Some New Light on Substitution and Attraction Effects

ABSTRACT - An experiment was conducted to examine the mediation of substitution and attraction effects (Huber, Payne, and Puto 1982). Following Farquhar and Pratkanis (1987), it was hypothesized that assimilation in judgment would lead to substitution in choice and that contrast in judgment would lead to attraction in choice. Although context effects were found in both judgment and choice, the judgment-choice relationship was rather weak (r = .29, p < .03). The results are interpreted in terms of Tversky et al.'s (1988) compatibility principle.

Citation:

Frank R. Kardes, Paul M. Herr, and Deborah Marlino (1989) ,"Some New Light on Substitution and Attraction Effects", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 203-208.

Advances in Consumer Research Volume 16, 1989      Pages 203-208

SOME NEW LIGHT ON SUBSTITUTION AND ATTRACTION EFFECTS

Frank R. Kardes, Massachusetts Institute of Technology

Paul M. Herr, Indiana University

Deborah Marlino, Massachusetts Institute of Technology

ABSTRACT -

An experiment was conducted to examine the mediation of substitution and attraction effects (Huber, Payne, and Puto 1982). Following Farquhar and Pratkanis (1987), it was hypothesized that assimilation in judgment would lead to substitution in choice and that contrast in judgment would lead to attraction in choice. Although context effects were found in both judgment and choice, the judgment-choice relationship was rather weak (r = .29, p < .03). The results are interpreted in terms of Tversky et al.'s (1988) compatibility principle.

INTRODUCTION

A great deal of research has shown that the manner in which consumers evaluate a target brand depends upon what background or contextual brands are salient at the time of brand choice (Burton and Zinkhan 1987; Huber, Payne, and Puto 1982; Huber and Puto 1983; Ratneshwar, Shocker, and Stewart 1987). For example, consider the case in which one brand dominates another on one dimension but is dominated on a second equally important dimension. When a third, "decoy" brand is added to the choice set, the choice probabilities for the two original brands change. Sometimes the decoy takes disproportionately more share from the similar (target) than from the dissimilar (competitor) brand, and sometimes the decoy inc eases the probability of choosing the target brand. These effects are known as substitution and attraction effects, respectively. The purpose of the present study is to examine the mediation of these effects.

Social judgment theory (Sherif and Hovland 1961; Sherif, Sherif, and Nebergall 1965) suggests one plausible explanation for substitution and attraction effects (see Farquhar and Pratkanis 1987). If a decoy serves as a reference point for judging a target brand, then judgments of the target should be displaced toward the decoy (assimilation) when the decoy is positioned near the target and judgments of the target should be displaced away from the decoy (contrast) when the decoy is positioned far away from the target. Hence, similar brands are seen as more similar and dissimilar brands are seen as more dissimilar than they actually are. As a consequence, a moderately discrepant inferior decoy may make the target appear less attractive (assimilation/substitution) and an extremely discrepant inferior decoy may make the target appear more attractive by comparison (contrast/attraction).

METHOD

To test the assimilation/contrast view, 84 undergraduates were randomly assigned to one of four conditions: Moderately discrepant decoy, extremely discrepant decoy, No decoy, and Target brand only conditions. Brand descriptions for four product categories (beer, cars, restaurants, ar,d health insurance; order of presentation was counterbalanced) were adapted from Huber et al. (1982) and Huber and Puto (1983). Unlike previous studies, the competitor always dominated the target on price and the target always dominated the competitor on a specific quality dimension (see Appendix). The quality ratings were provided on a scale ranging from 0 to 100. The decoy was always priced slightly lower than the target. In addition, the decoy was slightly lower in quality than the target brand in moderately discrepant decoy conditions and much lower (almost as low as the competitor) in extremely discrepant decoy conditions. The position of the decoy was manipulated on the quality dimension instead of the price dimension because previous research has shown that context effects are more pronounced on ambiguous than on nonambiguous dimensions (Herr 1986; Herr, Sherman, and Fazio 1983).

Another aspect of the present study that differs from previous research is that attitudinal judgments served as a primary dependent measure. Judgments were measured on three semantic differential scales ranging from 0 to 10 (favorable, satisfactory. pleasant). In addition, subjects were asked to choose one brand from each product category and process measures (prototypicality ratings) were taken for two of the product categories, beer and cars. Subjects were asked to "imagine a brand of beer (car) that is average in price, average in taste quality (reliability) and average on all dimensions" and they were asked to rate the degree of similarity (0 - 10) between this prototypical brand and the target brand.

RESULTS AND DISCUSSION

Subjects' ratings of the target brand, the decoy, and the competitor were averaged across attitude scales (favorable, satisfactory, pleasant) and across product categories (beer, cars, restaurants, health insurance packages) to provide product evaluation indices for the target, the competitor, and the decoy. These indices were internally consistent. The Cronbach's alpha values were .82, .9(), and .81 for the target, the decoy, and the competitor, respectively.

Attitudinal Judgments

Overall evaluations of the target brand, the decoy, and the competitor as a function of the four experimental conditions are presented in Table 1. As Table 1 indicates, more favorable evaluations of the target were formed when the competitor was present (M = 6.52) as opposed to absent (M = 5.27), t (40) = 3.55, p c .001 (two-tailed tests were used in all analyses). This pattern indicates that judgments of the target were contrasted away from judgments of the dissimilar competitor, as predicted.

TABLE 1

BRAND ATTITUDES

Although the target was rated more favorably than the decoy across conditions (p's < .01), the mean difference between evaluations of the target and the decoy was greater in extremely discrepant (difference = 1.75) than in moderately discrepant (difference = .85) decoy conditions, t (40) = 2.28, p < .03. Hence, the discrepancy manipulation was effective.

It was predicted that judgments of the target should be displaced toward the decoy in moderately discrepant decoy conditions. Hence, less favorable evaluations of the target should be formed when a moderately discrepant inferior decoy is present as opposed to absent. However, more favorable evaluations of the target were formed in moderately discrepant decoy (M = 7.31) than in no decoy (M = 6.52) conditions, 1(40) = 2.42, p < .02. Thus, a contrast effect was observed when an assimilation effect was anticipated. This finding is rather surprising given that (a) the discrepancy manipulation was effective and that moderate discrepancy should lead to assimilation (Herr et al. 1983; Manis, Nelson, and Shedler 1988; Sherif and Hovland 1961); (b) the contextual stimuli and the target stimulus were presented simultaneously (as opposed to separating the induction and test phases), which should increase the likelihood of assimilation (Manis, Paskewitz, and Cotler 1986); and (c) the brands were differentiated on an ambiguous dimension, which should also increase the likelihood of assimilation (Herr 1986, Herr et al. 1983).

Manis and his colleagues (Manis and Paskewitz 1984a, 1984b, Manis et al. 1986) have also obtained contrast effects in conditions under which assimilation was expected. Manis et al. (1986, 1988) suggested that judgment tasks may elicit assimilation and contrast simultaneously and that observed results depend on the relative strength of these opposing forces. Assimilation should be enhanced by increasing the salience of a moderately discrepant reference point. If subjects eliminated the decoy from their consideration set quickly in the present study, then their evaluations of the decoy may not have been sufficiently salient to permit assimilation to occur. Indeed, the decoy may not have even served as a standard of comparison. This is a strong possibility given that the stimuli used as reference points are known to vary across individuals and across situations (Fazio 1979, Higgins and Stangor 1988, Kahneman and Miller 1986, Puto 1988, Srull and Gaelick 1983). Subjects may have (a) abstracted a reference point by integrating information about the decoy and the competitor, (b) used an internal reference point based upon prior knowledge and experience, or (c) they may have combined these two comparison strategies in some way.

It was predicted that judgments of the target should be displaced away from the decoy in extremely discrepant decoy conditions. Hence, more favorable evaluations of the target should be formed when the extremely discrepant decoy is present as opposed to absent. However, evaluations of the target brand did not differ as a function of the presence or absence o' the extremely discrepant decoy (M's = 6 62 vs 6 52. p > .20). Perhaps the decoy was too discrepant to be used as a reference point for judging the target brand. If the target and the decoy were perceived as members of two different sub-categories, the decoy should not be employed as a standard for judging the target.

Choice

Choice probabilities for the target brand, the decoy, and the competitor as a function of the type of decoy (moderately discrepant decoy, extremely discrepant decoy, no decoy) are presented in Table 2. The results indicated that a smaller proportion of subjects chose the target when the moderately discrepant decoy v.as present as opposed to absent (.52 vs. .67), chi-square = 3.96, p < .05. Thus, a reliable substitution effect was obtained.

No attraction effect was found across product categories (.69 vs .67, p > .20). However, in the car category, a greater proportion of subjects tended to choose the target when the extremely discrepant decoy was present as opposed to absent (.86 vs. .76, p < .10). Hence, a weak attraction effect was found with the car stimuli.

TABLE 2

BRAND CHOICE

The Judgment-Choice Relation

A choice index was computed to enable us to assess the degree of correlation between Judgment and choice. If a subject chose the target in all four product categories, a score of four was assigned. If a subject chose the target in three categories, a score of three was assigned, etc. Hence, the choice index ranged from 0 to 4. A weak but statistically reliable relationship was found between judgment and choice, r(61) = .29, p < .03.

This weak relationship is father surprising given that the product stimuli and the judgment and choice measures were included on the same page of the questionnaire . Although subjects' evaluative judgments were externally available and salient at the time of brand choice, judgment accounted for less than 9% of the variance in choice.

To recapitulate, it was predicted that assimilation effects in judgment should lead to substitution effects in choice and that contrast effects in judgment should lead to attraction effects in choice. However, in conditions under which assimilation effects were expected, contrast effects were observed in judgment and substitution effects were found in choice. Further, in conditions under which contrast effects were expected, neither contrast nor attraction was observed. Although context effects were obtained rather easily, the direction of these effects was difficult to predict.

Prototypicality Judgments

Prototypicality judgments were assessed to determine whether the judgmental context effects observed involved a change in perception (i.e., a "true" attitude change) or a change in the way subjects responded to the measurement instrument (i.e., a response effect). Some theorists maintain that context effects reflect a change in the way an attitude object or issue is perceived (Helson 1964, Sherif and Hovland 1961), whereas others argue that context effects represent a change in the way an attitude is expressed rather than a change in the underlying attitude itself (Ostrom and Upshaw 1969, Upshaw 1984). We treat this controversy as an empirical issue. Context effects are sometimes perceptual and sometimes linguistic in nature. The boundary conditions are not well understood.

Both direct and indirect measures should be provided to separate perceptual versus response language effects. A perceptual change should have several important repercussions. For example, a change in the manner in which an object is perceived should influence the way the object is perceived relative to other objects and should affect subsequent overt behavior toward the object. Conversely, response effects should be limited to a given response scale and should not generalize to measures of related constructs (however, response effects can influence subsequent overt behavior if the response is sufficiently salient; see Sherman, Ahlm, Berman, and Lynn 1978).

Similarity judgments between the target brand and the prototypical brand as a function of the four experimental conditions are presented in Table 3. A one-way analysis of variance revealed that subjects tended to judge the target beer as more similar to the prototype in the moderately discrepant decoy condition than in the remaining conditions, F(3. ,95 = 1.96, p = .13. A significant main effect was found on prototypicality judgments of the target car, Ff 3, 79) = 4.08, p < .01. Least significant difference tests indicated that the target car was rated as more prototypical in the moderately discrepant decoy condition than in the extremely discrepant decoy a-d no decoy conditions (p's < .05). Thus, the target brand is perceived as more prototypical when a similarly-positioned brand is salient. Further, recall that response effects should be localized to a single response scale. Because the contextual brands influenced evaluations of the target and perceptions of the target relative to the prototype, the observed judgmental context effects were at least partially influenced by perceptual changes.

TABLE 3

SIMILARITY TO PROTOTYPE

APPENDIX

PRODUCT STIMULI

CONCLUSION

The most striking finding of the present experiment is that context effects are easy to obtain but the direction of these effects is difficult to predict. Although context effects in both judgment and choice were observed, only a weak correspondence between judgment-specific and choice-specific context effects was found.

One fruitful direction for future research on the mediation of substitution and attraction effects involves focusing on specific judgments rather than on global product evaluations. Farquhar and Pratkanis (1987) and Simonson (1987) found that importance weights on a focal dimension are higher when a decoy is moderately as opposed to extremely discrepant and these importance weights influence subsequent choice. The prototypicality-judgment methodology outlined in the present paper should be useful for determining when shifts in importance weights reflect a change in perception as opposed to measurement reactivity.

Interestingly, Tversky, Sattah, and Slovic (1988) have also focused on the role of importance weights in judgment and choice, but Tversky et al. were concerned with the judgment-choice relation rather than with substitution and attraction effects (although we believe that these two issues are related). They found that the more important of two dimensions is weighted more heavily in choice than in judgment. Hence, choice is more lexicographic than judgment. Tversky et al. argue that choice involves the ordering of dimensions and alternatives, whereas judgment involves the integration of information across dimensions within alternatives. Consequently, qualitative information is weighted more heavily in choice and quantitative information is weighted more heavily in judgment.

When stimulus information is not compatible with the response (i.e., when quantitative information is used in a choice task or when qualitative information is used in a judgment task), stimulus information must be translated into compatible information. Unfortunately, this cognitively effortful translation process decreases the diagnosticity of the stimulus information and increases the likelihood of bias and error. Thus, preferences are likely to be particularly unstable when consumers are confronted with a choice task involving quantitative information pertaining to two alternatives that dominate the other on one of two equally important dimensions. This is the typical structure of the choice task in studies of substitution and attraction effects.

REFERENCES

Burton, Scot and George M. Zinkhan (1987), "Changes in Consumer Choice: Further Investigation of Similarity and Attraction Effects," Psychology & Marketing, 4 (Fall), 255-265.

Farquhar, Peter H. and Anthony R. Pratkanis (1981), "Phantom Choices: The Effects of Unavailable Alternatives on Decision Making," Decision Research Program Technical Report 87-2, Carnegie-Mellon University, Pittsburgh, PA.

Fazio, Russell H. (1979), "Motives for Social Comparison: The Construction-Validation Distinction," Journal of Personality and Social Psychology, 37 (October), 1683-1698.

Helson, Harry (1964), Adaptation-Level Theory. An Experimental and Systematic Approach to Behavior, New York: Harper & Row.

Herr, Paul M. (1986), "Consequences of Priming: Judgment and Behavior," Journal of Personality and Social Psychology, 51 (December), 11061115.

Herr, Paul M., Steven J. Sherman, and Russell H. Fazio (1983), "On the Consequences of Priming: Assimilation and Contrast Effects," Journal of Experimental Social Psychology, 19 (July), 323340.

Higgins, E. Tory and Charles Stangor (1988), "A 'Change-of-Standard' Perspective on the Relations Among Context, Judgment, and Memory," Journal of Personality and Social Psychology, 54 (February). 181-192.

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

Huber, Joel and Christopher Puto (1983), "Market Boundaries and Product Choice: Illustrating Attraction and Substitution Effects," Journal of Consumer Research, 10 (June), 31-44.

Kahneman, Daniel and Dale T. Miller (1986), "Norm Theory: Comparing Reality to Its Alternatives,' Psychological Review, 93 (April), 136-153.

Manis, Melvin, Thomas E. Nelson, and Jonathan Shedler (1988), "Stereotypes and Social Judgment: Extremity, Assimilation, and Contrast," Journal of Personality and Social Psychology, 55 (July), 2836.

Manis, Melvin and Joan R. Paskewitz (1984a), "Specificity and Contrast Effects: Judgments of Psychopathology," Journal of Experimental Social Psychology, 20 (May), 217-230.

Manis, Melvin and Joan R. Paskewitz (1984b), "Judging Psychopathology: Expectation and Contrast," Journal of Experimental Social Psychology, 20 (July), 363-381.

Manis, Melvin, Joan Paskewitz, and Scott Cotler (1986), "Stereotypes and Social Judgment," Journal of Personality and Social Psychology, 50 (March), 461 -473.

Ostrom, Thomas M. and Harry S. Upshaw (1968), "Psychological Perspective and Attitude Change,' in Psychological Foundations of Attitudes, eds. Anthony G. Greenwald, Timothy C. Brock, an Thomas M. Ostrom, New York. Academic Press, 217-242.

Puto, Christopher P. (1987), "The Framing of Buying Decisions," Journal of Consumer Research, 14 (December), 301-315.

Ratneshwar, Srinivasan, Allan D. Shocker, end David W. Stewart (1987), "Toward Understanding the Attraction Effect: The Implications of Product Stimulus Meaningfulness and Familiarity," Journal of Consumer Research, 13 (March), 520-533.

Sherif, Carolyn W., Muzafer Sherif, and Roger E. Nebergall (1965), Attitude and Attitude Change. The Social Judgment-Involvement Approach, Philadelphia: W. B. Saunders Company.

Sherif, Muzafer and Carl I. Hovland (1961), Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Change, New Haven, CT: Yale University Press.

Sherman, Steven J., Karin Ahlm, Leonard Berman, and Steven Lynn (1978), "Contrast Effects and Their Relationship to Subsequent Behavior," Journal of Experimental Social Psychology, 1A (July), 340350.

Simonson, Itamar (19873, "Justification Processes in Consumer Choice," Working Paper, University of California, Berkeley.

Srull, Thomas K. and Lisa Gaelick (1983), "General Principles and Individual Differences in the Self as a Habitual Reference Point: An Examination of Self-Other Judgments of Similarity," Social Cognition, 2, 108- 121.

Tversky, Amos, Shmuel Sattah, and Paul Slovic (1988), "Contingent Weighting in Judgment and Choice," Psychological Review, 95 (July), 371384.

Upshaw, Harry S. (1984), "Output Processes in Judgment," in Handbook of Social Cognition, eds. Robert S. Wyer and Thomas K. Srull, Hillsdale, NJ: Lawrence Erlbaum Associates. 237-256.Table 1

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