Explaining Negativity Biases in Evaluation and Choice Behavior: Theory and Research

David E. Kanouse, The Rand Corporation
ABSTRACT - The question of how people integrate positive and negative information about persons or objects is of interest to researchers studying attitudes and behavior in many different settings. This paper describes the consistent finding of a "negativity bias" in studies of information integration, discusses several types of explanations that have been offered for it, and draws on related work on social judgment to suggest directions for future research.
[ to cite ]:
David E. Kanouse (1984) ,"Explaining Negativity Biases in Evaluation and Choice Behavior: Theory and Research", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 703-708.

Advances in Consumer Research Volume 11, 1984      Pages 703-708

EXPLAINING NEGATIVITY BIASES IN EVALUATION AND CHOICE BEHAVIOR: THEORY AND RESEARCH

David E. Kanouse, The Rand Corporation

[James P. Kahan provided helpful comments on a draft.]

ABSTRACT -

The question of how people integrate positive and negative information about persons or objects is of interest to researchers studying attitudes and behavior in many different settings. This paper describes the consistent finding of a "negativity bias" in studies of information integration, discusses several types of explanations that have been offered for it, and draws on related work on social judgment to suggest directions for future research.

NEGATIVITY IN INFORMATION INTEGRATION

People are often called upon to form overall judgments about objects that possess multiple attributes, some of them more desirable than others. Presumably, such judgments require people to integrate separate items of information into a unitary impression of the object. Anderson (1965, 1971) proposed a simple weighted averaging model to account for these judgments. According to the model, a person's overall evaluation of an object will be a weighted average of the person's evaluations of its individual attributes.

Although the averaging model has provided a useful description of behavior on a wide range of integration tasks, people's judgements seem to depart from the model's predictions in a number of ways. One of the most important is that people tend to weigh negative information more heavily than positive information (Kanouse and Hanson, 1972). This tendency has been documented most thoroughly in the impression format-on literature, where many studies have found that impressions based on a combination of positive and negative traits are more negative than would be predicted from the scale values of the traits considered separately (e.g., Hamilton and Huffman, 1981; Hamilton and Zanna, 1972 Hodges, 1974; Levin and Schmidt, 1969; Miller and Rowe, 1967; Wyer and Watson, 1969). Similar effects have been reported when the task requires integration of information about specific behaviors rather than traits (Birnbaum, 1972 Richey, Koenigs, Richey, and Fortin, 1975). Other studies have found that negative first impressions are more difficult to change (Briscoe, Woodyard, and Shaw, 1967; Freedman and Steinbruner, 1964) and that negative information is more likely than positive information to have an enduring effect (Cusamano and Richey, 1970; Gray-Little, 1973; Richey, McClelland, and Shimkunas, 1967). Rosenbaum and Levin (1969) found that the impact of negative information, unlike that of positive information, does not vary much with the credibility of the information's source. Lutz ( 9'5) found that the symmetrical presentation of belief-relevant information concerning the attributes of a product led subjects to shift their attitudes further in the negative direction when the implications of the information were negative than they shifted in the positive direction when the implications were positive. Studies of employment interviews have found that the negative characteristics of prospective employees exert a disproportionate influence on hiring decisions (Bolster and Springbett, 3561; Carl son, 1971; Constantin, 1976; Crissy and Regan, 1951; Springbett, (1958).

All the above is consistent with the notion that when people combine information, they weigh negative information more heavily than positive; i.e., that the whole is evaluated more negatively than the average of its parts. A dozen years ago, Kanouse and Hanson (1972) reviewed evidence for the existence of this "negativity bias" and offered several possible explanations for it. Since then, the research literature on social cognition has grown enormously (e.g., Nisbett and Ross, 1980; Wyer and Carlston, 1979). Accordingly, it seems a good time to take a fresh look at this particular negativity bias in the light of recent developments in relevant theory and research. [Recently, Amabile and Glazebrook (1982) described a different sort of "negativity bias" that they believe occurs in certain social situations; namely, a bias to evaluate other people negatively in the interests of self-enhancement. This situationally-triggered tendency to be hypercritical of others should not be confused with the information processing bias at issue here.] This paper attempts to do that, briefly reviewing several possible explanations for the recurrent findings and suggesting promising directions for future research.

EXPLANATIONS FOR NEGATIVITY

Explanations of the negativity bias are many and varied. In sorting them out, it is useful to distinguish between micro-level explanations that seek to elucidate the perceptual and judgmental processes underlying the phenomenon and macro-level explanations that seek to place the phenomenon in a larger context wherein the bias makes sense. Below, we consider several explanations of each type, beginning with the micro level.

Response Scale Artifacts

One possible explanation for the phenomenon is that it reflects a response bias rather than a judgmental bias. Subjects in information integration studies not only make judgments about an object but must also translate those judgments into a response on a scale supplied by the experimenter. There is no guarantee that the function relating the underlying judgments to the overt responses is linear (Parducci, 1965; Upshaw, 1969). For example, subjects may have to project Judgments from an asymmetric underlying distribution onto a symmetric scale. In such a case, one might well find apparent shifts in the scale values for integrated judgments from those predicted by a linear combinatorial rule. However, Birnbaum (1974) thoroughly investigated this possibility and concluded that the negativity bias is a genuinely judgmental phenomenon; no reasonable transformation of the response scale will eliminate s ha loin<:

Selective Attention

Judgments of the type we are considering are the end product of a sequence of information processing activities that begins with the perceptual registration of the relevant items of information (e.g., trait descriptors) and ends with a response describing the overall judgment. The bias favoring negative information could be introduced at any or all of these stages. Fiske (1980) provides evidence that the bias toward weighing negative information more heavily is reflected in the amount of perceptual attention given to that information. She presented subjects with captioned slides depicting people engaged in any of eight different activities that pretesting had shown would produce differential ratings of the likability of the stimulus person: very positive, moderately positive, moderately negative, or very negative. Dependent measures included the amount of time spent looking at each slide as well as likableness ratings. The results showed that negative information not only received disproportionate weight in subject's overall judgments, but also captured a greater share of their attention, as reflected in viewing times. These effects occurred independently of the evaluative extremity of the information, although extreme information also received more attention and more weight.

Fiske's results suggest that negative information may be more "attention-grabbing" at the outset (McArthur, 1981; Taylor and Fiske, 1978). The differential weight it receives in integration judgments could reflect this initially greater salience rather than any additional prominence the information takes on during the information integration task itself.

The Judgment Process

Perhaps the most common view of the greater impact of negative information is that it reflects an asymmetry in the importance of the information under the typical judgment strategy. That is, people weigh negative information more heavily than positive because, implicitly, that is how they think it should be weighed. Several possible macro-level explanations have been offered for why this should be the case, but for the moment we are concerned with attempts to understand what it is that people do rather than why it is that it might make sense for them to do it. One possibility is that Anderson's (1965, 1971) averaging model gives a fairly good account of how integrative judgments are made, except that in "averaging" individual impressions, people give more weight to those that are negative or extreme.

Several investigators, however, have argued that what people actually do may be quite different from the averaging process assumed in Anderson's model. For example, Warr (1974) has argued that when people are asked to make a judgment on the basis of several items of information, they select the item with the most extreme evaluation implications and make a provisional judgment based on that item. Then they examine the implications of the remaining items and modify the provisional judgment accordingly. Such a process would, of course, result in an "overweighing" of more extreme information (or more negative information; see Warr and Jackson, 1977), but as a byproduct of the way in which the judgment is made. The process would involve anchoring rather than differential weighting of items considered in parallel.

Wyer (1973) has proposed a model that views the entire judgment process as a set of categorical probability judgments similar to those made in a concept identification task. People use different items of information about an object to circumscribe the categories to which the object can belong. The overall scale rating assigned to an object represents the expected value of a subjective probability distribution; i.e., the most representative category of those to which the object could belong. Separate items of information are neither summed nor averaged. Rather, their joint implications depend on the (estimated) conjunctive probability distribution, which may be quite different from the average of the individual distributions.

By treating scale ratings as the outcome of a subjective probability estimation task, Wyer's model emphasizes the importance of uncertainty. Other things equal, the most influential items of information about an object will be the least ambiguous, because such items will most strongly circumscribe the categories to which the object could belong. Wyer (1973) found that negative trait adjectives were not only more influential than equally polarized positive adjectives, but were also rated as less ambiguous.

Leon (1981) has also argued against the averaging model, proposing that negative information is treated categorically. For example, people may first screen for the presence of any negative information and if they find any, reject the object outright. Only after an object passes this initial screen is it evaluated for degree of favorability.

Perhaps the most common criticism of the averaging model concerns its treatment of attributes as independent. Because adjectives (like other cues) are often ambiguous, their meaning can depend on the other adjectives they are paired with. For example, a "self-confident malicious" person may be judged more malicious than a "shy malicious" person. Hypotheses concerning contextual changes in meaning have led to a controversy that has little bearing on this discussion (see Ostrom, 1977); it is worth noting, however, that negative adjectives may lead to contextual shifts in meaning more often than positive ones do.

Retrieval and Accessibility

Very often, judgments about an object require retrieval of information from memory. Negativity biases can arise in the retrieval process just as they can in other processes. Indeed, several studies cited earlier found negativity biases that emerged only over time, a fact that implicates memory or retrieval processes.

Two points about memory and retrieval processes are pertinent. First, memory representations are often organized around overall impressions that label and interpret the original information. Over time, a person's judgments and inferences rely increasingly on the label and interpretation in memory rather than the original information on which it was based (Bartlett, 1932). For example, people may retain the affective impression of an object long after they have forgotten the specific features that gave rise to the affect (Burnstein and Schul, 1983).

Second, when people are called upon to make a judgment about an object, they do not generally carry out an exhaustive memory search for all relevant information; instead, they sample only a subset of the information that is most accessible (Tversky and Kahneman, 1973, 1974; Taylor and Fiske, 1978). Among the factors that seem to increase the accessibility of information are its novelty and distinctiveness (Hamilton and Gifford, 1976; Hastie and Kumar, 1979). Negative attributes and behaviors are often presumed to be more novel and distinctive. Consistent with the two points above, Carlston (1980) found that a week after subjects were exposed to positive and negative information about traits and events, they recalled negative information more accurately and confidently than positive information

Although several studies have pointed to differences in memory or retrieval of positive and negative information, most such findings have been incidental to the original purpose of the investigation. Lingle, Kukerich and Ostrom (1983) are among the few investigators who have explicitly tested hypotheses about how people access positive and negative information in forming overall judgments. Their study followed an earlier study by Lingle and Ostrom (1979) that found evidence that judges of occupational suitability selectively search for unfavorable (disqualifying) attributes, or alternatively, that they selectively search for hypothesis-disconfirming evidence. Lingle, Kukerich and Ostrom (1983) sought to pin down the type of preference involved by varying the wording of the question; e.g., "Would this person be a success (failure) as a lawyer?" The results supported the idea of a hypothesis-disconfirming preference rather than a negativity preference. That is, judges selectively searched for negative information when Judging probable success and positive information when judging probable failure.

The Figure-Ground Hypothesis

So far we have largely focused on the processes involved in attention, encoding, retrieval, and integration of information and how negativity biases might occur at each stage. we will now briefly review explanations that relate negativity biases to the goals that motivate evaluative judgments or to phenomena that are external to the judgment process itself.

Several explanations for the relative weight given negative attributes are based, somewhat ironically, on the overwhelming prevalence of positive attributes in the world or at least in the world as people describe it. Positive words are used much more frequently than negative ones (Boucher and Osgood, 1968); most people both expect and report high levels of personal happiness (Bradburn and Caplovitz, 1965; Cantril, 1965); and people tend to evaluate other people, including strangers, quite favorably (Sears and Whitney, 1973). The fact that people's impressions of the world are so diffusely pleasant may make it easier for negative information to stand out by contrast.

Novelty and Distinctiveness

The prevalence of positive information means that negative information is both more novel and more distinctive; that is, it is less frequently encountered in and of itself and also more likely to facilitate distinctions among objects. Its greater novelty should increase the probability that it will be attended to and remembered, and its distinctiveness should give it more weight in judgments that require differentiation among objects (Fiske, 1980).

Perhaps the most interesting implication of this notion is that negativity biases critically depend on the positivity of the underlying distribution. Judgments in domains where positive attributes are not the rule should show no negativity bias; similarly, judgments about objects that are more easily distinguished by their positive attributes than by their negative ones should also show no such bias. Eliminate a negative attribute's "surprisingness" (Feldman, 1966) or its "informativeness" (Fiske, 1980) and you will reduce its impact. Although plausible, this hypothesis has not been subjected to the kind of systematic testing that it deserves.

Definitiveness

Jones and Davis (1965) have suggested that because people have mostly positive characteristics, they are often assigned less responsibility for their positive traits than for their negative ones. Normative traits are taken for granted, and the individual is held responsible only for distinguishing traits. Because negative traits are more likely to be non-normative, they are apt to weigh more heavily in evaluations of the individual.

Although this line of reasoning was developed in connection with perceptions of people rather than judgments of objects an analogous point can be made for objects. It seems plausible that the attributes an object shares with others of it class are less definitive of the object than are its distinctive features. For example, a particular model of auto mobile is more likely to be characterized by (and evaluated in terms of) the features that distinguish it from other models than by the features it shares with them. Truly "standard equipment" is taken for granted until one moves to the more general category "automobile", where it is the common features that become definitive (Rosch and Lloyd, 1978).

Evaluations have an implicit (and often explicit) referent: I evaluate a particular model of car as an automobile, or as a sports car or station wagon. Its distinctive features with respect to the referent class are more likely to be salient in my evaluation than its common features, and I am more likely (if asked) to report them as definitive attributes. Which features are common and which distinctive, will, of course, depend on the referent class. This dependence could be exploited to test hypotheses about how positive and negative features of an object affect the way it is evaluated under conditions where the features are shared or not shared with the referent class.

Credibility

When information about an object comes through the opinions or recommendations of another person, negative information may be more credible than positive information. The reason is that there are often strong normative pressures to say positive things, so that a person who expresses negative feelings is more likely to be considered sincere (Jones and Davis, 1965). Mizerski (1982) has recently tested a hypothesis regarding beliefs about unfavorable product ratings that is based on this general notion. Additional studies that examine how people assess the credibility of positive and negative information could prove worthwhile.

Interference Effects

Kanouse and Hanson (1972) suggested that negative attributes can often interfere with the enjoyment of positive attributes. For example, one rancid ingredient can spoil the finest soup (Kanouse, 1972). A fast car with transmission troubles is no longer a fast car, although it is very much a car with transmission troubles. It is much more difficult to find examples of positive attributes that nullify the effects of negative ones. Thus, interactions among various attributes may tend to lower evaluations below what would be expected from the favorability of the attributes considered separately. Examples are amusing to consider and, up to a point, instructive; however they cannot answer the question of how prevalent such interference effects are. To do that requires systematic research examining interference effects within specified content domains.

THE FRAMING OF EVALUATIONS

The notion of referent class discussed above touches on the larger question of how the evaluative judgments we have discussed are related to other types of judgments, such as those involved in choice. In an earlier discussion of negativity (Kanouse and Hanson, 1972), we examined both the impression formation literature and the literature on risky choices and decision-making, because we thought there was a connection between the judgments and the biases found in these two literatures. I still do, although space permits only a sketchy elaboration on this point.

Psychological studies of risky choices indicate that people commonly make risk averse choices in situations involving gains, whereas they commonly make risk seeking choices in situations involving losses (Tversky and Kahneman, 1974). For example, most people prefer a sure gain of $80 to a gamble that gives them an 85 percent chance of winning $100 and a 15 percent chance of winning nothing. Such a preference is "risk averse" because the monetary expectation of the gamble is higher. However, when people are asked to choose between a sure loss of $80 and a risky choice that gives them an 85 percent chance of losing $100 and a 15 percent chance of losing nothing, most people prefer the gamble to the sure loss. In this case, the choice is "risk seeking" rather than "risk averse." Often it is possible to reverse people's risk preferences simply by framing the decision problem differently (Tversky and Kahneman, 1981). The reason is that people tend to evaluate alternatives not only in relation to each other but also in relation to a neutral reference point that determines the amounts that are "gained" or "lost" under each alternative. Depending on the decision frame adopted, this reference point may be the status quo, or it may be the actual or anticipated state of things at some earlier or later point. The relative evaluation of alternatives can vary markedly depending on which frame is chosen.

Evaluative judgments of the type we have examined are, of course, not the same as the evaluations involved in risky choices. For one thing, outcomes in risky decision problems are typically disjunctive and single-attributed, whereas in information integration tasks they are typically conjunctive and multi-attributed. However, both types of judgments require the assessment of complex bundles of attributes of varying hedonic value, and for both types of judgments people seem to weigh negative attributes more heavily than positive ones. (A consistent finding in the risk literature is that people find losses more aversive than equivalent gains are pleasurable).

The importance of framing in information integration tasks is not known. However, I would argue that its potential for affecting judges' evaluation strategies (and the attendant outcomes) is great enough to warrant further study. Certainly it seems that when a judge's reference point is very high and the number of judgments required is very large, an evaluation strategy that involves overweighting negative information makes sense. This will often be the case in employment interviews, especially when there are many more qualified applicants than openings. It should also be fairly frequent in consumer product selection.

Results from the study by Lingle, Dukerich, and Ostrom (1983) provide initial evidence that framing may have powerful effects on judgments in information integration tasks, just as it does on risky choices. However, this study is only a beginning, and needs to be followed by additional work that explores the kinds of frames judges may employ and identifies their most common or typical effects. Studies of how the evaluation context affects the frame chosen may also be useful; for example, Peter Wright's work on the effects of time pressure (Wright, 1974) and time horizon (Wright and Weitz, 1977), which suggests that the frame chosen may be sensitive to time perspective.

CONCLUSION

This paper has attempted to take a fresh look at a well-documented bias; namely, people's tendency to weigh negative information more heavily than positive when they form overall judgments of objects, especially other people. Despite the burgeoning research over the past few years on social cognition and on systematic biases that affect human judgment, there has been very little theory-driven research designed to illuminate this negativity bias - to determine the range of situations in which it occurs, the stages of information processing where bias enters, and what the bias implies about basic cognitive processes involved in evaluative judgment. J hope that this brief review will help to rectify that situation by suggesting some approaches to this problem that appear potentially fruitful.

REFERENCES

Amabile, Teresa M. and Ann R. Glazebrook (1982). "A Negativity Bias in Interpersonal Evaluation," Journal of Experimental Social Psychology, 18 (1), 1-22.

Anderson, Norman H. (1965), "Adding Versus Averaging as a Stimulus Combination Rule in Impression Formation, " Journal' of Experimental Psychology, 70 (4), 394-400.

Anderson, Norman H. (1971), "Integration Theory and Attitude Change," Psychological Review, 78 (3), 171-206.

Bartlett, Frederic C. (1932), Remembering: A Study in Experimental and Social Psychology, Cambridge: Cambridge University Press, 1932.

Birnbaum, Michael (1972), "Morality Judgments: Tests of an Averaging Model," Journal of Experimental Psychology, 93 (1), 35-42.

Birnbaum, Michael H. (1974), "The Nonadditivity of Personality Impressions," Journal of Experimental Psychology, 102 (3). 543-561.

Bolster, B. T. and B. M. Springbett (1965), "The Reaction of Interviewers to Favorable Information," Journal of Applied Psychology, 45 (2), 97-103.

Boucher, Jerry and Charles E. Osgood (1969), "The Polyanna Hypothesis," Journal of Verbal Learning and Verbal Behavior, 8 (1), 1-8.

Bradburn, Norman and David Caplovitz (1965), Reports on Happiness, Chicago: Aldine.

Briscoe, May E., Howard D. Woodyard and Marvin E. Shaw (1967), "Personality Impression Change as a Function of the Favorableness of First Impressions," Journal of Personality, 35 (2), 343-357.

Burnstein, Eugene and Yaacov Schul (1983), "The Informational Bases of Social Judgments: Memory for Integrated and Nonintegrated Trait Descriptions," Journal of Experimental Social Psychology, 19 (1), 49-57.

Cantril, Hadley (1965), The Pattern of Human Concerns, New Brunswick. NJ: Rutgers University Press.

Carlson, Robert E. (1971), "Effect of Interview Information in Altering Valid Impressions," Journal of Applied Psychology, 55 (1), 66-72.

Carlston, Donal E. (1980), "The Recall and Use of Traits and Events in Social Inference Processes," Journal of Experimental Social Psychology, 16 (4), 303-328.

Constantin, S. W. (1976), "An Investigation of Information Favorability in the Employment Interview," Journal of Applied Psychology, 61 (6), 743-749.

Crissy, W. J. E., and .James J. Regan (1951), "Halo in the Employment Interview," Journal of Applied Psychology, 35, 338-341.

Cusumano, Donald R. and Marjorie R. Richey (1970), "Negative Salience in Impressions of Character: Effects of Extremeness of Stimulus Information," Psychonomic Science, 90 (9), 81-83.

Feldman, Sheldon (1966), "Motivational Aspects of Attitudinal Elements and Their Place in Cognitive Interaction," in Cognitive Consistency: Motivational Antecedents and Behavioral Consequents, ed. Sheldon Feldman, New York: Academic Press.

Fiske, Susan T. (1980), "Attention and Weight in Person Perception: The Impact of Negative and Extreme Behavior," Journal of Personality and Social Psychology 38 (6), 889-906.

Freedman, Jonathan L. and John D. Steinbruner (1964), "Perceived Choice and Resistance to Persuasion," Journal of Abnormal Psychology, 68 (6), 678-681.

Gray-Little, Bernadette (1973), "The Salience of Negative Information in Impression Formation Among Two Danish Samples," Journal of Cross-cultural Psychology, 4 (2), 193-206.

Hamilton, David L. and Robert K. Gifford (1976), "Illusory Correlation in Interpersonal Perception: A Cognitive Basis of Stereotypic Judgments," Journal of Experimental Social Psychology, 12 (4), 392-407.

Hamilton, Davit L. and Leroy J. Huffman (1971), "Generality of Impression-formation Processes for Evaluative and Non-evaluative Judgments," Journal of Personality and Social Psychology, 20 (2), 200-207.

Hamilton, David L. and Mark P. Zanna (1972), "Differential Weighting of Favorable and Unfavorable Attributes in Impressions of Personality," Journal of Experimental Re- search in Personality, 6 (2-3), 204-212.

Hastie, Reid and Purohit Anand Kumar (1979), "Person Memory: Personality Traits as Organizing Principles in Memory for Behaviors," Journal of Personality and Social Psychology, 37 (1), 25-38.

Hodges, Bert H. (1974), "Effect of Valence on Relative Weighting in Impression Formation," Journal of Personality and Social Psychology, 30 (3) 378-381.

Jones, Edward E. and Keith E. Davis (1965), "From Acts to Dispositions: The Attribution Process in Person Perception," in Advances in Experimental Psychology, Vol. 2., ed. Leonard Berkowitz, New York: Academic.

Kanouse, David E. (1972), "Language, Labeling, and Attribution," in Attribution: Perceiving the Causes of Behavior, eds. Edward E. Jones et al., Morristown, NJ: General Learning Press.

Kanouse, David E. and L. Reid Hanson, Jr. (1972), "Negativity in Evaluations," in Attribution: Perceiving the Causes of Behavior, eds. Edward E. Jones et al., Morristoun, NJ: General Learning Press.

Leon, Federico R. (1981), "The Role of Positive and Negative Outcomes in the Causation of Motivational Forces," Journal of Applied Psychology, 66 (1), 45-53.

Levin, Irwin P. and Charles F. Schmidt (1969), "Sequential Effects in Impression Formation with Binary Intermittent Part 1), 283-287.

Lingle, John H., Janet M. Dukerich and Thomas M. Ostrom (1983), "Accessing Information in Memory-Based Impression Judgments: Incongruity versus Negativity in Retrieval Selectivity," Journal of Personality and Social Psychology, 44 (2), 262-272.

Lingle, John H. and Thomas M. Ostrom (1979), "Retrieval Selectivity in Memory-Based Impression Judgments," Journal of Personality and Social Psychology, 37 (2), 180-194.

Lutz, Richard J. (1975), "Changing Brand Attitudes Through Modification of Cognitive Structure," Journal of Consumer Research, 1 (4), 49-59.

McArthur, Leslie Z. (1981), "What Grabs You: The Role of Attention in Impression Formation and Causal Attribution," in Social Cognition: The Ontario Symposium, Vol. L, ed. E. Tory Higgins, C. Peter Herman, Mark P. Zanna, Hillsdale, NJ: Lawrence Erlbaum Associates.

Miller, James W. and Patricia M. Rowe (1967), "Influence of Favorable and Unfavorable Information Upon Assessment Decisions," Journal of Applied Psychology, 51 (5), 432-435.

Mizerski, Richard W. (1982), "An Attribution Explanation of the Disproportionate Influence of Unfavorable Information," Journal of Consumer Research, 9 (3), 301-310.

Nisbett, Richard E. and Lee D. Ross (1980), Human Inference: Strategies and Shortcomings in Social Judgment, New York: Prentice-Hall.

Ostrom, Thomas M. (1977), "Between Theory and Within Theory Conflict in Explaining Context Effects in Impression Formation," Journal of Experimental Social Psychology, 13 (5), 492-503.

Parducci, Allen (1965), "Category Judgment: A Range-Frequency Model," Psychological Review, 72 (6), 407-418.

Richey, Marjorie R., Robert J. Koenigs, Harold W. Richey and Richard Fortin (1975), "Negative Salience in Impressions of Character: Effects of Unequal Proportions of Positive and Negative Information," Journal of Social Psychology, 97 (December), 233-241.

Richey, Marjorie H., Lucille McClelland, and Algimantas Shimkanas (1967), "Relative Influence of Positive and Negative Information in Impression Formation and Persistence," Journal of Personality and Social Psychology, 6 (3), 322-327.

Rosch, Eleanor and Barbara B. Lloyd (1978), Cognition and Categorization, Hillsdale, NJ: Lawrence Erlbaum Associates.

Rosenbaum, Milton E. and Irwin P. Levin (1969), "Impression Formation as a Function of Source Credibility and the Polarity of Information," Journal of Personality and Social Psychology, 12 (1), 34-37.

Sears, David O. and Richard Whitney (1973), "Political Persuasion," in Handbook of Communication, eds. Ithiel de Sola Pool et al., Chicago: Rand McNally.

Springbett, B. M. (1958), "Factors Affecting the Final Decision in the Employment Interview," Canadian Journal of Psychology, 12 (1), 13-22.

Taylor, Shelley E. and Susan T. Fiske (1978), "Salience, Attention, and Attribution: Top of the Head Phenomena," in Advances in Experimental Social Psychology, Vol. 11, ed. Leonard Berkowitz, New York: Academic Press.

Tversky, Amos and Daniel Kahneman (1973), "Availability: A heuristic for Judging Frequency and Probability," Cognitive Psychology, 5 (September), 207-232.

Tversky, Amos and Daniel Kahneman (1978), "Judgment Under Uncertainty: Heuristics and Biases," Science. 185 (September 27), 1124-1131.

Tversky, Amos and Daniel Kahneman (1981), "The Framing of Decisions and the Psychology of Choice," Science, 211 (January 30). 453-458.

Upshaw, Harry S. (1969), "The Personal Reference Scale: An Approach to Social Judgment," in Advances in Experimental Social Psychology, Vol. 4, ed. Leonard Berkowitz, New York, Academic Press.

Warr, Peter (1974), "Inference Magnitude, Range, and Evaluative Direction as Factors Affecting Relative Importance of Cues in Impression Formation," Journal of Personality and Social Psychology, 30 (2), 191-197.

Warr, Peter and Paul Jackson (1977), "Salience, Importance, and Evaluation in Judgments About People," British Journal On Social and Clinical Psychology, 16 (Part 1), 35-45.

Wright, Peter (1974), "The Harassed Decision Maker: Time Pressures, Distractions, and the Use of Evidence," Journal of Applied Psychology, 59 (5), 555-561.

Wright, Peter and Barton Weitz (1977), "Time Horizon Effects on Product Evaluation Strategies," Journal of Marketing Research, 14 (November), 429 443.

Wyer, Robert S. (1973), "Category Ratings as 'Subjective Expected Values': Implications for Attitude Formation and Change," Psychological Review, 80 (6), 445-467.

Wyer, Robert S. and Donal E. Carlston (1979), Social Cognition Inference, and Attribution, Hillsdale, NJ: Lawrence Erlbaum Associates.

Wyer, Robert S. and Stanley F. Watson (1969), "Context Effects in Impression Formation," Journal of PersonalitY and Social Psychology, 12 (1), 22-33.

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