The Negative Attraction Effect? a Study of the Attraction Effect Under Judgment and Choice

Jennifer Aaker, University of California at Berkeley
ABSTRACT - This study focuses on the interaction between judgment vs. choice and the impact of justification on the attraction effect, which refers to the increase in the choice probability of the alternative most similar to a newly added inferior alternative brand in a choice set. One unexpected result, a strong negative attraction effect that appeared in the choice/low justification cell, could be caused by the appeal of a "black sheep" alternative. A second unexpected finding, the appearance of the attraction effect in the judgment/high justification cell, may be due to the force of the justification manipulation which alters the judgment task to one of choice.
[ to cite ]:
Jennifer Aaker (1991) ,"The Negative Attraction Effect? a Study of the Attraction Effect Under Judgment and Choice", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 462-469.

Advances in Consumer Research Volume 18, 1991      Pages 462-469



Jennifer Aaker, University of California at Berkeley

[The author wishes to acknowledge the helpful comments and assistance from Professor Philip Tetlock and Itamar Simonson.]


This study focuses on the interaction between judgment vs. choice and the impact of justification on the attraction effect, which refers to the increase in the choice probability of the alternative most similar to a newly added inferior alternative brand in a choice set. One unexpected result, a strong negative attraction effect that appeared in the choice/low justification cell, could be caused by the appeal of a "black sheep" alternative. A second unexpected finding, the appearance of the attraction effect in the judgment/high justification cell, may be due to the force of the justification manipulation which alters the judgment task to one of choice.

Choice modeling has stimulated considerable research in the fields of consumer behavior and psychology (Luce, 1959; Tversky, 1972; Payne, 1982). One issue in marketing focuses on how the addition of an inferior brand into a two brand set affects choice. The conflicting predictions that can be hypothesized are based upon: the principle of regularity (Luce, 1959), the attraction effect (Huber et al., 1982), and a "negative attraction effect".

The principle of regularity, assumed by most choice models (Luce, 1977), indicates that the addition of a new object into a two object choice set cannot change the relative attractiveness of the two original objects (1959). It is based upon the Luce axiom that the choices are made by adding up the utility of the attributes.

A context effect violating regularity, the attraction effect, demonstrates that Luce's axiom does not apply in certain instances (Payne, 1982). The attraction effect refers to the increase in choice probability of the alternative most similar to a newly added inferior alternative brand in the original choice set (Huber, Payne and Puto, 1982; Huber and Puto, 1983; Ratneshwar, Shocker and Stewart, 1987). Consider a set of two equally-valued brands; Brand A is superior or "dominant" in one dimension (i.e. cost), while Brand B is superior or "dominant" in a second dimension (i.e. quality). By adding a third "asymmetrically inferior" alternative C (closest to alternative B) to the choice set, the choice probability of B increases (Huber et al., 1982). (An alternative is "asymmetrically inferior" if it is dominated by one alternative in the set, but not by the second alternative).

The choice literature has explored regularity vs. the attraction effect. However, the negative attraction effect, where an asymmetrically inferior alternative actually decreases the attractiveness of those with which it is most similar has been left out of the literature--possibly because the attractiveness of the dissimilar or "black sheep" alternative always decreases or because no choice situations have been tested in which the negative attraction effect occurs. In the study to be described a negative attraction effect was found in one of the cells. We will suggest reasons why it emerged.


One purpose is to focus on the two decision making tasks, judgment and choice, to determine if the attraction effect (or negative attraction effect) occurs in judgment as it does in choice. The second purpose is to explore the concept of the justification effect in another context from that which was used in Simonson's study (1989).

The Justification Explanation of the Attraction Effect

One partial explanation for the attraction effect is that people select the alternative that is easiest to justify to themselves and others (Simonson, 1989). When an asymmetrically inferior alternative is added to a choice set, the asymmetrically dominating alternative becomes easier to justify--hence the attraction effect.

Simonson (1989) supports the justification hypothesis by showing that when subjects are asked to justify their decision to an authority or reference group (a condition of high justification), the attraction effect is large. In contrast, in low justification conditions, the attraction effect is much smaller if it exists at all (Ratneshwar, Shocker et al., 1987; Huber, Payne and Puto, 1982). In the present paper, the levels of justification (low and high) will both be studied in order to examine the effect of justification on the decision making process in another context.

Judgment vs. Choice

Past research in decision making in general and the attraction effect in particular has focused upon choice (i.e. one alternative is selected from the choice set) rather than judgment (i.e. all alternatives are evaluated using a seven point scale.) In fact, the question of what happens under judgment has not really been considered. Many researchers have used the terms, judgment and choice interchangeably (Slovic and Lichtenstein 1971). Only recently have the two response modes been distinguished (e.g., Bettman, 1982; Einhorn & Hogarth, 1981; Payne, 1982).

Different decision-making strategies, non-compensatory and compensatory, have been associated with the two response tasks (Einhorn & Hogarth, 1981). Under choice, a non-compensatory decision making strategy is more likely to be used (Johnson & Russo, 1984), where less information is utilized (Billings and Sherer 1988). For example, given a choice the respondent may simply eliminate an alternative based on one attribute. Therefore, if one alternative appears to stand out or be in an attractive position based on an attribute, the choice will be easily made.

Under judgment, a more compensatory process is more likely be used which involves more deliberative thinking, time and effort (Einhorn et al. 1979). Therefore, when asked to rate (a judgment task) the alternatives, the respondent may make the initial assessment of the alternatives (with respect to one attribute and perhaps their relationship to each other), but then s/he may continue to consider each alternative individually, weighing each of the alternatives respective attributes against each other. The compensatory process involves more effort to evaluate all the available information and rank order the alternatives. Therefore, the attraction effect bias (which emerges using a non-compensatory strategy-elicited under choice) should be lessened using a compensatory strategy--elicited under judgment. Thus, the attraction effect should differ under judgment and choice, especially in the high justification cells where DMs (decision makers) feel more accountable for their decisions. In low justification, Simonson has shown that little, if any, attraction effect occurs even under choice (1989). Therefore, logically there should be no real attraction effect under judgment in low justification.

The hypotheses follow:

1a. Under a choice task, the attraction effect will emerge in a high justification condition.

1b. Under a choice task, a small attraction effect will appear in a low justification condition

2a. Under a judgment task, the attraction effect will not emerge in a high justification condition.

2b. Under a judgment task, the attraction effect will not appear in a low justification condition.


Subjects and Design

Undergraduate psychology students (N=450) at the University of California at Berkeley were asked to participate in the experiment for class credit. The response tasks, judgment vs. choice, were within-subject, whereas the two other variables, justification level and number of people in the choice set, were between-subject. Approximately an equal number of males and females were in four cells; the two conditions of justification (low vs. high) and two vs. three people in the choice set (Person A, Person B, with or without Person C). The two person choice set was included as a control set which will be the basis for comparisons of the three person choice set.


The context of people evaluation to test the attraction effect was selected for three reasons. First, Ratneshwar, Shocker and Stewart (1987) suggest that the subject may perceive the "microworld" of products used in attraction effect studies differently from the realistic world of brands that they confront everyday in the market place. Most products have brand names which carry associations and attitudes. Therefore, it is suggested that the decision making process using products without brands is relatively unrealistic. In an evaluation context where brands are not involved, this problem does not arise.

Second, a validated scale was found in which adjectives.describing people were known to be equal in value (Anderson, 1965). Equally-valued adjectives provide a more sensitive experiment because the choice task is more likely to result in 50/50 splits. In brand choice contexts, it is not necessarily true that commonly used traits such as quality and price are equally-valued in all product classes.

Third, the stimuli are described by common characteristics which are easy to conceptualize. Whereas, it is difficult to understand the value of a TV set with a quality rating of 70 (vs. 50) and a distortion level of 2.5% (vs. 1.5%), the idea that a person is "very humorous" (an 8 on a 1-10 scale), but "not at all open-minded" (a 2 on a 1-10 scale) is more understandable.

In the questionnaires, subjects were asked to choose to befriend one of two or three people and evaluate (rate) each of them. There were three choice tasks. In the first, people were described by their level of "imaginativeness" and "intelligence", in the second "open-mindedness" and "humor" were used and in the third "friendliness" and "thoughtfulness" provided the dimensions. Person C was inferior to Person A in one trait, thoughtfulness but not on friendliness, so dominance here does not necessarily exist. (see Figure la). In addition, the third added Person C was clearly inferior to Person B on both traits, thoughtfulness and friendliness; thereby making Person B asymmetrically dominating.

After the evaluation, subjects were asked to write a short paragraph justifying each of their choices. The length of the paragraphs provides one manipulation check for the level of justification felt. The theory behind such a check is that people may tend to write more if they feel highly accountable. In addition, the Crown-Marlowe Social Desirability Test (1967) was included to test for individual differences in social desirability and to provide covariates. Lastly, as another manipulation check on the justification manipulation, three questions were added which assessed how accountable the individual subjects felt. (e.g. "How accountable do you feel to provide justifications for your decisions?")

Fifty undergraduate subjects at the University of California, Berkeley participated in a pretest to find pairs of personality characteristics which were 1) equally valued and 2) unique in meaning. (Although the adjectives picked were rated equal in value in Anderson's list of 555 personality-trait words (1965), certain adjectives were too similar in meaning or not as equally-valued).



Justification. Subjects in the low justification condition were told that their responses would remain totally confidential and would be used for statistical purposes only. They were instructed to not write their names on the questionnaire. In the high justification condition subjects were informed that their responses would be evaluated by a team of psychological researchers and they would be asked to justify their decisions in the questionnaire and possibly in an interview. Subjects were required to include their name and telephone number before starting and their initials on each page of the questionnaire.

Number of People. There were either two or three people in each choice set. In the first choice set, used for a control, only two people were described using two attributes (e.g. friendliness and thoughtfulness). For each attributes, a specific value (ranging from one to ten) was given.

In the second choice set, as shown in Figure la-c, a third asymmetrically inferior Person C was added. Person C was described as the worst "1" on thoughtfulness and a mediocre "5" on friendliness, whereas Person B was a "2" on thoughtfulness and a high "8" on friendliness. The third person, Person C, was clearly inferior to one of the two alternatives, Person B.

Dependent Measures

Subjects were asked to select the person they would choose to befriend, a measure of choice. In addition, subjects were asked, "How much do you think you would like this person?", a measure of judgment. They responded using a seven-point scale ranging from 1 ("dislike very much") to 7 ("like very much"). The order of the choice and judgment tasks was counterbalanced to test the existence of an affect of one dependent measure on the other.


Manipulation Checks

A significant difference in the justification paragraph length suggested that the manipulation worked--average of words: 35.3 vs. 47.7, p<0.01. A second manipulation check involved three 7-point scales assessing accountability felt during the experiment. The high justification subjects felt more sensitive to what others would think of their selection (5.12 vs. 4.89; p<0.1), and felt accountable to others in case they had to justify their decision (5.04 vs. 4.89; p<0.05). The third scale (whether others would understand the subject's decisions) did not result in significant differences but was less on target. No effect due to response task order existed. The Crown-Marlowe scale did not detect individual differences.

Lastly, in order to prove that the asymmetrically dominated alternative (Person B) was indeed asymmetrically dominating, the choices and ratings of Person C were compared to those of B. The percentage preference ratio of B over C was more than 50 to 1 in all three choice sets, thereby demonstrating that B was the asymmetrically dominating.

Hypothesis 1

The choice results, which were based on the percent of respondents who rated person B on a 1-7 likability scale, are shown in Figure 2. Hypothesis la is clearly supported. There was a significant attraction effect in the aggregate (t=4.83, p>0.01) and in each of the individual cases for the high justification condition (t-5.46, p<0.01; t=4.80, p<0.05; t=4.75, p<0.05). However, hypothesis 1b was not supported--there was no attraction effect in the low justification condition. In fact, there was actually a strong negative attraction effect (t=5.03, p<0.01). Note the sharp negative slope in the aggregate and in the individual cases (all significant at the 0.01 level).

Hypothesis 2

Hypothesis 2b was supported in that there was no attraction effect under low justification in the judgment results, which were based on the percent of respondents who rated person B on a 1-7 likability scale (see Figure 3). In fact, again there was a negative attraction effect. However, it was not significant in the aggregate and only appeared in one of the three cases.

A second unexpected finding was that hypothesis 2a was not supported. In fact, there was a strong attraction effect under high justification under the judgment task.


In this discussion we will focus upon the two unexpected findings--the negative attraction effect that appeared especially in the choice/low justification cell and the attraction effect that appeared in the judgment /high justification cell.

The Negative Attraction Effect Under Choice and Low Justification

Clearly the attraction effect applies in the high justification condition. However, in the low justification condition other forces are at work.

One possible explanation is Tversky's low of similarity (1972), where the choice probability of Person A (the dissimilar alternative) increased because Person C took a disproportionate share from Person B. Based on the elimination-by-aspects model, the law of similarity refers to the idea that similar items divide the DM's loyalty. However, in the present study the increase in choice of Person C was not significant, so the negative effect of similarity does not apply.

Another possibility is that the dissimilar alternative A simply looks more interesting because it is different than the others. DMs may prefer the "black sheep" alternative because it stands apart. The DM may feel that the B and C types are more common. A preference for the unusual of different is consistent with the variety-seeking literature in consumer behavior.

In what conditions would the DM "go for" the black sheep alternative? One condition may be low justification in which the DM is not held accountable to anyone and the DM's responses are kept completely confidential. The state of an individual, who is not seen or treated like an individual, is termed "de-individuation" by Festinger, Pepitone and Newcomb. (1952). When a subject is assured of total confidentiality and told that their responses will be combined for statistical measures only, subjects feel free from restraints, less inhibited and able to indulge in forms of behavior in which, under high justification, they would not indulge (Festinger et al., 1951). Without self-monitoring or self-attention, subjects may become "more reactive to stimuli and emotions and are unresponsive to norms and long-term consequences of their behavior". (Festinger et al., 1951, p. 210) The "norms" and "long-term consequences" in this case may be the justifications needed for a decision. Without a need to justify the decision to others, DMs will "go for" the dissimilar "black sheep".

Research in social perception consistent with the "black sheep"/ de-individuation hypothesis exists. Under certain conditions (i.e. low justification or confidentiality), individuals prefer friends who are perceived unusual and who stand away from the crowd (Heider, 1958; Newcomb, 1953). Individuals may prefer the "dissimilar" alternative because s/he is more interesting. However, such reasoning is not easily justified to others.





Why is the negative attraction effect under low justification stronger in choice than in judgment? Perhaps it is due to the different decision making strategies which are associated with the two response tasks (Johnson & Russo 1984). In choice, given a difficult trade-off decision, a noncompensatory strategy is more likely where alternatives are simply eliminated. In judgment, a compensatory strategy is more likely where alternatives are weighed and DMs may rate the alternatives equally (Einhorn and Hogarth 1981). Therefore, the preference for the black sheep alternative will be more likely in choice.

The different levels of negative attraction effect in the two response tasks is particularly interesting. It could be related to the general problem in the decision making literature where preferences are not necessarily accurate predictors of choice

The Positive Attraction Effect under Judgment and High Justification

One influence of the asymmetrically inferior alternative is to make the asymmetrically dominating alternative (which dominates on one or two attributes) easier to justify to oneself and others. However, why would the strong positive attraction effect occur under judgment, where DMs may evaluate the alternatives separately and can rate the alternatives equally? There is no choice to justify.

One possible explanation is that the forces underlying the attraction effect under high justification are so strong that they apply even under an judgment task. In fact, in a strong justification manipulation, judgment tasks may resemble choice tasks. By making subjects highly accountable for their decisions, the experimenter implicitly implies a sort of decision (or choice) must be made. In such a situation, the decision making strategies used under choice may be elicited even though the response task is judgment.


First, the present research is based on the premise that "more is better"--with respect to the attributes describing the people in each choice set. Specifically, it is assumed that a person who is very friendly (an "8" on a 1-10 scale) would bc more attractive than another who is unfriendly or moderately friendly (a "2" or a "5"). The optimal level of an attribute was not tested in the present paper. As in product examples, the promise that "more is better" is not always true.

Moreover, the meaning of each attribute may change when paired to another specific attribute. For example, friendliness may bc a positive attribute when paired with thoughtfulness but a negative one when paired with humor (both of which were equally-valued according to Anderson's list). Future research could focus on the effect of one attribute on the other as well as the optimal level of each attribute.

Second, further study may be done on the negative attraction effect and conditions under which it occurs. In addition, research may focus on why the negative attraction effect occurs by taking process measures. In particular, it would be useful to show if, indeed, a focus on choice occurs in a judgment task under high justification.

Third, no steps were taken to show that the different response modes, judgment and choice, actually elicited different decision making strategies. The presence of these decision processes may be tested as Tversky, Slovic and Sattath did in Contingent Weighting in Judgment and Choice (1988). For example, a coding scheme, such as Integrative Complexity Coding, may be used to code the level of complexity in the DM's reasoning and investigate changes m cognitive processing or decision-making strategies.

Fourth, future research would be productive in the realm of cross-cultural literature concerning the attraction effect. Specifically, does the attraction effect occur in other cultures? If so, do the same explanations of the phenomenon apply (e.g. justification)?

The present study was replicated in Paris with 40 subjects. Results showed similar findings as the present experiment, thus indicating that the attraction effect is cognitive process, not a cultural norm. However, such an interpretation is tentative. The evidence is weak because of small sample size, different age and profession in subject pool and translation differences. Much room remains in the social perception and consumer behavior literature for such cross cultural research.

Lastly, "real world" marketing implications of the present research can be seen in the realm of new product introduction or development. Given a situation where two (or more) equally-valued brands are on the market, an existing competing manufacturer may consider the possibility of introducing an asymmetrically inferior new brand in order to increase the market share of their existing (asymmetrically dominating) brand and decrease the market share of the competitor(s).

More interestingly, a manufacturer could explore the realm of justification to change market share in a specific market. By increasing the level of accountability through advertising or direct marketing, a manufacturer may increase market share for their brand. For example, given a brand situation with three alternatives positioned as in Figure 1, under low justification (a "normal" choice situation, where the consumer does not have to justify his/her choice to anyone), the consumer may be choosing brand A---the dissimilar alternative (i.e. the negative attraction effect occurs). By holding the consumer accountable for their decision by first asking them why they choose their choice and then listing some of the reasons why the DM should choose their brand, the choice probability of the manufacturer's brand B may increase.


Abelson, Robert (1964), "The Choice Theories," in S. Moscow and A. Brayfield (Eds.) Decision and Choice, NY: McGraw Hill.

Anderson, Norman H. (1981), "Likableness Ratings of 555 Personality-Trait Words," Journal of Personality and Social Psychology, 9, 272-279.

Assar, Amardeep and Dipankar Chakravarti (1984), "Attribute Range Knowledge: Effects on Consumers' Evaluation of Brand-Attribute Information and Search Patterns in Choice," in EMA Educators' Proceedings, eds. Russell W. Belk et al., Chicago: American Marketing Association, 62-67.

Bettman, J.R. (1982), "A Functional Analysis of the Role of Overall Evaluation of Alternatives in Choice Processes", Advances in Consumer Research, 9, 87-93.

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.

Chaiken, Shelly (1980), "Heuristic Versus Systematic Information Processing and the Use of Source Versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 45 (March), 560-570.

Curley, Shawn P., J. Frank Yates, and Richard A. Abrams (1986), "Psychological Sources of Ambiguity," Organizational Behavior and Human Decision Processes, 38 (2), 230-256.

Dipboye, Robert L. (1977), "Alternative Approaches to Deindividuation", Psychological Bulletin, 84 (6), 1057- 1075.

Einhorn, H.J., and Hogarth, R.M. (1981), "Behavioral Decision Theory: Processes of Judgment and Choice", Annual Review of Psychology, 32, 52-88.

Festinger, Leon, "A Theory of Cognitive Dissonance," Evanston, Illinois: Row, Peterson, 1957.

Festinger, L., Pepitone, A., Newcomb, T. (1952), "Some Consequences of Deindividuation in a Group", Journal of Abnormal and Social Psychology, 47, 382-389.

Hagafors, R. and B. Brehmer (1983), "Does Having to Justify One's Judgments Change the Nature of the Judgment Process?" Organizational Behavior and Human Performance, 31, 223-243.

Heider, Fritz, "The Psychology of Interpersonal Relations," New York, Wiley, 1958.

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, 90-98.

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

Johnson, E.J., and Russo. (1984), "Product Familiarity and Learning New Information", Journal of Consumer Research, 11 542-550.

Kahneman, D., and Tversky, A. (1984), "Choices, Values, and Frames", American Psychologist, 39,341 -350.

Lerner, Melvin J. and Selwyn Becker (1965), "Interpersonal Choice as a Function of Ascribed Similarity and Definition of the Situation."

Luce, R. Duncan (1959), Individual Choice Behavior, New York: Wiley.

Luce, R. Duncan (1977), 'The Choice Axiom After Twenty Years," Journal of Mathematical Psychology, 15, 215-233.

McAllister, D.W., Mitchell, T.R., & Beach, L.R., (1979), "The Contingency Model for the Selection of Decision Strategies: An Empirical Test of the Effects of Significance, Accountability, and Reversibility", Organizational Behavior and Human Performance, 24, 228-244.

Newcomb, Theodore, M., (1953), "Social Psychology and Group Processes," Annual Review of Psychology, (4), 183-214.

Payne, John W. (1982), "Contingent Decision Behavior," Psychological Bulletin, 92 (Sept.), 382-402.

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

Simonson, Itamar (1989), "Choice Based on Reasons: The Case of Attraction and Compromise Effects", Journal of Consumer Research, 16, 158-173.

Slovic, Paul (1975), "Choice Between Equally-Valued Alternatives," Journal of Experimental Psychology: Human Perception and Performance, 1, 280-287.

Slovic, Paul, Baruch Fischhoff, and Sara Lictenstein (1982), "Response Mode, Framing and Information Processing Effects in Risk Assessment," New Directions in Methodology of Social and Behavioral Science: The Framing of Questions and the Consistency of Response, San Francisco, Calif.: Josey Bass.

Tetlock, Philip E. (1985), "Accountability: The Neglected Social Context of Judgment and Choice," Research in Organizational Behavior, 7, 297-332.

Tetlock, Philip E. (1985), "Accountability: A Social Check on the Fundamental Attribution Error," Social Psychology Quarterly, Vol. 48, 3, 227-236.

Tversky, Amos, Paul Slovic, Samuel Sattath (1988), "Contingent Weighting in Judgment and Choice", Psychological Review, 95, (3), 271384.

Tversky, Amos (1977), "Features of Similarity", Psychology Review, 84, 327-352.