Negativity in Intradimensional Judgments of Presidential Candidates

Jill Gabrielle Klein, INSEAD
ABSTRACT - Character judgments of candidates in three presidential elections were examined using data from the 1984, 1988, and 1992 National Election Studies. When differences were perceived between the candidates on a particular attribute, that difference was more predictive of overall evaluations for the candidate viewed more negatively on that attribute than for his more positively-viewed opponent. Thus, support was found for the use of intradimensional strategies and negativity in candidate evaluations.
[ to cite ]:
Jill Gabrielle Klein (1998) ,"Negativity in Intradimensional Judgments of Presidential Candidates", in NA - Advances in Consumer Research Volume 25, eds. Joseph W. Alba & J. Wesley Hutchinson, Provo, UT : Association for Consumer Research, Pages: 574-577.

Advances in Consumer Research Volume 25, 1998      Pages 574-577


Jill Gabrielle Klein, INSEAD

[The author would like to thank Richard Ettenson, Gregory Fischer, Andrew John, Donald Kinder, and Christie Nordhielm for their help with this paper.]


Character judgments of candidates in three presidential elections were examined using data from the 1984, 1988, and 1992 National Election Studies. When differences were perceived between the candidates on a particular attribute, that difference was more predictive of overall evaluations for the candidate viewed more negatively on that attribute than for his more positively-viewed opponent. Thus, support was found for the use of intradimensional strategies and negativity in candidate evaluations.

Every four years the American public must choose among a cast of presidential contenders. By the time the election nears, voters are likely to have been exposed to vast amounts of information about the final contenders, and on the basis of this information voters select their preferred candidate. Not surprisingly, judgments of candidate personality strongly influence both overall candidate evaluations and final voting decisions (Kinder and Abelson 1981; Kinder and Sears 1985). It is unclear, however, whether candidates are compared on a given personality attribute (an intradimensional strategy) or whether an overall evaluation of each candidate is formed and then compared (a holistic strategy).

Consumers use intradimensional strategies in a variety of tasks (Abelson and Levi, 1985; Johnson and Puto, 1987; Payne, 1976; Payne, Bettman and Johnson 1988, 1992; Malhotra, 1982). For example, decision-makers have been found to use this strategy when selecting scholarship candidates (Russo and Dosher, 1983), and when making purchase decisions, such as choosing a car or selecting an apartment (Russo and Rosen, 1975; Payne, 1976). Intradimensional decision rules are most likely to be utilized when the number of alternatives (e.g., brands) is small (Abelson and Levi 1985; Payne, 1976). Giventhat most presidential elections involve the choice between two major candidates, it seems likely that voters will employ intradimensional strategies when choosing among candidates.

While candidates make great efforts to project positive images to the public, it appears to be campaign gaffes that attract voters’ attention. This is consistent with psychological research which suggests that negative information is weighted more heavily than positive information when impressions of others are formed: negative items appear to have a stronger "pull" on overall impressions. Such a tendency is known as the negativity effect (e.g., Anderson, 1965; Fiske 1980; Hamilton and Huffman, 1971; Hamilton and Zanna, 1972; Kanouse and Hanson, 1972; Skowronski and Carlston, 1989). Consumers also exhibit negativity in product evaluations and choices (Lutz 1975; Mizerski 1982; Weinberger, Allen, and Dillon 1980; Weinberger and Dillon 1986).

Negativity also characterizes political judgments. In an analysis of the 1968, 1972, and 1980 National Election Studies (NES), Lau (1982, 1985) found that reasons offered by respondents for disliking a candidate were more predictive of overall evaluations than were reasons for liking a candidate. The 1984, 1988 and 1992 NES surveys each included a trait inventory of positively-worded attributes, and respondents were asked how well each attribute described the candidate. Thus, respondents could indicate a perceived character strength by saying that the trait was descriptive of the candidate, or they could indicate a perceived weakness by saying that the trait was not descriptive of the candidate. Klein (1991, 1996) examined the relative weighting of these candidate attributes and found consistent evidence for negativity across the three elections. Perceived candidate weaknesses were found to be more predictive of overall evaluations and voting than were perceived strengths.

These analyses support the presence of negativity by comparing traits within a given candidate. For example, in 1992, Bill Clinton’s "honesty" was rated as more negative than his "compassion" and honesty was weighted more heavily than compassion in evaluations of Clinton (Klein 1996). [Note that this does not translate into an election loss for Clinton. The same phenomenon was also present for Bush whose negatives (for example, compassion) were also weighted more heavily than his strengths.] Thus, this procedure for analyzing negativity takes a holistic approach: within a candidate’s personality profile, attributes are weighted based on their relation to the other attributes in the profile. This test of negativity did not assess the impact of perceived attribute differences across the candidates, and therefore did not specifically assess intradimensional comparisons.

If intradimensional strategies are employed, then the difference in judgments of a given attribute across the two candidates should be predictive of overall evaluations. Specifically, if the absolute difference between judgments of an attribute is large, the intradimensional suggests that the attribute should play an important role in overall evaluations. Further, the literature on negativity suggests that such differences will be particularly relevant for the candidate who is the loser on the comparison. For example, in 1992 Bill Clinton was rated as more compassionate than George Bush. If voters compare the two candidates on this attribute (an intradimensional comparison) then this difference should have an impact on overall evaluations, and this difference should be more predictive of evaluations of Bush than of evaluations of Clinton.


H1: The perceived difference of a given attribute across candidates will be a stronger predictor of overall evaluations of the candidate who has the more negative rating on the attribute than of the candidate who has the more positive rating.


The National Election Study surveys from 1984, 1988, and 1992 were analyzed to test the hypothesis. The NES uses a random probability sampling technique to select respondents, thus creating a representative sample of eligible voters in the United States. Data from the pre-election phase of the surveys were analyzed. These interviews were carried out in-person, from September to election day. The analyses presented below included a sample size of 1411 for the 1984 NES, 1265 for the 1988 NES, and 721 for the 1992 NES. [NES data are available from the Center for Political Studies, University of Michigan, 426 Thompson St., Ann Arbor, MI 48106-1248.]

In the 1984 survey, respondents were asked how well each of the following 12 traits described Walter Mondale and Ronald Reagan: "hard-working," "decent," "compassionate," "commands respect," "intelligent," "moral," "kind," "inspiring," "knowledgeable," "sets a good example," "cares about people like you," "and provides strong leadership." In the 1988 survey, respondents were asked to rate George Bush and Michael Dukakis on nine traits: "intelligence," "compassionate," "moral," "inspiring," "provides strong leadership," "decent," "cares about people like you," "knowledgeable," and "honest." In 1992, the trait "decent" was replaced with the trait "gets things done." The same response scale was used in all three elections: respondents indicated that the trait described the candidate: "extremely well," "quite well," "not too well," or "not well at all." [In 1984, a special weekly version of the NES was conducted in which respondents were presented with the trait items but were given the following response options: "a great deal," "somewhat," "a little," and "not at all." Nearly identical relative trait weightings (regression slopes) were obtained with this response scale and the scale used in the standard NES. Thus, relative weightings did not depend on which response scale was used.] Respondents completed the entire inventory for a single candidate, before proceeding to the next candidate.

A thermometer score, which is a measure of overall evaluation of each candidate, was also obtained in the surveys. Respondents were asked to indicate how favorable or unfavorable they felt towards each candidate on a 0 to 100 scale, with higher numbers indicating feelings of warmth or favorability, and lower numbers indicating feelings of coldness or unfavorability.


In order to measure the importance of attribute differences, the absolute difference between ratings of the two candidates was computed for each attribute. (Trait ratings were converted to a 0 to 1 scale.) For example, if a respondent said that "intelligent" described Ronald Reagan "extremely well" (1.00) but that intelligence described Walter Mondale only "quite well" (.67) the respondent’s score on the attribute distinctiveness variable would be .33. The same score would be obtained if the attribute ratings for the two candidates were reversed. This variable provided a measure of the absolute difference (ABDIF) between judgments of the two candidates on the same attribute. The absolute difference was used in the regressions described below in order to allow for an examination of the effect of differencesCregardless of which candidate was viewed as superior or inferior on an attributeCon overall evaluation. The predictive power (slope coefficients) of these absolute differences will then be compared across candidates.

Thus, the regression equations took the following form:

y(thermometer score) = a + b1(intelligent ABDIF) + b2(compassionate ABDIF) + b3(moral ABDIF) bn(nth trait ABDIF) + bn+1 (race) + bn+2(gender) + bn+3(Democrat) + bn+4(Republican) + bn+5(liberal) + bn+6(conservative)

where n is the number of traits rated in a given election, and bn to bn+6 are coefficients for traditional candidate evaluation control variables (Kinder and Abelson, 1981). Six regression models were estimated, one for each candidate in each of the three elections. [The thermometer score was used as the dependent variable, rather than candidate choice, in order to examine the differential impact of the difference between two attributes for each of the competing candidates. A model with candidate choice as a dummy dependent variable wold have included the same independent variables and dependent variable for both the candidates in a given election.]

While all of the traits were included in the regression equations, only those traits on which the candidates differed were used to test the hypothesis. [The mean difference between the candidates on a given attribute had to be larger than plus or minus .02 to be included in the test of the hypothesis.] For example, in 1992 Clinton and Bush were viewed as equally intelligent. Thus, this attribute did not distinguish between the candidates and could not be used to test the hypothesis. There were 9 such attributes on which the candidates did not differ in the three NES surveys analyzed.

The slope coefficients resulting from the regression analyses are presented in Table 1. To test H1, 21 comparisons were possible: 9 on which the Republican candidate was rated most favorably, and 12 on which the Democratic candidate was rated most favorably. To examine the relative magnitude of each attribute’s predictive power, the absolute values of the coefficients were compared. For 16 of the 21 comparisons, the difference in the slope coefficients was in the predicted direction: the absolute value of the slope was larger for the candidate who was the loser on the comparison. Of these 16, four were significantly greater (and one was marginally greater): "compassionate" and "commands respect" in 1984, t(1392)=3.44, p<.001 and t(1392)=2.35, p<.05, respectively; "decent" in 1988, t(1249)=2.02, p<.05; and "compassionate" and "moral" in 1992 , t(705)=3.16, p<.001 and t(705)=1.86, p<.10, respectively. Of the five comparisons not in the predicted direction, only oneC"inspiring" in 1984Cwas significant, t(1392)=2.02, p<.05.

The above analyses were based on pair-wise comparisons of the slope coefficients across candidates. In Figure 1, the absolute values of the slope coefficients from Table 1 are grouped according to whether a candidate is worse than his opponent or better than his opponent on a particular attribute. On average, the slopes for the candidate who is viewed as inferior to his opponent on a given dimension were significantly higher than when the candidate was viewed as superior to his opponent, t(38)=2.06 p<.05. Thus, H1 receives support, both when comparisons are made across individual pairs of slopes, and when slope averages are compared.


The findings show that when voters detect attribute differences, the candidate who loses out in the comparison is harmed more than the candidate who wins is helped. For example, Clinton was viewed as more compassionate than Bush in the 1992 election. This difference was more costly to Bush than it was beneficial to Clinton. Thus, intradimensional comparisons are more likely to be detrimental than beneficial to a candidate’s overall evaluation.

The results of this study support previous findings of negativity in candidate evaluations (Klein 1991, 1996; Lau 1982, 1985), and go beyond previous work by showing that intradimensional differences in candidate haracter are significant predictors of candidate evaluations. Research in marketing has shown that consumers adapt their strategies to the decision-making task and context, and switch between holistic and intradimensional strategies (Payne, Bettman and Johnson 1988). The findings here suggest that, at some point in the election process, voters make within-attribute comparisons across candidates.



Political advertisements often attempt to highlight the negative attributes of a candidate’s opponent. The results presented here suggest that the effects of this type of advertisement may not be symmetrical in affecting judgments of the two candidates. Any evaluation of the effectiveness of negative advertisements should include measures of potential differences in attribute judgments across candidates, and the relationship between these differences and overall judgments of the sponsoring candidate and the opponent.




Abelson , R. P., and A. Levi (1985), "Decision making and Decision Theory," in The Handbook of Social Psychology, 3rd ed., Vol. 1, G. Lindzey, and E. Aronson, eds. New York: Random House.

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

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

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

Hamilton, D. L., and M. Zanna (1972), "Differential Weighting of Favorable and Unfavorable Attributes in Impressions of Personality," Journal of Experimental Research in Personality, 6, 204-212.

Johnson, Michael D., and Christopher P. Puto (1987), "A Review of Consumer Judgment and Choice," in Review of Marketing, Michael J. Houston, ed. Houston, Chicago: American Marketing Association, 236-292.

Kanouse, D. E., and L. R. Hanson (1972), "Negativity in Evaluations," in Attribution: Perceiving the Causes of Behavior, E. E. Jones et al., eds. Morristown, NJ: General Learning Press.

Kinder, D. R., and R. P. Abelson (1981, September), "Appraising presidential candidates: Personality and affect in the 1980 campaign," Paper presented at the meeting of the American Political Science Association, New York.

Kinder, D. R., and D. O. Sears (1985), "Public Opinion and Political Action," in The Handbook of Social Psychology, 3rd ed, Vol. 1., G. Lindzey, and E. Aronson, eds. New York: Random House.

Klein, J. G. (1991), "Negativity Effects in Impression Formation: A Test in the Political Arena," Personality and Social Psychology Bulletin, 17, 412-418.

Klein, J. G. (1996), "Negativity in Impressions of Presidential Candidates Revisited: The 1992 Election," Personality and Social Psychology Bulletin, 22, 289-296.

Lau, R. R. (1982), "Negativity in political perception," Political behavior, 4, 353-378.

Lau, R. R. (1985), "Two explanations for negativity effects in political behavior," American Journal of Political Science, 29, 119-138.

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

Malhorta, N. K. (1982), "Information load and consumer decision making," Consumer Research, 8, 419-430.

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

Payne, J. W. (1976), "Task complexity and contingent processing in decision making: an information search and protocol analysis," Organizational Behavior of Human Performance, 16, 366-387.

Payne, John W., James R. Bettman, and Eric J. Johnson (1988), "Adaptive Strategy Selection in Decision Making," Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 534-552.

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

Russo, J. E., and L. D. Rosen (1975), "An eye fixation analysis of multialternative choice," Memory and Cognition, 3, 267-276.

Russo, J. Edward, and Barbara Anne Dosher (1983), "Strategies for Multiattribute Binary Choice," Journal of Experimental Psychology: Learning, Memory, and Cognition, 9, 676-696.

Skowronski, J. J., and D. E. Carlston (1989), "Negativity and Extremity Biases in Impression Formation: A Review of Explanations," Psychological Bulletin, 105, 131-142.

Weinberger, M. G., C. T. Allen, and M. R. Dillon (1980), "Negative Information: Perspectives and Research Directions," in Proceedings of the Association for Consumer Research, J. Olson, ed. Ann Arbor, MI: Association for Consumer Research, 398-404.

Weinberger, M. C., and W. R. Dillon (1986), "The Effects of Unfavorable Product Rating Information," Advances in Consumer Research, 13, 528-532.