Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Jewish Leaders
Adam J. Berinsky
and
Tali Mendelberg
Department of Politics
Princeton University
Princeton, NJ 08544-1012
(609) 258-4750
talim@princeton.edu
Abstract
We hypothesize that a stereotype can have an indirect impact over judgment even if some of its
pieces are rejected. We test this proposition with a national survey-experiment that describes a
hypothetical candidate either as “Jewish”, “Jewish” and “shady”, or neither “Jewish” nor
“shady”. We find that once cued, a social stereotype trait (“shady”), even though it is rejected as
illegitimate, can activate another, more acceptable political trait (“liberal”) that historically has
been linked with “shady”. In turn, voters cued with the social trait give more weight to the
acceptable political trait in evaluating the candidate, and the candidate’s support suffers,
especially among conservative voters. This indirect influence of discredited stereotypes has
implications for our understanding of the way stereotypes influence political judgments and for
the ability of groups to overcome a legacy of discrimination.
An earlier version of this paper was presented at the Conference on Experimental Methods,
Harvard University, May 11, 2001. The authors are listed in alphabetical order. We thank
Stephen Ansolabehere, Larry Bartels, James Druckman, James Glaser, and Deborah Schildkraut
for their thoughtful feedback, Paul Gerber for dedicated research assistance, Ed Freeland and the
staff at the Princeton University Survey Research Center for conducting the study, and Princeton
University for a research grant.
The Indirect Effects of Discredited Stereotypes: Social and Political Traits in Judgments of
Jewish Leaders
Introduction
Increasingly, research on voting has recognized the crucial role that stereotypes may play
in shaping how voters view and evaluate political candidates. Yet we know little about
differences in the nature and power of stereotypes. Some stereotypes are well known but
disreputable and widely rejected. For example, the notions that Jews are greedy, women are
incapable of governing, and African Americans are born with inferior intelligence were once the
norm but have since passed from the realm of acceptable social and political discourse. Other
stereotypes are equally well known but widely accepted. For example, it is widely assumed that
Jews, women and African Americans are liberal. In this paper we examine the effects of different
kinds of group stereotypes on voters’ jud gments and evaluations of political candidates. In
particular, we ask whether stereotypes may have an indirect impact on political judgments
We focus here on the impact of political and social stereotypes about Jews in
candidate for Vice-President in 2000, suggests that stereotypes of Jews probably do not have a
direct effect on evaluations of candidates, but may have an indirect one. Lieberman was widely
identified by his own campaign and by the media as an observant Jew. While the coverage of
Lieberman’s candidacy was positive, the heavy attention to his Jewish identification, and anti-
Jewish references in other campaigns, prompt us to ask whether voters’ stereotypes of Jews may
We begin with a general discussion of the nature and functions of stereotypes. We then
move to a specific discussion of the role played by stereotypes of Jews. At the heart of the paper,
1
we describe an experiment, embedded in a national telephone survey, designed to test our
hypotheses. We find that direct cues to social stereotype traits are quite powerful despite the
illegitimacy of these traits. Voters respond to these cues indirectly, by giving more weight to
The term “stereotype” refers to a cognitive structure consisting of a category or label, and
its corresponding traits. 1 These traits are linked together into a coherent structure that resides in
long-term memory and can become activated – ready for use – in subsequent judgments (Judd
and Downing 1995). People use stereotypes in general as heuristics or shortcuts – to fill in
information they lack, to arrive at quick inferences and judgments, and to save the effort of
mindfulness (Lippman 1922). Stereotype traits may be correct or incorrect, but notably for our
purpose, even when largely correct, they are over-generalized, resistant to contrary evidence, and
may be applied in error (Allport 1954; Lippman 1922; Nelson, Acker and Manis 1996).
Stereotypes of groups, however, may have something more: an affective charge (Hurwitz
and Peffley 1998, 91; Jussim et al 1995). Often group stereotypes contain “cultural baggage” –
the traces of a history of group inequality and denigration (Allport 1954; Operario and Fiske
1998: 34-35). As a consequence, negative traits continue to reside in people’s stereotype of the
group even after it achieves a significant measure of equality. Though individuals consciously
reject its validity, the stereotype may be activated nevertheless, and used in subsequent
judgments, often without intent or awareness (Banaji and Greenwald 1994; Bargh 1996; Devine
1989; Fiske 1998; Macrae, Milne and Bodenhausen 1994; Mendelberg 2001). People can of
course control the activation of stereotypes they reject, if they are motivated to do so and their
cognitive resources are not taxed (Devine and Vasquez 1998; Fiske and Neuberg 1990; Monteith
1
Other definitions include expectancies and knowledge (Hamilton and Trolier 1986: 133).
2
1993). But overriding automatically activated stereotypes is an arduous task, especially where
stereotypes of social groups are concerned (Nelson, Acker and Manis 1996; Operario and Fiske
1998).
A key feature of group stereotypes is that they are rich cognitive structures whose various
traits are linked in interconnected “associative network” structures (Anderson 1985; Lodge and
Stroh 1993). Once a particular stereotype trait is activated, or brought to working memory, the
other pieces of the stereotype may become available through a process of spreading activation
across the stereotype’s associative network. Thus, it is possible that once a disavowed trait has
been activated, it may activate another, more acceptable trait. Consequently, voters may rely on a
more legitimate component of the stereotype even if they avoid relying on a less legitimate one.
The result may be erroneous and inequitable decision- making about political figures.
We are especially interested in two components of stereotypes about groups: social and
political traits (see also Huddy 1998). Social traits describe social characteristics of groups in an
exaggerated or over-generalized manner. Many of these social stereotype traits are now
considered – at the conscious level – to be violations of the modern norm that groups should be
treated equally in a democracy (Mendelberg 2001). Nevertheless, the impact on social and
political judgments of stereotypes about African Americans and women, which contain traits
such as “violent” or “feminine” respectively, is well documented (Gilens 1999; Gilliam and
Iyengar 2000; Huddy and Terkildsen 1993; Kahn 1994; Kinder and Mendelberg 1995;
Mendelberg 2001; Hurwitz and Peffley 1998; Sigelman et al 1995; Sniderman and Piazza 1993).
representation of the way groups behave in politics. For that reason, they are seen as legitimate
beliefs. Political stereotype traits may refer to the overall liberal or conservative orientation of a
3
group, or to more specific characteristics such as “feminist” (Huddy 1998). Research in political
science has often focused on the impact of political stereotypes of political groups, such as the
Democratic and Republican parties (Lau 1989; Lodge, McGraw and Stroh 1989; Rahn 1993).
But recent scholarship suggests that voters link political and social traits in evaluating members
of social groups active in politics (e.g., using the trait “feminine” to evaluate a “feminist” woman
While social and political traits are distinct, they are linked in people’s minds through the
cognitive structure of the group stereotype. The social and political traits may not be related
logically. Nevertheless, the two are likely to be associated and linked with the group’s name in
the cognitive structure of the stereotype. When a social trait is activated by a political campaign,
a political trait may as well, through its link in the group stereotype.
Voters are likely to view political traits as more legitimate political considerations than
social traits. Thus, while most voters reject social stereotype traits as illegitimate, they are likely
to use political stereotype traits. If an illegitimate social trait is activated it may not receive added
weight in voters’ considerations because it will be consciously rejected by the individual. But –
and here is the key – voters may give added weight to the more legitimate political trait, even if it
does not apply, because the overall structure of the group stereotype has been activated.
hypothetical Jewish politician. As for many social groups, the stereotype of Jews can be divided
into social and political components. In the U.S., the social traits of Jews as greedy and cunning
have very old roots and are well known, but they have been widely repudiated as illegitimate.
The political stereotype trait of Jews as liberals, however, is widely believed and relatively more
4
socially acceptable. 2 If a social trait is cued during a campaign, it may not harm a Jewish
candidate’s support directly, but it may do so indirectly by activating the legitimate trait of Jews
as too liberal. After all, while their contents may be unrelated, the political and social traits,
together with the category label “Jew”, are likely to be linked together in the memory of many
voters. The stereotype of Jews thus might carry a political penalty for Jewish candidates, given
The social stereotype of Jews has a centuries-old history in the U.S. (Dinnerstein 1994;
Higham 1955). According to several surveys conducted since the 1960s, it contains such traits as
“clannish”, “cunning”, and “greedy” (Martire and Clark 1982; Quinley and Glock 1979;
Selznick and Steinberg 1969). While levels of anti-Semitism have declined dramatically over the
past several decades, the exact nature and magnitude of the decline is uncertain (Smith 1993).
But the key for our study is that these social stereotypes of Jews are sufficiently known,
even if they are rejected by a large majority of voters. While it is rare that such stereotypes come
directly into play in electoral politics, there are indications that they still exist in regular political
discourse. On the national level, Patrick Buchanan has made stereotypical references to Jews on
the social dimension, which have been widely condemned (Chanes 1995, 15; Foltin 1997, 142;
Foltin 2000, 151-152; Wertheimer 1995, 52). On a local level we have only suggestive
anecdotes. For example, a recent influx of observant Jews into the small Iowa town of Postville
provoked a petition signed by 126 voters (in a town of 1,600 residents) to oppose the first Jew
appointed to the town council and force him to run in a special election. In a newspaper
interview, the mayor, who opposed the appointment, explained the opposition in classic social-
stereotypic terms: “The Jewish people are not guilt- free on this. And it’s not totally about
2
In our study, when asked whether the statement “Jews are liberal politically” describes Jews, a minority (43%)
5
religion. The typical Jewish person is going to bargain. The typical Jewish person will delay
Though social stereotypes of Jews are discredited, political stereotypes are less so in part
because they appear accurate. Since the beginning of their large-scale involvement in electoral
politics during the 1930s, Jewish citizens and leaders have taken consistently liberal positions on
various issues. Today they continue to be unusually liberal on the separation between church and
state, social issues relating to sexuality, spending fo r social programs, civil rights and
government aid to African Americans (Cohen and Liebman 1997; Glaser 1997). Throughout the
1980s, 40% to 50% of Jews identified as liberal, a rate two to three times higher than that of
The belief that Jews are liberal is accurate. But if it functions as a component of voters’
overall stereotype of Jews, it might not be used accurately and judiciously in all cases.
Lieberman, for example, has taken moderate and even some conservative positions on a number
of policies. As the New York Times reported after the 2000 Democratic Convention, “behind the
traditional Democratic oratory of his convention speech is a senator who has described himself
as pro-business, pro-trade and pro-econo mic growth" (New York Times 8/27/00). As a Senator,
Lieberman actively promoted the interests of the insurance, high technology and health care
6
industries. He has also made a national name advocating traditional morality in popular culture
and criticizing President Clinton’s extra- marital affairs. His overall record is much more
accurately classified as moderate than liberal. If the stereotype of Jews has potency, however,
people may judge a candidate like Lieberman as more liberal than he really is.
More generally, the political trait “liberal”, if it functions as part of a stereotype of Jews
that exaggerates and oversimplifies reality, may lead to perceptions of Jewish leaders as liberal,
even when their actual positions indicate that they are not. A leader whose Jewish membership is
emphasized may be perceived as much more liberal than he is in fact. Since many voters do not
identify as liberal, being perceived as a liberal is likely to carry a political liability for leaders.
In sum, while most voters ma y reject a message that plays on the stereotypic social traits
of Jews, that message may nevertheless prompt some voters to give greater weight to their
stereotypic political trait of Jews as too liberal – even if the message makes no allusion to the
political trait. Negative, discredited social traits may not be used in evaluations of the candidate
because they are deemed an unacceptable basis for judgment. However, the activation of an
illegitimate social trait may in turn activate the more acceptable political trait. The candidate may
lose support from the latter even as he escapes a penalty from the former.
Method
political leaders. For this purpose, we asked respondents to evaluate a fictitious candidate,
Howard Wilson. 5 Respondents to a national phone survey were randomly read one of six short
4
Also, during the past several decades Jews have been significantly more loyal to the Democratic party than
Catholics or Protestants (Abramson, Aldrich and Rhode 1998, 110).
5
Respondents, of course, were not told that Wilson was not a real politician.
6
Note that the three variables are not fully crossed with each other. The survey was conducted by the Survey
Research Center at Princeton University. The survey was fielded from early February to April 2001 using random-
digit dialing and a national voting-age sample. The sample characteristics are displayed in the Appendix.
7
The profiles were identical across the six conditions except for the words that related to
the stereotypical dimensions of Jews (see Table 1). 7 The first dimension concerned the social
stereotypes of Jews. In the control condition (“no-label”) Wilson had no religious or ethnic
affiliation and was reported to be facing a lawsuit “filed against him by a business client who
accuses Wilson of neglecting and ignoring his business practices”. In the “label-only” condition,
Wilson was identified as “Jewish” (among a list of other, neutral characteristics), and was
reported to be facing the same lawsuit by a business “client ” over “neglecting and ignoring” his
business practices. In the social stereotype condition, Wilson was identified as “Jewish” (in the
same way as in the “label-only” condition) but was also reported to be facing a lawsuit filed “by
a business competitor who accuses Wilson of engaging in shady business practices to get
ahead.”
(Table 1 Here)
manipulation concerning political stereotypes of Jews. One- half of the respondents were given a
somewhat liberal issue profile of Wilson. Specifically, they were told “Wilson has voted to raise
spending on assistance to the poor, to increase government involvement in health care and for a
tax relief plan for middle-class families.” The remaining respondents were given a somewhat
conservative issue profile: “Wilson has voted to reduce spending on assistance to the poor, to
decrease government involvement in health care and for a tax relief plan for middle-class
families.”
These six conditions effectively manipulate three variables. First, they provide or
withhold the Jewish label. Specifically, two-thirds of the respondents were told that Wilson is
Jewish; the other third were given no informatio n concerning Wilson’s religious practices.
7
A randomization check is provided in the Appendix.
8
Second, the conditions either introduced or withheld a cue relating to the social stereotype of
Jews as a group that practices shady business dealings in order to make more money.
Specifically, one-half of the respondents who were told that Wilson is Jewish (one-third of all
respondents) were also told that he is accused of being “shady.”8 Third, we either cued or
countered the political stereotype of Jews as liberal. In particular, one half of respondents was
given information indicating that he is; the other half was given information indicating that he is
a conservative.
condition that cued the social stereotype without the Jewish label. We are highly skeptical that
such a condition would, in fact, measure the impact of the stereotypical dimension (greed and
dishonest business practices) independent of the Jewish identity of the candidate. A person of
unidentified ethnicity who is accused of using shady business practices to advance his business at
the expense of competitors may well be perceived as a Jew. We could of course proceed with
such a condition and simply ask respondents whether they perceived the politician as a Jew,
Catholic, Protestant, or something else. But this question probably underestimates respondents’
true perception of the candidate’s Jewish ethnicity. Instead, we have chosen to forego a no- label,
social stereotype condition and rely on a pattern of evidence to test the link between the social
After hearing one of the six candidate profiles, respondents were asked to evaluate
Wilson in general and to place him on a number of specific political attitude and trait scales. 9 For
comparison purposes they were also asked to place themselves and several real political figures
on some of these scales (including Joseph Lieberman). At the end of the questionnaire we also
8
It is important to note here that we hold the presence of a potential scandal constant across conditions and vary the
type of scandal.
9
All items have been used in major data collection efforts.
9
measured whether the respondent accepts or rejects various stereotypes of Jews (using
established questions developed for this purpose and extensively tested on national samples).
Results
Our main expectation is that the mere invocation of the ”Jewish” label will activate the
stereotypic political traits attached to the stereotype of Jews. Moreover, we expect that the
invocation of the social stereotypic traits will increase the activation of the overall Jewish
stereotype, political traits included, even if the respondent consciously rejects that stereotype.
Therefore, we expect that 1) the description of the Jewish candidate that cues the stereotype will
not cause people to judge the candidate in more socially stereotypic terms, but 2) will cause them
to judge the candidate in more politically stereotypic terms. Specifically, our hypothesis is that
Wilson will be perceived as more liberal in the “Jewish label” profile than in the no- label
condition and more liberal still in the “shady Jewish label” condition. If the link between the
social and political stereotypes is sufficiently strong, then we will see this stereotype activation
even in the profile that counters the political stereotype with information that the candidate is, in
fact, politically conservative. Real political ramifications may follow. If people perceive a
socially stereotyped candidate as more politically stereotypical (that is, as more liberal), the
candidate’s support may suffer among the majority of voters, who are moderate or conservative.
As a first step, we ask whether the Jewish label, with or without the social stereotype,
leads voters to think in more socially stereotypical terms about the candidates. 10 We expect it
will not. We regressed the perception of the candidate as greedy and the perception of him as
dishonest (coded on the 0-1 interval) on a dummy variable for the “Label-only” condition and
another dummy variable for the “shady” condition (see Appendix for wording). Neither the
10
Jewish label nor the label with the stereotype cue lead people to judge Wilson as either more
dishonest or more greedy. 11 In all cases, the coefficients are insignificant in both a statistical and
a substantive sense. Thus, as expected, the experimental treatments do not affect respondents’
(Table 2 Here)
More crucially, however, we expect these ethnic cues to have important political
consequences through the more indirect judgment of the candidate’s political ideology. Our
second step therefore is to test the hypothesis that cuing the Jewish stereotypes – through the
“label only” and the “shady” conditions – will activate the political stereotypic trait of liberal. 12
Table 2 presents the means and standard errors in each of the six conditions. Regardless of
whether Wilson is portrayed as liberal or conservative, cuing the Jewish stereotype increases his
perceived liberalism. Merely identifying Wilson as Jewish pushes the respondents’ impression of
the candidate toward the liberal end of the seven-point scale. Adding the “shady” trait pushes
Wilson leftward still. Voters, it seems, are indeed quite ready to rely on the political stereotypic
trait even when provided clear, direct information about the candidate’s ideology through his
issue positions. 13
10
All the analyses performed in this paper were conducted using only respondents who did not identify themselves
as Jewish. Including the 15 Jews in the analyses does not change the results.
11
The OLS coefficients, with standard errors in parenthesis, are as follows: GreedyCandidate = .40 + .02 (.03)
Label-Only + .03 (.03) Shady; n = 599, adjusted R-squared = -.002. DishonestCandidate = .40 + .001 (.03) Label-
Only - .01 (.03) Shady; n = 623, adjusted R-squared = -.003.
12
The analyses used here make use of a seven-point branching liberal-conservative scale, where 7 is the most liberal
response, and 1 is the most conservative response. We get the same results if we use only the first branch of the
question, and when we use ordered probit instead of OLS. By way of calibrating the results, the ideology ratings of
Joe Lieberman and Dick Cheney are 4.97 and 2.52, respectively.
13
The impact of the “shady” condition is equivalent to the impact of moving from a moderately conservative to a
moderately liberal candidate profile in the control conditions. However, we do not wish to over-emphasize the
comparison between the impact of candidate ideology and candidate social trait: by design we avoided extreme
variance in the manipulation of candidate ideology, making it rather moderate in all conditions, in order to
generalize to most contests. We also note that the fact that candidate ideology has an impact suggests that our
ideology manipulation worked well – voters saw a statistically significant but moderate difference between two
moderately different ideological profiles.
11
Recall that we also manipulated the political stereotype trait – we activated it with a
liberal issue profile or countered it with a conservative set of issues. The question now is whether
candidate information that counters the political stereotype trait moderates the impact of the
social trait. If social stereotype traits have limited power, then when we tell voters that the
candidate holds conservative issue positions, the impact of the social trait should diminish. To
find out, we regressed perception of the candidate’s ideology on a pair of dummy variables
representing the “Label-only” and “shady” conditions, separately for the liberal and conservative
(Table 3 Here)
candidate is no less susceptible than a liberal candidate to the effects of social stereotyping. A
liberal candidate is perceived as somewhat more liberal merely for being labeled a Jew, and as
slightly more liberal still if he is portrayed as a “shady” Jew (the effects are in the right direction
but are not statistically significant from the control candidate). Among the conservative
candidates, the Jewish and “shady” Jewish candidates are each perceived as substantially more
liberal than their counterpart in the control condition (Table 3 shows that each is statistically
significant relative to the control candidate; the difference between them is not statistically
of the “shady” social stereotype trait, and may smooth the way for stereotype activation (though
the last column in Table 3 shows that the differences between the effects on the liberal and
14
The effects in Table 3 obtain among both liberal and conservative respondents.
12
An Alternative Hypothesis: Perhaps They are Shady if They are Liberal, Jewish or Not
However, perhaps what we would interpret as the effect of activating the stereotype on
perceptions of ideology is a spurious effect. People may not associate the Jewish label and its
stereotypical social traits with liberal ideology, as we hypothesize. Rather, our respondents may
instead associate liberals (of all ethnic backgrounds) with dishonesty or greed. Perhaps if we had
asked our respondents about a Protestant “shady” candidate, they would have rated him as liberal
too. In that case, it is not the stereotypic link between social and political traits that causes the
the candidate’s greed and dishonesty in the liberal and the conservative control conditions (see
Appendix for wording). If people associate liberal positions with greed or dishonesty without
consulting their stereotype of Jews, then perceptions of the candidate’s greed and dishonesty
should be higher in the liberal than the conservative conditions that do not refer to Jews.
The results suggest otherwise. With perceptions coded on the 0 to 1 interval, the
unstandardized OLS estimate on a dummy variable coded 1 for the liberal control condition and
0 for the conservative control condition is -.05 for “greedy,” with a standard error of .04. The
results for the “dishonest” item are even clearer; there, the coefficient on the dummy variable is
.003, with a standard error of .04. The liberal non-Jewish Wilson is perceived to be no more
greedy or dishonest than the conservative non-Jewish Wilson. There is no evidence for the
alternative hypothesis that voters associate a candidate’s greed and dishonesty with liberal
ideology, his ethnic membership notwithstanding. Thus it seems that the causal story
emphasizing the link between social and political traits specific to stereotypes of Jews is correct.
13
The Impact of Candidate’s Perceived Liberal Ideology on Candidate Evaluation
incur a political penalty from stereotypes. If the stereotype is in fact affectively charged,
activating it should lead to a negative judgment, not merely to a biased perception of ideology.
Does the Jewish candidate lose support through his perceived ideology because he is Jewish?
And if so, is it merely because of his Jewish identity, or only when he is also attacked as a
socially stereotypic Jew? Finally, does the candidate’s actual conservative position moderate the
To find out, we regressed the feeling thermometer for the candidate on perception of the
candidate’s ideology in each of the six conditions. 15 The common sense of party politics suggests
that the penalty paid by the Jewish candidate applies primarily among people who do not identify
themselves as liberal. After all, for liberal voters, the candidate’s liberal ideology would not be a
liability (and might be an asset). Not so for conservative voters, for whom the candidate’s
perceived liberal ideology should constitute grounds for rejection. To find out whether
conservative and liberal voters react in these opposite ways, we estimated the model separately
(Table 4 Here)
(Table 5 Here)
The OLS coefficients and standard errors are presented in Table 4, and the corresponding
estimated means in Table 5. Recall that many voters tend to misperceive the socially stereotyped
candidate in more politically stereotyped terms. The results in Tables 4 and 5 suggest that
conservative voters go the next step and give more weight to their exaggerated perception. Table
4 shows that in comparison with the control and “label-only” conditions, the “shady” trait leads
14
conservative voters to give much more weight to their perception of the candidate’s ideology.
This holds for the liberal candidate (b = -4.78, se=1.42; difference from “label-only” is
significant at p < .05) and especially for the conservative candidate (-6.94, se=1.41; differences
from “label-only” and from control are each significant at p < .05). It is the cue to the social
stereotype trait of “shady,” and not the mere Jewish label, that sets indirect stereotyping in
motion. The candidate’s actual conservative positions do little to inhibit indirect stereotyping
(Table 4 shows that there is no statistically significant difference between the liberal and
conservative candidates). Table 5 makes clear what these effects mean: the “shady” Jewish
liberal. A “shady” candidate who manages to avoid the perception that he is liberal essentially
inhibits the process of indirect stereotyping and suffers few ill effects.
Liberal voters’ evaluations grow colder with the social stereotype too, though in a
different way. Tables 4 and 5 show that when the conservative candidate is not Jewish, his
perceived liberalism is a reason to support him (b = 4.98, se = 1.91). But when he is a “shady”
Jew, his perceived liberalism ceases to be a plus and may become a reason to oppose him (the
coefficient takes a negative sign though it is not statistically significant, b = -1.77, se = 1.68). His
evaluations plummet from 61 (for a non-Jew) to 34 (for a “shady” Jew). Again, the mere Jewish
label carries no statistically significant ill effects relative to the control condition; only the
“shady” trait does so. By contrast with the conservative candidate, the liberal candidate suffers
no ill effects from the social stereotype (Table 4 shows that the difference between the
stereotyped liberal candidate were to try to shift his actual policy positions in a more
15
The model includes six dummy variables representing the six conditions, and six variables representing the
perceptions of candidate ideology multiplied by each dummy variable. We estimated the model without a constant,
for ease of interpretation.
15
conservative direction (and came to resemble the conservative candidate in our study), he would
Let us summarize these results. 17 We found evidence for the indirect effect of a social
stereotype cue. Conservatives not only see the “shady” Jewish candidate as more liberal than his
non-Jewish counterpart, but also find the candidate’s liberal ideology much more objectionable.
The mere label does not carry ill effects; the candidate suffers a more severe penalty for his
Notably, the socially stereotyped candidate only suffers the loss of support indirectly, through a
stereotyped perceived ideology. The process of stereotyping is indirect: the social trait (“shady”)
activates a stereotyped political trait (“liberal”), which in turn decreases support. Without the
perceived political trait, the social cue does nothing to erode the candidate’s support. Also
notable is the inability of the candidate who takes actual conservative positions to escape the
consequences of indirect stereotyping. Finally, we found that liberal voters also respond to the
social stereotype cue, but only in evaluating a conservative candidate. Indirect stereotyping
Conclusion
In this paper we have attempted to gauge whether negative stereotypes of a social group
may work in a subtle yet potent fashion. We focused on stereotypes of Jews as a difficult test of
this hypothesis. Surveys have shown consistently that negative stereotypes of Jews have declined
dramatically in the United States. Yet the results here suggest that when campaigns cue
16
Validating the intuition that liberals prefer someone they perceive as liberal while conservatives prefer the
converse, the effects in the control conditions show that the impact of the candidate’s perceived liberal ideology is
negative among conservatives, and positive among liberals. But this effect does not hold among liberal voters
evaluating the liberal candidate. That makes for a problematic comparison with the effects on liberals who are
evaluating a liberal Jewish or shady candidate.
17
By way of calibrating the effects, one might contrast them with the impact of Lieberman’s and Cheney’s
perceived ideology on their respective feeling thermometer ratings (-3.09, se = .56, and 1.33, se = .70 respectively).
We get the same results when we recode the feeling thermometer to three categories (positive, 51-100; neutral, 50;
negative, 0-49), and stronger results when we focus on voters who gave feeling thermometer ratings other than 50.
16
stereotypic social traits – even those widely discredited – we may witness an indirect process of
the sense that stereotypic social and political traits are linked together within a cognitive
structure that resides in memory. The discredited social traits are activated but controlled, while
the political, more legitimate political trait is activated but not over-ridden. Because the social is
linked to the political, it can carry an important indirect consequence for political judgments.
The process of stereotyping, however, is not only cognitive but also affective. Some of
the social stereotype traits of social groups with a history of disadvantage are negative. We found
that only when a negative social trait is cued does the target of the stereotype suffer adverse
candidacy does not fit our “label-only” simulation, since his campaign and press coverage
highlighted not only the Jewish label but also the socially stereotypic trait of religiosity. Neither
does his candidacy clearly fit our social trait condition. The trait of religiosity, while social, is
much more positive than the social trait “shady.” Lieberman was portrayed as a deeply religious
man with the associated positive characteristics of moral integrity and attachment to the valued
institutions of family and place of worship. There was a small exception, however, that hints at
the kind of campaign we simulated here. For a brief time, Lieberman was cast by the Republican
National Committee as “Slumlord Joe” for his role in evicting a tenant from a building he
administered for his “rich uncle’s” estate. 18 Talk radio host Rush Limbaugh charged him with
18
The Hartford Courant, 9/14/2000, “Records Refute GOP ‘Hypocrite” Claim, by Mike McIntire.
17
hypocrisy in social stereotypic terms (“can you spell ‘chutzpah,’ Joe?!”). 19 Still, this episode was
What we can say with some confidence is that absent positive information about a Jewish
candidate, intensive attention to a candidate’s Jewish ethnicity is likely to pose a liability with an
electorate composed of more than trivial numbers of conservative voters. Moreover, if a Jewish
candidate were to run against a candidate in the mold of Patrick Buchanan (who has made
consistent statements unfavorable to Jews), or in an environment rife with tensions between Jews
and others (as in the Iowa town of Postville, or at certain moments in New York City), the
candidate may find himself characterized in negative stereotypical terms. It may not be the
opposing candidate who does so; cues might present themselves in the course of casual exchange
among ordinary political discussion partners, on talk radio, or in internet forums. Regardless of
their source, if such cues are available, anti-Jewish stereotypes may well become activated. The
Notably, however, the process by which Jewish candidates’ support can erode is an
indirect one, and thus potentially difficult to counter. Voters do not rely on social stereotypic
reasons for opposing a stereotyped candidate. In our study, voters did not reject the candidate
because he was a dishonest or greedy person. Rather, voters rely on their misperceived reading
of the candidate’s ideology. The cue to the negative social trait sets in motion a process by which
the candidate is perceived as more liberal, and in turn judged more harshly for it.
But the temptation that this presents for candidates – to inoculate themselves by moving
rightward – is probably a false one. Most likely, a candidate’s attempt to present a moderately
conservative policy profile will fail, and may even backfire. The results presented here suggest
that moderately conservative candidates can no more escape the effects of stereotyping than can
19
http://www.rushonline.com/visitors/goreism.htm
18
moderately liberal candidates. Thus the strategy that may seem most sensible is unlikely to
succeed.
A strategy that these results can recommend takes as its point of departure the indirect
nature of the stereotype process. If voters come to understand that their response derives from
stereotypical social traits that they reject, they may be more motivated to override the stereotype.
The literature on stereotypes provides some hope in this regard. When people are motivated to
interfere with the activation of their own stereotypes, they can often weaken the impact of their
stereotypes on their judgments, if not completely eradicate it (e.g., Bodenhausen and Macrae
1998; Devine 1989; for a somewhat more pessimistic conclusion, see Nelson, Acker and Manis
1996).
Today, members of social groups with a history of disadvantage are protected from harm
by virtue of their citizenship in a democracy. But those among them who seek to join the
governing class may, under specific circumstances, continue to face obstacles rooted in social
suspicio n. The key to the remedy is to understand those circumstances and to investigate the
19
References
Abramson, Paul, John Aldrich and David Rhode. 1998. Change and Continuity in the 1996
Elections. Washington, DC: Congressional Quarterly Press.
Anderson, John R. 1985. Cognitive Psychology and Its Implications, 2nd ed. New York: W.H.
Freeman.
Banaji, M. R., and A. G. Greenwald. 1994. “Implicit Stereotyping and Prejudice.” In The
Psychology of Prejudice: The Ontario Symposium, ed. M.P. Zanna and J.M. Olson.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Chanes, Jerome A. 1995. “Antisemitism and Jewish Security in America Today: Interpreting the
Data.” In Antisemitism in America Today: Outspoken Experts Explode the Myths. Ed.
Jerome A. Chanes. New York: Birch Lane Press. Pp. 3-32.
Cohen, S. M., and C. S. Liebman. 1997. “American Jewish Liberalism: Unraveling the Strands.”
Public Opinion Quarterly 61:405−430.
Devine, Patricia G. 1989. “Stereotypes and Prejudice: Their Automatic and Controlled
Components.” Journal of Personality and Social Psychology 56:5−8.
Devine, Patricia G., and Kristin A. Vasquez. 1998. “The Rocky Road to Positive Intergroup
Relations.” In Confronting Racism: The Problem and the Response, ed. Jennifer Lynn
Eberhardt, and Susan T Fiske. Thousand Oaks, CA: Sage Publications, Inc.
Dinnerstein, Leonard, and Mary Dale Palsson. 1973. “Introduction.” In Jews in the South, ed.
Leonard Dinnerstein and Mary Dale Palsson. Baton Rouge: Louisiana State University
Press.
Fiske, Susan T. 1998. “Stereotyping, Prejudice, and Discrimination.” In The Handbook of Social
Psychology, 4th ed., ed. Daniel T. Gilbert, Susan T. Fiske, and Gardner Lindzey. New
York: McGraw Hill.
Fiske, Susan T., and Steven L. Neuberg. 1990. “A Continuum Model of Impresssion Formation,
from Attention and Interpretation.” In Advances in Experimental Social Psychology, ed.
Mark P. Zanna. New York: Academic Press.
20
Foltin, Richard T. 1997. “National Affairs.” In American Jewish Year Book 1997, ed. David
Singer and Ruth R. Seldin. New York: The American Jewish Committee.
Foltin, Richard T. 2000. “National Affairs.” In American Jewish Year Book 2000, ed. David
Singer and Lawrence Grossman. New York: The American Jewish Committee.
Gilens, Martin. 1999. Why Americans Hate Welfare. Chicago: University of Chicago Press.
Gilliam Frank D., and Shanto Iyengar. 2000. “Prime Suspects: The Influence Of Local
Television News On The Viewing Public.” American Journal of Political Science.
44:560−573.
Glaser, James M. 1997. Toward an Explanation of the Racial Liberalism of American Jews.
Political Research Quarterly 50:437−458.
Glazer, Nathan. 1995. “The Anomalous Liberalism of American Jews.” In The Americanization
of the Jews, ed. Robert M. Seltzer, and Norman J. Cohen. New York: New York
University Press.
Hamilton, David L., and Tina K. Trolier. 1986. “Stereotypes and Stereotyping: An Overview of
the Cognitive Approach.” In Prejudice, Discrimination, and Racism, ed. John F.Dovidio,
and Samuel L. Gaertner. San Diego, CA: Academic Press, Inc.
Higham, John. 1955. Strangers in the Land. New Brunswick: Rutgers University Press.
Huddy, Leonie.1998. "The Social Nature of Political Identity: Feminist Image and Feminist
Identity." Paper presented at the annual meeting of the American Political Science
Association, Boston, September, 1998.
Huddy, Leonie, and Nayda Terkildsen. 1993. “Gender Stereotypes and the Perception of Male
and Female Candidates.” American Journal of Political Science 37:119−147.
Hurwitz, Jon and Mark Peffley. 1998. “Whites’ Stereotypes of Blacks: Sources and Political
Consequences. In Jon Hurwitz and Mark Peffley (eds.), Perception and Prejudice: Race
and Politics in the United States. New Haven: Yale University Press.
Jaher, Frederic Cople. 1994. A Scapegoat in the New Wilderness: The Origins and Rise of Anti-
Semitism in America. Cambridge, Massachusetts: Harvard University Press.
Judd, Charles M. and James W. Downing. 1995. “Stereotypic Accuracy in Judgments of the
Political Positions of Groups and Individuals.” in Political Judgment: Structure and
Process, ed. Milton Lodge and Kathleen M. McGraw. Ann Arbor: The University of
Michigan Press.
Jussim, Lee, Thomas E. Nelson, Melvin Manis and Sonia Soffin. 1995. Prejudice, Stereotypes,
and Labeling Effects: Sources of Bias in Person Perception. JPSP 68: 228-246.
21
Kahn, Kim F. 1994. “Does Gender Make a Difference? An Experimental Examination of Sex
Stereotypes and Press Patterns in Statewide Campaigns.” American Journal of Political
Science 38(1): 162-195.
Kinder, Donald R. and Tali Mendelberg. 1995. “Cracks in American Apartheid.” Journal of
Politic 57: 402-424.
Lau, Richard. 1989. “Political Schemata, Candidate Evaluations, and Voting Behavior.” In
Political Cognition: The 19th Annual Carnegie Symposium on Cognition. Richard R. Lau
and David O. Sears (eds.). Hillsdale, NJ: Lawrence Erlbaum.
Layman, Geoffrey C. 1997. “Religion and Political Behavior in the United States: The Impact of
Beliefs, Affiliations, and Commitment from 1980 to 1994.” Public Opinion Quarterly
61:288−316.
Lippmann, Walter. 1922. Public Opinion. New York: Harcourt, Brace, and Co.
Lipset, Seymour Martin, and Earl Raab. 1995. Jews and the New American Scene. Cambridge,
Massachusetts: Harvard University Press.
Lodge, Milton, Kathleen M. Mcgraw, and Patrick Stroh. 1989. “An Impression Driven Model Of
Candidate Evaluation.” American Political Science Review 83:399−420.
Lodge, Milton and Patrick Stroh. 1993. “Inside the Mental Voting Booth: An Impression-Driven
Process Model of Candidate Evaluation.” In Explorations in Political Psychology, ed.
Shanto Iyengar, and William J. McGuire. Durham: Duke University Press.
Macrae, C. Neil, Alan B. Milne, and Galen V Bodenhausen. 1994 “Stereotypes as Energy Saving
Devices: A Peek Inside the Cognitive Toolbox.” Journal of Personality and Social
Psychology 66:37−47.
Martire, Gregory and Ruth Clark.1982. Anti-Semitism In the United States. New York: Praeger.
Mendelberg, Tali. 2001. The Race Card: Campaign Strategy, Implicit Messages, and the Norm
of Equality. Princeton, NJ: Princeton University Press.
Monteith, Margo J. 1993. “Self- Regulation of Prejudiced Responses: Implications for Progress in
Prejudice-Reduction Efforts.” Journal of Personality and Social Psychology 65:469−485.
Nelson, Thomas E., Michele Acker and Melvin Manis. 1996. Irrepressible Stereotypes. Journal
of Experimental Social Psychology 32: 13-38.
Operario, Don and Susan T. Fiske. 1998. “Racism Equals Power Plus Prejudice: A Social
Psychological Equation for Racial Oppression. In Confronting Racism: The Problem and
the Response. Jennifer L. Eberhardt and Susan T. Fiske (eds.). Thousand Oaks, CA: Sage.
Pp. 33-53.
22
Perdue, Charles W., John F. Dovidio, Michael B. Gurtman, and Richard B Tyler. 1990. “Us and
Them: Social Categorization and The Process of Intergroup Bias.” Journal of Personality
and Social Psychology 59:475−486.
Quinley, Harold E., and Charles Y. Glock. 1979. Anti-Semitism in America. New York: The Free
Press.
Rahn, Wendy M. 1993. “The Role of Partisan Stereotypes in Information Processing about
Political Candidates.” American Journal of Political Science 2:472−496.
Selznick, Gertrude J. and Stephen Steinberg. 1969. The Tenacity of Prejudice. New York: Harper
and Row.
Sigelman, Lee. 1991. “’If You Prick Us, Do We Not Bleed? If You Tickle Us, Do We Not
Laugh?’ Jews and Pocketbook Voting.” Journal of Politics 53: 977-92.
Sigelman, Carol K., Lee Sigelman, Barbara J. Walkosz, and Michael Nitz. 1995. “Black
Candidates, White Voters: Understanding Racial Bias in Political Perceptions.” American
Journal of Political Science 39:243−265.
Smith, Tom W. 1993. “Actual Trends or Measurement Artifacts? A Review of Three Studies of
Anti-Semitism.” POQ 57: 380-93.
Sniderman, Paul and Thomas Piazza. 1993. The Scar of Race. Cambridge: Harvard University.
23
Appendix I: Candidate Profiles
Now we are going to ask you some questions to see how voters such as yourself view candidates
for public office. I'm go ing to read you a description of a politician. Please listen and remember
what you can from the description.
Profile 1. Howard Wilson is a state representative who is considering running for an open seat
in Congress. Wilson is a 48 year-old businessman who is married with two children. He has
served in the state Senate for two terms. While in the Senate, Wilson has voted to raise spending
on assistance to the poor, to increase government involvement in health care and for a tax relief
plan for middle-class families. As a state senator he worked to expand his state's economic ties
with several countries overseas. One potential problem for Wilson is a lawsuit filed against him
by a business client who accuses Wilson of neglecting and ignoring his business practices,
causing the client to lose a great deal of money. Wilson has denied the charge.
Profile 2. Howard Wilson is a state representative who is considering running for an open seat
in Congress. Wilson is a 48 year-old Jewish businessman who is married with two children. He
has served in the state Senate for two terms. While in the Senate, Wilson has voted to raise
spending on assistance to the poor, to increase government involvement in health care and for a
tax relief plan for middle-class families. As a state senator he worked to expand his state's
economic ties with several countries overseas. One potential problem for Wilson is a lawsuit
filed against him by a business client who accuses Wilson of neglecting and ignoring his
business practices, causing the client to lose a great deal of money. Wilson has denied the
charge.
Profile 3. Howard Wilson is a state representative who is considering running for an open seat
in Congress. Wilson is a 48 year-old Jewish businessman who is married with two children. He
has served in the state Senate for two terms. While in the Senate, Wilson has voted to raise
spending on assistance to the poor, to increase government involvement in health care and for a
tax relief plan for middle-class families. As a state senator he worked to expand his state's
economic ties with several countries overseas. One potential problem for Wilson is a lawsuit
filed against him by a business competitor who accuses Wilson of engaging in shady business
practices to get ahead, causing the competitor to lose a great deal of money. Wilson has denied
the charge.
Profile 4. Howard Wilson is a state representative who is considering running for an open seat
in Congress. Wilson is a 48 year-old businessman who is married with two children. He has
served in the state Senate for two terms. While in the Senate, Wilson has voted to reduce
spending on assistance to the poor, to decrease government involvement in health care and for a
tax relief plan for middle-class families. As a state senator he wo rked to expand his state's
economic ties with several countries overseas. One potential problem for Wilson is a lawsuit
filed against him by a business client who accuses Wilson of neglecting and ignoring his
business practices, causing the client to lose a great deal of money. Wilson has denied the
charge.
24
Profile 5. Howard Wilson is a state representative who is considering running for an open seat
in Congress. Wilson is a 48 year-old Jewish businessman who is married with two children. He
has served in the state Senate for two terms. While in the Senate, Wilson has voted to reduce
spending on assistance to the poor, to decrease government involvement in health care and for a
tax relief plan for middle-class families. As a state senator he worked to expand his state's
economic ties with several countries overseas. One potential problem for Wilson is a lawsuit
filed against him by a business client who accuses Wilson of neglecting and ignoring his
business practices, causing the client to lose a great deal of money. Wilson has denied the
charge.
Profile 6. Howard Wilson is a state representative who is considering running for an open seat
in Congress. Wilson is a 48 year-old Jewish businessman who is married with two children. He
has served in the state Senate for two terms. While in the Senate, Wilson has voted to reduce
spending on assistance to the poor, to decrease government involvement in health care and for a
tax relief plan for middle-class families. As a state senator he worked to expand his state's
economic ties with several countries overseas. One potential problem for Wilson is a lawsuit
filed against him by a business competitor who accuses Wilson of engaging in shady business
practices to get ahead, causing the competitor to lose a great deal of money. Wilson has denied
the charge.
25
Appendix II: Sample Characteristics, Variables and Randomization
Checks
Total N = 756. AAPOR’s RR 3 response rate is 16%.
20
The 2000 NES, unlike our study, provided an option of “haven’t thought about it much.” Most likely, those who
selected this answer chose moderate or don’t know in our study.
26
Continued.
Religious affiliation
Protestant 39.4 35.2
Roman Catholic 23.7 23.3
Jewish 2.9 1.8
Other religion 16.5 19.6
No religion 16.1 19.6 21
Refusal/don’t know 1.3 0.4
Religion, important to respondent
Extremely 35.2 38.0
Very 27.0 21.9
Somewhat 23.8 16.2
Not too 12.3 23.6
Don’t know 0.9 0.1
Refusal 0.8 0.2
Female 56.0 56.3
Politically informed
Blair 45.0 33.9
Rehnquist 36.1 9.6
Majority in the House 62.2 54.6
Conservative –
1 2 3 4 5 6 7
Liberal Scale
Percentage of
9.3 19.1 19.4 3.4 18.8 21.0 9.0
Respondents
Note: 7 is the most liberal response, and 1 is the most conservative response
Thermometer
0-24 25-49 50 51-74 75-100
Ratings
Percentage of
15.2 23.9 40.7 13.6 6.6
Respondents
21
The NES asked whether the respondent had an affiliation at all first, then followed up with which affiliation. Our
questionnaire, by comparison, provided the no affiliation option in the same question as which affiliation.
27
Randomization Check
The number of subjects is randomly distributed across conditions, with cell size ranging from
122 to 130. Every relevant variable is randomly distributed across conditions with the exception
of education. When we included education in our basic models the results were substantially the
same as those we report in the text.
1. I am going to read a list of words and phrases people may use to describe Howard
Wilson. For each, tell me whether the word or phrase describes him extremely well, quite
well, not too well, or not well at all. What about DISHONEST?
2. We hear a lot of talk these days about liberals and conservatives. When it comes to
politics, do you usually think of yourself as liberal, moderate or middle of the road or
conservative? (IF LIBERAL) Would you say that you are extremely liberal or just
somewhat liberal? (IF CONSERVATIVE) Would you say that you are extremely
conservative or just somewhat conservative? (IF MODERATE) If you had to choose,
would you say that you are a liberal or a conservative?
3. What about Howard Wilson? Do you think he is liberal, moderate or middle of the road,
or conservative? (IF LIBERAL) Would you say that he is extremely liberal or just
somewhat liberal? (IF CONSERVATIVE) Would you say that he is extremely
conservative or just somewhat conservative? (IF MODERATE) If you had to choose,
would you say that he is a liberal or a conservative?
4. Now I'd like to get your feelings toward some of our political leaders and other people
who are in the news these days. I'll read the name of a person and I'd like you to rate that
person using something we call the feeling thermometer. Ratings between 50 degrees and
100 degrees mean that you feel favorable and warm toward the person. Ratings between
0 degrees and 50 degrees mean that you don't feel favorable toward the person and that
you don't care too much for that person. You would rate the person at the 50 degree mark
if you don't feel particularly warm or cold toward the person. How about Howard Wilson,
the person I described earlier? How do you rate your feelings toward him?
5. Now I am going to read some statements about Howard Wilson, including some that may
not apply. For each statement, please tell me whether the word or phrase describes him
extremely well, quite well, not too well, or not well at all. Wilson is too greedy to be a
good politician. How well does this statement describe Wilson?
6. Now I am going to read a list of statements about Jews, including some that were made
many years ago. For each, please tell me whether it describes Jews extremely well, quite
well, not too well, or not well at all . . . Jews are liberal politically. (Would you say this
statement describes Jews extremely well, quite well, not too well, or not well at all?
28
Table 1
Study Design
Political Stereotype
Profile 1 Profile 4
Control Condition
No Label, Not Shady No Label, Not Shady
Profile 2 Profile 5
Label-Only Condition
Jewish Label, Not Shady Jewish Label, Not Shady
Profile 3 Profile 6
Shady and Label
Jewish Label, Jewish Label,
Condition
“Shady” “Shady”
Table 2
Mean Perception of Candidate’s Liberalness
Political Stereotype
Political Stereotype
Liberal Candidate Conservative Candidate
Social Stereotype Coefficient (SE) Coefficient (SE) Difference (SE)
Control Condition (constant) 4.06 (0.17)*** 3.30 (0.20) *** 0.76 (0.26)***
Label-Only Condition
(1 = Label condition, 0 = control or 0.24 (0.25) 0.58 (0.28)** 0.34 (0.37)
Shady conditions)
N 309 287
Political Stereotype
Liberal Candidate Conservative Candidate
Social Stereotype Coefficient (SE) Coefficient (SE) Difference (SE)
Shady and Label Condition -4.78 (1.42)*** -6.94 (1.41)*** 2.16 (2.00)
Political Stereotype
Shady and Label Condition 4.03 (1.53)*** -1.77 (1.68) 5.80 (2.27)**
Political Stereotype
Political Stereotype