT will likely require a program of research studies (some of which are discussed at the end of this paper), we can assess whether support for the theory currently exists by examining data collected for another purpose. Carlson, Russo, and Meloy (2000) collected intermediate attribute assessments in order to determine the extent to which predecisional distortion could be ameliorated by a warning. Since these data were not collected with testing DPT in mind, statistical inference will be kept to a minimum, focusing rather on an exploratory investigation of the data.

In the aforementioned study, attribute information for two backpacks was presented sequentially to 86 student participants. Participants were given information about six attributes, each of which described both backpacks, and were asked to make a hypothetical product choice between them. All attributes, except Price, were described as a combination of multiple sub-attributes. The two attributes, Price and Bottom of the Backpacks, follow:

Price

Backpack H is priced at $38.95 and Backpack J is priced at $39.95.

Bottom of the Backpacks

Backpack H is described as having a bottom which is double-stitched and double-ined for durability. In addition, the material has been chemically treated to enhance its strength and water repellency while retaining its suppleness. Backpack J is described as having a layer of leather sewn into the bottom of the pack to enhance the integrity of the pack. The leather has been treated with a thin coat of oil for a softer, richer feel. In addition, a layer of water repellent material has been sewn in between the interior layer and the leather exterior.

Measures

After each attribute, three questions were asked. First, participants were instructed to indicate (on a 9-point scale) which of the two backpacks the attribute favored (1="Strongly Favored Backpack H", 9="Strongly Favored Backpack J"). The second question introduced a metaphor that equated the backpacks to horses in an ongoing race and asked participants to identify the current leader. The horserace analogy made clear that the leading alternative could change at any time. A third question was used to allow participants to express their confidence in the identified current leader. After reading all six attributes, participants were asked to choose one of the two backpacks and to make a final indication of their confidence on a 50% to 100% scale. The 86 participants who completed this version of the task are referred to as the standard choice group (SCG).

Controls and Design

An additional 25 participants, the no choice group (NCG), evaluated the same attribute information, but did so outside the context of making a choice. Participants in the NCG evaluated the same attributes under the guise that each attribute represented a different pair of backpacks. Therefore, participants believed they were seeing 12 backpacks (six pairs) rather than two backpacks. Only the first (rating) question was asked of the NCG participants since the leader and confidence questions were meaningless without a cumulated preference. Analysis of the NCG responses indicated that all six attributes slightly favored Backpack H. That is, the mean NCG rating for each attribute fell between 3.96 (Materials) and 4.81 (Price) on the 1-9 scale. Thus, while any one individual in the SCG could have preferences that would deviate from these diagnosticity ratings (as calculated by the NCG evaluations), systematic deviations over many participants cannot be explained solely by preferences over the content of the information. Attributes were presented to both the NCG and the SCG in one of two orders: Price-First, in which Price was the first attribute seen, or Price-Last, which was the reverse order of Price-First.

Choice Proportions

Of the 86 participants in the SCG, 62.8% (54 of 86) chose Backpack H after viewing all the information. Given the minor amount of diagnosticity favoring Backpack H in each attribute, together with numerous opportunities for true preferences to influence an individual’s leaning, it was not surprising that 32 of 86 participants selected Backpack J. However, what is of interest is that the choice proportions were not equal for the SCG Price-First and Price-Last participants. Since all the attributes favored Backpack H, we might expect that primacy would have helped Price-First SCG participants lean toward Backpack H. This leaning, together with the small diagnosticity in each attribute, should (in the absence of DPT) have made Backpack H the landslide winner for the Price-First SCG participants. In fact, the opposite result held. Backpack H was selected 55.6% of the time (25 of 45) in the Price-First order and 70.7% of the time (29 of 41) in the Price-Last order.

While the unequal proportions above are troublesome for most theories of consumer decision making, they are easily handled by DPT. This theory suggests that for the Price-First SCG participants, the $1 gap signaled a low degree of disparity between thee two products. Note that Price (as evaluated by the NCG) slightly favored Backpack H, and that no Price-First SCG participants indicated that Price favored Backpack J, it is clear that Price was not signaling that Backpack J was of higher quality. This expectation of low disparity presumably influenced how Price-First SCG participants interpreted subsequent attributes (i.e., the attributes were perceived as less favorable toward Backpack H when compared to Price-Last SCG participants). [An economist might approach this by noting that both backpacks are currently in the market, so the consumer has expectations about the price quality relationship between the two products. Namely, if either of the two products had a price-quality ratio that was dramatically different from the other, then one of the two products would dominate in the market. However, since both backpacks are available in the market, this cannot be the case and, therefore, the small price difference (of $1 for a $40 backpack) is a signal of low disparity between the backpacks. That is, these two backpacks compete in the same quality category, and likely appear side-by-side on retail shelves. Thus when the price appears first, consumers are less likely to employ product separation techniques such as information distortion and are more likely to be willing to accept either backpack.] Had this experiment been designed to test DPT, a one-sided test for a difference in the choice proportions would have revealed the difference to approach reliability, z=-1.48, p<0.069.

This result has important implications for how information should be revealed to or acquired by consumers. For example, if a manufacturer has a product that is slightly better (relative to a competitor’s offering) on every dimension (as is Backpack H relative to Backpack J), the manufacturer will be best served by withholding attributes that might signal low disparity till late in the evaluation process. Revealing an attribute that signals low-disparity too early in the information evaluation process can have a big impact on the (subjective) interpretation of subsequent information, thereby influencing consumers’ ability to create disparity between products.

Attribute Variability

If, as predicted by DPT, Price signaled low disparity between the backpacks, there should be evidence of less variance in the attribute evaluations of the Price-First participants, compared to the Price-Last participants. In fact, average variance was much lower for the Price-First SCG participants (2.76) than for the Price-Last SCG participants (4.10). In addition, comparison of the mean and variance of the Price attribute across the Price-First SCG (mean=4.91, var=0.31) and the Price-Last SCG (mean=4.0, var=3.65) participants reveals two very interesting differences. First, as DPT would suggest, Price is a much more subjective attribute (has higher variance) when it appears in the last position than when it appears in the first position. Given the concomitant change in choice under the two orders, one can interpret this as reflecting an ability of participants to evaluate Price more subjectively when it appears last than when in appears first. Second, the evaluation of Price is more favorable to Backpack H when it follows the other five attributes (all of which slightly favor Backpack H) than when it appears first. This suggests that participants interpreted the $1 difference as favoring Backpack H to a greater degree when the Price attribute was preceded by a developing preference for Backpack H.

Predecisional Information Distortion

The above is consistent with DPT, but it offers no insight as to how consumers distort information or re-weight attributes in order to bring disparity observations into coherence with disparity expectations. In this example, DPT predicts that Price-First participants will expect less disparity between the backpacks, and therefore, will have less need to create disparity between them (via information distortion) during the choice process. To consider how consumers achieve coherence through information distortion, the process tracing and predecisional information distortion techniques developed by Russo et al. (1996, 1998) were employed.

Predecisional information distortion was measured across both orders, yielding predecisional information distortion of 0.20 for the 41 Price-First SCG participants and distortion of 0.60 for the 45 Price-Last participants. This difference in distortion is exactly what we would predict if the Price attribute reflected an expectation of low disparity between the backpacks. The conclusion one reaches in light of DPT is that there is less reason for participants to distort the information when expectations of low disparity between the two products have been signaled by a small price difference. [Unfortunately, the data do not permit analysis of the re-weighting contention forwarded by DPT.]

Observed Disparity

If disparity expectations influence observed disparity between alternatives, one might assess the level of disparity between two alternatives by choice confidence, choice postponement rates, and willingness to pay measures. Using the Carlson et al. (2000) data, the influence that disparity expectations had on final choice confidence was examined. Note that low confidence suggests relatively low disparity between alternatives. As DPT predicted, Price-First SCG participants were less confident (6.76 out of 10) in their choice than Price-Last SCG participants (7.28).

THE ROLE OF DPT IN CHOICES

To choose one alternative over another implies that one has discovered some disparity between the alternatives (or has flipped a coin to decide). Presumably, given the commitment that people express over their choices, most consumers have indeed perceived disparity among alternatives at the time of the choice. This disparity may be enhanced by post-choice cognitive dissonance reduction. However, even when it is created or discovered pre-choice it may serve a useful purpose to the consumer, namely, to help establish coherence between expected and perceived disparity.

Though the role of disparity expectations in choice seems ideally suited for the evaluation component of consumer choice, disparity expectations might also play a role in the alternatives we accept for consideration. That is, the number of options in the consideration set may be altered to reduce or increase disparity, thereby making the choice less consequential or more obvious, respectively. An example of the former occurs when a consumer intentionally invites a new and clearly inferior alternative into the consideration set as a means of bolstering support for the current leader. An example of the latter case might be the typical brand line in which manufacturers offer a range of high brands (i.e., brand lines where the premium product dominates the second tier brand that dominates the third tier brand and so on).

APPLICATIONS OF DPT

This section provides a broad overview of several areas in which disparity, disparity expectations, and disparity pursuit theory might have application. The few areas discussed here are not intended as a comprehensive list of the potential applications of disparity, but rather as an introduction to the potential usefulness of these concepts.

Niches and Advertising

Disparity and expected disparity might be helpful concepts for explaining market niches and understanding the benefits of advertising. A consumer’s expected disparity is presumed to differ across domains. While factors such as expertise likely moderate expected disparity, the underlying actual (i.e., true) disparity in a domain presumably accounts for much of the disparity that consumers expect. Some product categories (or domains) have successfully differentiated themselves from their larger parent category and, in doing so, have revealed themselves as disparate from it (e.g., sport utility vehicles and small trucks from the more general domain of trucks). The oddity here is that the alternatives in the new domain, such as sport utility vehicles, now likely suffer from a lack of disparity among each other. In such cases, attempts to create disparity often emphasize previously trivial distinctions (e.g., the new Nissan Exterra touts a built-in first aid kit and an aluminum roof rack as distinct features). It is likely that advertising and brand loyalty play a role in attracting and holding customers to a particular alternative in these (newly constructed) low disparity domains.

Brand Extensions

Consider for a moment what the typical computer manufacturer’s line looks like. The top-of-the-line computer has all the best or tied-for-best features among the features possessed by the computers in the manufacturer’s line. Likewise, the bottom line computer has all the worst or tied-for-worst features. This setup maximizes disparity among the products in a manufacturer’s product line. The purpose of this disparity creation is to make the choice as easy as possible once the consumer has selected a brand. That is, within the brand line, disparity among the alternatives is much larger than disparity across brands, and even though a price-quality trade-off remains to be made, this is likely easier than choosing from a low disparity brand line.

Independence of Irrelevant Alternatives

DPT might have implications for the literature on independence of irrelevant alternatives (Huber et al. 1982). In DPT no alternative is truly irrelevant nor independent since every alternative provides information about the disparity among the alternatives in the set. For example, the disparity information provided by an alternative added to the choice set can have two clear impacts. First, introduction of another alternative can either increase or decrease the disparity in a set. A case of the latter is when a new product is introduced that is qualitatively orthogonal and relatively equal (in value) to the currently considered alternatives. A case of the former is when the new alternative is distinctly inferior in quality and very different in its features, relative to the currently considered alternatives. Second, addition of an alternative to the choice set can signal the relative importance of the attributes over which the disparity and the qualitative judgements are assessed. This is reflected in the consumer’s desire to avoid difficult choices and thus to pursue alternatives and dimensions along which choices are easier.

FUTURE WORK

To date, no published work has empirically or experimentally explored the role of disparity expectations on disparity pursuit. As a starting point, one might assess whether consumers actually have and use disparity expectations in product choices. To do so, participants might be asked to provide an assessment of disparity for several domains. The sample could then be parsed into two groups on the basis of their having either high or low disparity expectations for a given domain. After some delay (say a week or longer) participants could be given a choice task in which they make intermediate attribute evaluations. Support for DPT would come in the form of high-disparity participants committing more predecisional distortion, re-weighting importance weights to a greater degree, and achieving higher choice confidence, relative to low-disparity participants.

In work currently in progress (Carlson 2000), I manipulated expected disparity between two alternatives in a passive choice task as a means of conducting a clean test of DPT. One group (low disparity) was told in advance that both of the products were rated well by an external source, while the other group (high disparity) was told that only one of the products was rated well. These two groups’ attribute evaluations will be used to determine whether the attributes for the low-disparity group are distorted less and perceived as less subjective (i.e., less variable). Participants’ decisions, choice confidence, and attribute importance weights will be used to reveal perceived disparity and the ability of participants to re-weight the attributes to achieve separation. Finally, the study will reveal whether disparity expectations can be created external to the choice process (via a cover story) as well as internal to the choice process (via key attributes such as Price and overall Ratings) as suggested b the data presented above.

REFERENCES

Boulding, William, Ajay Kalra, and Richard Staelin (1999), "The Quality Double Whammy," Marketing Science (forthcoming).

Carlson, Kurt A. (1999). "The Impact of Expected Disparity on Consumer Decisions," Working Paper, Cornell University.

Carlson, Kurt A. and Lisa R. Klein (1999). "Predecisional Distortion in Real Product Choice: the Role of User Control in an Interactive Multimedia Environment," Working Paper, Cornell University.

Carlson, Kurt A., J. Edward Russo, and Margaret G. Meloy (2000). "The Impact of Warnings on the Reduction of Predecisional Distortion," (in preparation).

Huber, Joel, John W. Payne, and Christopher P. Puto (1982), "Adding Asymmetrically Dominated Alternatives: Vialoations of Regularity and the Similarity Hypothesis," Journal of Consumer Research, 9 (June), 90-98.

Johnson, Eric J. and John W. Payne (1985), "Effort and Accuracy in Choice," Management Science, 31(4), 395-414.

Oppewal, Harmen, Jourdan J. Louviere, and Harry J.P. Timmermans (1994), "Modeling Hierarchical Conjoint Processes With Integrated Choice Experiments," Journal of Marketing Research, 31 (February), 92-105.

Raaij, W. Fred van (1991), "The Formation and Use of Expectations in Consumer Decision Making," in Handbook of Consumer Behavior, T.S. Robertson and H. H. Kassarjian, Prentice-Hall, New Jersey

Russo, J. Edward, Victoria H. Medvec, and Margaret G. Meloy (1996), "The Distortion of Information During Decisions," Organizational Behavior and Human Decision Processes, 66(1), 102-110.

Russo, J. Edward, Margaret G. Meloy, and Victoria H. Medvec (1998), "Predecisional Distortion of Product Information," Journal of Marketing Research, 35 (November), 438-452.

Svenson, Ola (1992), "Differentiation and Consolidation Theory of Human Decision Making Decision Making." Acta Psychologica, 80, 143-168.

Svenson, Ola (1996), "Decision Making and the Search for Fundamental Psychological Regularities: What Can be Learned from a Process Perspective?" Organizational Behavior and Human Decision Processes, 65(3), 252-67.

Tyszka, Tadeusz (1998), "Two Pairs of Conflicting Motives in Decision Making," Organizational Behavior and Human Decision Processes, 74(3), 189-211.

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