investigate Americans’ stereotypes of both Muslims and Muslim-Americans
Abstract We investigate Americans’ stereotypes of both Muslims and Muslim-Americans. We find that negative stereotypes relating to violence and trustworthiness are commonplace and that little distinguishes Muslims from Muslim-Americans in the public’s mind. Furthermore, these stereotypes have consequences: those with less favorable views of Muslims are more likely to support several aspects of the War on Terror.
1
Since September 11, 2001, American politics and governance has largely focused on the “War on
Terror” and the wars in Afghanistan and Iraq. Despite many differences, these wars have a common feature:
an interaction between the United States and the Muslim world. This interaction raises two important but
overlooked questions: First, what do Americans think about Muslims living in the United States and
elsewhere? And second, do these attitudes toward Muslims shape attitudes toward the War on Terror and the
wars in Afghanistan and Iraq?1
Political issues often have a group-centric basis, whereby the group directly implicated by an issue is
central to the politics of that issue and to attitudes about that issue. The War on Terror may be such an issue,
and yet relatively little research has explored the group-centric basis of American attitudes toward the War on
Terror. In particular, few studies have examined Americans’ views of Muslims themselves and the role these
views play—even though the “enemy” in both wars has been repeatedly identified by its religious identity.
References to Islam range from sweeping generalizations—e.g., when the Reverend Franklin Graham called it
“a very evil and a very wicked religion”—to more nuanced differentiations of mainstream Muslims from
violent extremists—e.g., when President George W. Bush singled out “Islamo-Fascists.” Although it is clear,
both in reality and often in the rhetoric of American political leaders, that the War on Terror implicates a
small subset of Muslims, in the minds of some Americans this distinction may give way to a generalized
conception of the enemy that implicates Muslims more broadly. Thus, despite attempts to differentiate
extremists like al-Qaeda from Islam writ large, group-centrism may affect public opinion about the War on
Terror, with those having derogatory attitudes about Muslims more likely to support these wars.
Before we can clarify the effect of attitudes toward Muslims on support for the War on Terror, we
must first understand how Americans view Muslims themselves. Although survey data reveals unfavorable
attitudes toward Muslims, Muslim-Americans, and Islam generally, political scientists know little about the
specifics of these attitudes or their consequences. Our first contribution is to investigate the content of
attitudes toward Muslims and Muslim-Americans. We focus on stereotypes, or generalizations about the
traits, of both groups. We draw on theories of stereotype content that identify salient dimensions of 1 For economy of expression, we will often refer to these wars collectively as the “War on Terror,” mindful of the controversy over this label and over whether the Iraq War is truly part of the fight against terrorism.
2
stereotypes and then develop expectations about how Americans will evaluate Muslims and Muslim-
Americans along those dimensions. Evaluations of outgroups—and, as we will show, of Muslims and
Muslim-Americans—are not uniformly negative. Stereotypes of Muslims and Muslim-Americans as violent
and untrustworthy are particularly common. Thus, people value and de-value different groups, including
Muslims, for different reasons. We also compare and contrast stereotypes of Muslim and Muslim-Americans
with those of other American ethnic groups. This allows us to better map the contours of prejudice.
Our second contribution is to delineate how stereotypes of Muslims and Muslim-Americans
structure support for the War on Terror. This is valuable in three respects. First, we establish that attitudes
toward Muslims, not just an overall denigration of outgroups—i.e., ethnocentrism—are associated with
support for the War on Terror. This finding differs significantly from previous research. In the War on
Terror, it is less important that some Americans disparage outgroups than that they disparage Muslims in
particular. Second, we establish not just the role of attitudes toward Muslims, but the role of specific
stereotypes. Although the group-centric basis of policy preferences is commonly assessed with omnibus
measures of attitudes toward groups, specific stereotypes might be related in different ways to policy
preferences depending on how policy and policy debates intersect with particular beliefs about groups. We
show that Americans who consider Muslims to be violent and untrustworthy are more likely to support the
War on Terror. Finally, we extend the literature on group-centrism into a relatively new domain of public
policy. Although scholars have identified group-centrism in attitudes about many domestic policies, such as
welfare, and in attitudes about foreign policy during World War II and the Cold War, very few studies have
examined the War on Terror, even as this war has made Muslims even more salient to foreign policy debates.
Ultimately, our findings help to clarify the relationship between evaluations of groups and policy preferences
and produce a more theoretically and empirically refined portrait of group-centrism in the War on Terror.
Understanding stereotypes of Muslims and Muslim-Americans and their effects on public opinion
has obvious relevance for current politics. It also speaks to an important normative issue: the extent and
nature of prejudice in the United States. The literature on prejudice has focused mainly on racial groups, and
in particular African-Americans, although there is important work on religious groups, primarily Jews.
3
Attitudes toward African-Americans have changed dramatically over the past several decades, with traditional
forms of racism—those emphasizing innate deficiencies in intelligence—becoming rarer, if self-reported
attitudes are to be trusted. These shifts have sparked a debate about how much and what kind of prejudice
persists. We find significant prejudice against both Muslims and Muslim-Americans—sentiments that appear
difficult to dislodge absent dramatic shifts in how Muslims are portrayed in the popular and news media.
Theories of Stereotype Content
The tendency to stereotype is well-known. Stereotypes—what Lippman (1922) called “maps of the
world”—exemplify the universal human inclination to categorize. Individuals are placed into groups and
then the perceived characteristics of the group, the stereotypes, are imputed to those individuals. Frequently,
the valence of those stereotypes serves to elevate one’s own group above other groups (Tajfel and Turner
1979). But what are the characteristics attributed to the group? How should we understand the content of
stereotypes, and what expectations should we have about stereotypes of Muslims and Muslim-Americans?
However automatic the tendency to stereotype, the content of the stereotypes themselves varies widely across
groups. Both research from psychology on stereotype content and research within international relations on
image theory suggest two central dimensions of stereotypes, which in turn flow from patterns of intergroup
relations.
Perceptions of both persons and groups serve a fundamental purpose: to preserve our own safety
and well-being (Wojciszke 2005). That is, we make inferences about persons and groups in terms of
attributes that speak to how these persons and groups will affect us. This leads to two dimensions of
stereotypes. The first and primary dimension reflects a basic assessment of others: will they hurt me or help
me? Answering this question helps us determine whether a group poses a threat and thus whether we can
safely interact with it. This dimension of stereotype content thus centers on a group’s intentions. It has been
labeled “warmth” (Fiske et al. 2002), “morality” (Wojciszke 2005), and “social desirability” (Phalet and Poppe
1997). We adopt the label “warmth” here.
4
The second dimension of stereotypes gets not at actors’ intensions, but at their abilities. It captures
how capable or effective they are at accomplishing their goals. This too is relevant to assessing how another
person or group may affect us. It connotes their ability to act on their goals (Phalet and Poppe 1997) and
includes such attributes as intelligence, skill, and creativity (Fiske, Cuddy, and Glick 2007). It has been labeled
“competence” (Fiske et al. 2002; Wojciszke 2005) and “intellectual desirability” (Phalet and Poppe 1997). We
adopt the label “competence.”
A range of evidence suggests that trait perceptions fall along these dimensions of warmth and
competence. The centrality of warmth emerges in some of the earliest studies of person perception. Asch
(1946: 266) writes that qualities or warmth or coldness were “of special importance for our conception of a
person”: two individuals, differing only in whether they were “warm” or “cold” and otherwise described
using identical adjectives, were perceived quite differently. Both dimensions have been found to characterize
evaluations of persons (Rosenberg, Nelson, and Vivekananathan 1968), small groups (Bales 1950), and social
groups (Phalet and Poppe 1997; Fiske et al. 2002). Indeed, Wojciszke (2005) finds that 82% of the variation
in person perception can be accounted for by these two dimensions.
Within international relations, image theory generates expectations about how international actors
will be perceived (Alexander, Brewer, and Herrmann 1999). It also suggests dimensions similar to warmth
and competence, although it uses different terminology. One important component of image formation is
“goal compatibility.” Arguably, the compatibility of actors’ goals speaks to their intentions and thus to the
dimension of warmth: an actor whose goals differ from yours is perceived as more threatening. Hurwitz and
Peffley’s (1990) research on perceptions of the Soviet Union captures two traits related to warmth, trust and
threat. The second component of image formation is the relative status and power of actors. Status and
power speak directly to the abilities of the actor—that is, to how able they are to exert their will and get what
they want in the international arena. It is analogous to the dimension of competence.
Given these dimensions, how will specific groups be perceived in terms of warmth and competence?
The answer depends a great deal on prevailing patterns of intergroup relations and media coverage that often
reflects those patterns. Groups who do not compete with the salient ingroup will be seen as “warmer” and
5
less threatening (Fiske et al. 2002); such groups are “allies” in the typology of image theory (Alexander,
Brewer, and Herrmann 1999). Second, groups who have amassed power and status will be seen as more
competent, while lower status and less powerful groups will be viewed as less competent. An assortment of
evidence confirms these hypotheses. For example, subjects’ evaluations of the competence of different
groups are positively correlated with their evaluations of these groups’ statuses, and their evaluations of
warmth are negatively correlated with perceived competition with that group (Fiske et al. 2002). Ultimately,
research on stereotype content suggests that “warmth and competence are reliably universal dimensions of
social judgment across stimuli, cultures, and time” (Fiske, Cuddy, and Click 2007: 82) and that perceptions of
groups on these dimensions depend on prevailing social structural patterns.
The Content of Stereotypes about Muslims
What do these theories predict about how Americans stereotype Muslims? We expect that
Americans will stereotype Muslims negatively on the warmth dimension—that is, as threatening, violent, etc.
By and large, Americans have little reason to perceive that their goals and those of many Muslims are
compatible. Historical and contemporary depictions of Muslims in both the entertainment and news media
have emphasized their hostile intentions (Karim 2003; Poole 2002; Said 1997; Shaheen 2009). Content
analysis of news coverage of Muslims does not necessarily support the strongest claims of negative media bias
but does find patterns that may contribute to the perception that Muslims are threatening. Sheikh, Price, and
Oshagan’s (1995) study of newspaper coverage found that stories about Muslims often involved crises, war,
and conflict and tended to refer to Muslims in aggregated rather than differentiated ways, e.g., as a religious
group (“Muslims”) or by religion and nationality (“Iranian Muslims”). Although most stories were positive or
neutral in overall tone, slightly more than half of these stories also included terms such as “fundamentalist,”
“militant,” “terrorist, “radical,” or “extremist.” Similarly, Nacos and Torres-Reyna (2007) find that, in the
year prior to September 11, voices alleging American Muslim and Arab support for terrorism were twice as
common as those rejecting this allegation. In this period, 31 percent of textual depictions of American
Muslims and Arabs were negative and 44 percent were neutral or ambiguous. In the six months after
6
September 11, coverage became somewhat more positive (see also Schildkraut 2002: 523-25; Weston 2003)—
due in part to sympathetic stories about the challenges American Muslims faced—but this shift proved short-
lived. A year after September 11, the plurality of coverage (43%) was negative and most of the remainder
(37%) of coverage was neutral or ambiguous.2 There is, in short, considerable reason to believe that
Americans would tend to have negative stereotypes Muslims on the warmth dimension.
It is less clear, however, that Americans will stereotype Muslims as strongly on the dimension of
competence. On the one hand, we might expect Americans to stereotype Muslims are relatively
“incompetent”—that is, as deficient in effort, aptitude, or other indicators of ability. Some writing about the
Islamic world portrays Muslims as insufficiently rational or intellectual. Viorst (1994) writes of Muslims and
Arabs as having an “intellectual weakness” (67). Peretz (1996) writes of Arab countries, “Alas, these are
societies which cannot make a brick let alone a microchip.” At the same time, some depictions of Muslims or
related populations portray them as lazy. In Shaheen’s (2009) extensive survey of film portrayals of Arabs, he
notes the recurring figure of the “wealthy sheik,” who is “slothful,” “slovenly,” and “indolent” (25, 27).
Arabs and Muslims therefore appear to lack both warmth and competence. Said’s characterization (1997: 8)
implicitly suggests both dimensions. He argues that the prevailing portrait of Muslims “reduc(es) them all to
a special malevolent and unthinking essence.”
On the other hand, Muslims are also portrayed as somewhat competent. In terms of concepts like
power and status, Muslims are not necessarily portrayed as inferior. Episodes of Muslim violence and the
ubiquity of their portrayal as “villains” (Shaheen 2009) suggest their power. Moreover, the “wealthy sheik” is
at least wealthy, signaling a certain status that indigents would not have. In the typology of image theory,
Muslims may most resemble the category of “enemy”—one whose goals are incompatible with yours but is
2 Of course, simply because coverage was negative does not mean that it portrayed Muslims negatively in terms of the warmth dimension. However, we expect a significant portion of coverage focused on Muslims’ intentions. For example, Nacos and Torres-Reyna find that a prominent topic in both print and broadcast media before the attacks was political candidates’ relationships with American Muslims or Arabs and with groups allegedly tied to terrorist organizations.
7
also commensurate in power and status. Characteristics of “enemies”—“hostile, untrustworthy, monolithic,
and opportunistic” (Alexander, Brewer, and Herrmann 1999: 80)—evoke common depictions of Muslims.3
Ultimately, stereotypes of competence are less one-sided and appear less frequently in the media than
do stereotypes related to warmth. Thus, depictions of Muslims and Muslim-Americans will be more
consistently negative with regard to warmth than with regard to competence. Because negative information
about outgroups is more likely to be cognized and remembered (Ybarra, Stephan, and Schaberg 2000), we
expect that Americans’ evaluations of Muslims and Muslim-Americans will mirror these depictions:
Americans will tend to evaluate Muslims and Muslim-Americans more negatively in terms of warmth than in
terms of competence. This is consistent with what Fiske et al. (2002) report when they included “Muslims”
among a number of groups in their study of stereotypes dimensions (although their findings are based on
non-random samples).4
Will there be differences between evaluations of Muslims and evaluations of Muslim-Americans? We
expect that Muslim-Americans are likely to be evaluated more positively than Muslims, particularly in terms
of their warmth. Depictions of violence, particularly terrorism, involve Muslim perpetrators much more
often than Muslim-American perpetrators. Furthermore, in the wake of September 11, news accounts of
Muslim-Americans tended to be favorable, often discussing the backlash against Muslim-Americans during
this time (Nacos and Torres-Reyna 2003, 2007; Weston 2003); prime-time television also contained
sympathetic portrayals of Muslims (Alsultany 2008). Although there have been examples of Muslim-
Americans who have been linked to the Taliban or al-Qaeda since September 11—such as John Walker Lindh
and Jose Padilla—these are the exception rather than the rule.
3 By contrast, an actor with incompatible goals but who has lower status and power is a “dependent” that is characterized as “childish, incompetent, and divided”—a portrait at odds with the common depictions of Muslims. 4 Cuddy, Fiske, and Glick (2007) find that Arabs are also rated low on the warmth dimension but near the midpoint of the competence dimension, consistent with what one would expect if portrayals of competence were mixed.
8
Measuring Stereotypes of Muslims
There are little data on stereotypes of Muslims. Public polling speaks only in broad terms, suggesting
that many Americans do not hold Muslims, Arab-Americans, Muslim-Americans, or Islam in high regard (see
Nisbet, Ostman, and Shanahan 2007; Panagopolous 2006). For example, a series of Pew surveys from 2002-
2006 found that only about half of respondents had a favorable opinion of Muslims. Beliefs that Islam is
violent are also prominent: in a March 2006 ABC Poll, about one-third of respondents said that Islam
“encourages violence,” and 58 percent said that there are more “violent extremists” within Islam than in
other religions. Opinions of Muslims and Arabs living in the United States are similarly mixed. In a March
2003 Princeton Survey Research Associates poll, 51 percent of respondents said their opinion of Muslim-
Americans was favorable, while 24 percent said that it was unfavorable. But suspicions were also evident: in a
September 2002 CBS/New York Times poll, 33 percent said that Arab-Americans are “more sympathetic to
terrorists” than other Americans; in a September 2002 Gallup/CNN/USA Today poll, 44 percent said that
they had less trust in “Arabs living in this country” after September 11.
These findings are suggestive but insufficient in two respects. First, Americans’ assessments of
Muslims and Muslim-Americans are rarely juxtaposed with their assessments of other ethnic or religious
groups. It is thus difficult to tell whether American views of Muslims or Muslim-Americans are distinctive.
Some studies have examined feelings toward Muslim-Americans, Arab-Americans, “Arabs in the Middle
East,” Palestinians, and/or “Islamic fundamentalists” and found that attitudes toward these groups are more
unfavorable than attitudes toward other American ethnic and religious groups (Davis 2007; Traugott et al.
2002). However, these studies do not attempt to measure stereotype content. Second, although some survey
questions mention traits that speak to the content of stereotypes—e.g., questions about whether Islam is a
violent religion—there is no systematic comparison of different traits. Huddy et al. (2005) examine
stereotyping of Arabs as a part of their larger study of the effects of threat and anxiety on antiterrorism policy.
Their measures of stereotyping capture four characteristics—trustworthy, honest, violent and extremist—but
they do not report the distribution of responses on these items. Their approach differs from ours in that they
asked about Arabs rather than Muslims and in that they focus on the origins of these stereotypes rather than
9
their consequences. Thus, a more suitable set of indicators would involve multiple traits and also provide
comparisons among Muslims, Muslim-Americans, and other groups.
The former issue—that of comparison to other groups—can first be addressed by drawing on data
from the 2004 American National Election Study (ANES). For the first time, the ANES added “Muslims” to
the list of groups that respondents evaluate on the familiar “feeling thermometer”—a 0-100 scale where 0
indicates a very cool feeling, 100 a very warm feeling, and 50 a neutral feeling. The ANES thus allows us to
compare feelings toward Muslims to those toward whites, blacks, Catholics, Jews, and a host of others.
However, the ANES does not enable us to investigate the specific stereotypes that people have, or to
compare both Muslims and Muslim-Americans to each other and to other groups. To accomplish these tasks,
we collected data as part of the 2006 and 2007 Cooperative Congressional Election Studies (CCES). These
surveys were administered on-line to a cross-section of American adults and included both pre- and post-
election waves. (See section I of the auxiliary materials for more details.) We pool the two surveys in the
analysis that follows.
In the pre-election wave, respondents were asked to rank groups on four different trait dimensions:
peaceful-violent, trustworthy-untrustworthy, hardworking-lazy, and intelligent-unintelligent. In selecting
these dimensions to include in the CCES, three considerations were paramount. First, the duration of the
survey imposed limits on the number of attributes we could include. Second, where possible, we sought to
include attributes that had been asked about groups in other surveys, especially the ANES. This enables us to
compare the CCES sample to ANES samples—an important exercise given the differences in their respective
sampling designs and modes. This also enables us to link our findings to the large literature on stereotypes of
other American ethnic and religious groups, such as African-Americans. The latter three dimensions are part
of the ANES and the first stereotype dimension (peaceful-violent) has also figured in previous work on
stereotypes (Sniderman and Piazza 1993; Huddy et al. 2005).
Third, and most importantly, we sought to include dimensions that captured attributes of Muslims
and Muslim-Americans that are arguably prominent in news and other media. Given the frequency with
which Muslims are depicted solely or mostly in terms of violent acts, the inclusion of the peaceful-violent trait
10
dimension was obvious. The trustworthy-untrustworthy trait dimension is equally relevant, especially given
concerns, noted earlier, about Muslim-Americans’ loyalty in the wake of September 11, 2001. The intelligent-
unintelligent trait dimension speaks directly to portrayals of Muslims as un-intellectual, primitive, or
backwards. The hardworking-lazy dimension speaks to portrayals of Muslims as indolent. Of course, these
four trait dimensions may not capture all extant stereotypes of Muslims. However, they do capture trait
dimensions that figure prominently in media portrayals of and other writing about Muslims. Later we
demonstrate that these four trait dimensions do tap the two dimensions of stereotype content.
CCES respondents were asked to evaluate five different groups on each trait dimension. 5 The first
four groups, whites, blacks, Hispanic-Americans, and Asian-Americans, are the traditional focus of the ANES.
They also provide a baseline against which to compare attitudes toward Muslims and Muslim-Americans, as
most research on racial attitudes in the United States has focused on these groups and especially on blacks.
After evaluating these groups, respondents were randomly assigned to evaluate either Muslims or Muslim-
Americans. This experiment within the survey will demonstrate whether respondents evaluate Muslim-
Americans more favorably than Muslims.
The Contours of Stereotypes of Muslims and Muslim-Americans
We begin by comparing attitudes toward Muslims and other groups, drawing on the ANES feeling
thermometer items. Figure 1 presents the mean thermometer scores for 10 different groups. Each data point
is accompanied by a horizontal line that indicates the 95 percent confidence interval around the mean. Here
5 The battery of stereotype items began with the hardworking-lazy dimension and this preamble: “Now we have some questions about different groups in our society. We’re going to show you a seven-point scale on which the characteristics of people in a group can be rated. In the first statement a score of 1 means that you think almost all of the people in that group are ‘hard-working.’ A score of 7 means that you think almost all of the people in the group are ‘lazy.’ A score of 4 means that you think the group is not toward one end or the other, and of course you may choose any number in between that comes closest to where you think people in the group stand.” The remaining traits were asked in this order: intelligent, trustworthy, and peaceful. This follows the ANES order (hardworking, intelligent and trustworthy). For each trait, respondents were asked to evaluate whites first. The order of blacks, Hispanic-Americans, and Asian- Americans was then randomized. This sequence mirrors the ANES. Muslims or Muslim-Americans were always the last group in the sequence. Because the CCES was conducted on-line, we ensured that the group- trait combinations (20 in all) were displayed on-screen one at a time, which minimizes the likelihood that the rating of one group will affect the rating of other groups (Tourangeau et al. 2007: 94).
11
we focus on white non-Hispanic respondents, none of whom are Muslims. This is only to provide a clean
comparison between the ingroup (whites) and the other groups.
[insert Figure 1 about here]
The figure shows that these white respondents rated whites themselves most favorably, but rated
most other ethnic groups warmly as well. The mean ratings of blacks, Asian-Americans, and Hispanic-
Americans, while lower than those of whites (p<.05), are all in the high 60s. White respondents also tend to
feel warmly toward two other religious groups, Jews and Catholics, but less warmly toward Christian
Fundamentalists. However, they felt much more coolly toward Muslims, whose mean thermometer score is
only slightly above 50 and well below the rating given to Catholics and Jews. Only gays and lesbians and
illegal immigrants were evaluated less favorably. Black and Hispanic respondents also rated Muslims less
favorably than most of these groups (data not shown). Thus, while their feelings toward Muslims are slightly
favorable in absolute terms, Americans appear to like Muslims less than most other groups—a finding that
dovetails with those reported by Traugott et al. (2002) and Davis (2007).
Does this lack of favorability reflect negative stereotypes of Muslims and Muslim-Americans? Figure
2 presents the means of the CCES stereotype items for each of the five target groups. Again, we focus on
non-Muslim white respondents to ensure clearer comparisons among groups. In Figure 2 higher averages
indicate more unfavorable beliefs (i.e., more violent and less peaceful).6 Each plot in Figure 2 is scaled from
3 to 5, with a vertical line at the midpoint (4). These averages thus fall closer to the midpoint of the
underlying scale than to the extremes (1 or 7), suggesting that most respondents do not think any grou
strongly embodies either positive or negative stereo
p
types.7
6 The confidence intervals are wider for the Muslim and Muslim-American items because these items were asked only of half-samples. In Figure 2 we pool respondents from the 2006 and 2007 CCES; the pattern of results holds when each survey is examined separately. 7 The percentage of respondents who did not evaluate Muslims or Muslim-Americans is somewhat higher than the percentage of respondents who did not evaluate the other groups. In the pooled 2006-2007 CCES, the level of item non-response was, on average, 7 percent for whites, blacks, Hispanics, and Asians, and 18 percent for Muslims and Muslim-Americans. This latter figure is somewhat lower than the level of non- response in public poll questions about Muslims and Islam. For example, averaging over 5 Pew surveys conducted from 2001-2004, 25 percent of respondents chose “don’t know” when asked whether they were favorable or unfavorable toward Islam. This difference may arise because the CCES asked about specific traits rather than general favorability or because the CCES differs in both sampling design and mode of
12
[insert Figure 2 about here]
Nevertheless, there is considerable evidence of negative stereotypes and of differences among target
groups. Most importantly, our primary expectation is confirmed: respondents rated Muslims and Muslim-
Americans much less favorably in terms of warmth than in terms of competence. On average, respondents
saw both Muslims and Muslim-Americans as more violent than peaceful, and as more untrustworthy than
trustworthy. However, with regard to competence, the opposite is true: on average, Muslims and Muslim-
Americans were perceived as more hardworking than lazy, and as more intelligent than unintelligent.8
This finding comports with a variety of other evidence. It echoes some of the findings of the public
polls cited earlier, such as the belief among about one-third of Americans that Islam promotes violence. In
fact, if anything, these stereotype results suggest that perceptions of Muslim violence are even more prevalent
than public polls suggest. In the CCES sample, 45 percent of respondents placed Muslim-Americans on the
“violent” side of the spectrum (at 5, 6, or 7 on the 7-point scale). Fifty-one percent placed Muslims on this
side of the scale. These numbers are at or above the fraction of whites (44%) who placed blacks on this side
of the scale—and this commonplace portrayal of blacks is considered evidence that “negative racial
characterizations are openly and routinely expressed” (Sniderman and Piazza 1993: 37). Finally, that interview than phone surveys such as Pew. It is noteworthy, however, that non-response among CCES respondents with regard to Muslims and Muslim-Americans was lower on the peaceful-violent dimension (11% and 13% respectively). This suggests that more people felt they had enough information to evaluate Muslims and Muslim-Americans on this dimension—an intuitive finding given the historical prevalence of violence in media stories about Muslims. The presence of item non-response may indicate that some people are cloaking potentially prejudicial sentiments with a “don’t know” response (see Berinsky 2004). However, we found little evidence that the likelihood of evaluating Muslims and Muslim-Americans was systematically related to key correlates of prejudice. For example, we calculated the correlations between the respondent’s level of formal education and dichotomous variables capturing whether the respondent provided an answer for each group-trait combination. These correlations were never higher than .07 in absolute magnitude; only 3 of 24 were statistically significant at p<.05; and those 3 significant correlations were not all of the same sign. Thus, we do not think that item non-response is significantly affecting the aggregate distribution of opinion. 8 Although the number of Hispanic and black CCES respondents is small (N=131 and 127, respectively), we can extract a tentative sense of their attitudes toward Muslims and Muslim-Americans. In general, these mirror whites’ attitudes in several key respects. As was evident in Figure 2, whites generally rate both Muslims and Muslim-Americans less favorably than they rate whites. Black respondents also rate Muslims and Muslims-Americans as more violent and Muslims as less trustworthy than they do blacks (p<.05), but they do not rate them as more or less hardworking or intelligent. Hispanic respondents rate both Muslims and Muslim-Americans less favorably than Hispanics on every trait. However, their evaluations of Muslims and Muslim-Americans on the peaceful and trustworthy dimensions are less favorable in the absolute than their evaluations on the intelligent and hardworking dimensions, suggesting again that negative stereotypes of Muslims and Muslim-Americans involve traits related to “warmth.”
13
Americans tend to regard Muslims and Muslim-Americans as violent and untrustworthy appears to reflect the
content of media messages about Muslims, which often center on the violent actions of Muslim extremists.
Do respondents differentiate Muslims from Muslim-Americans? Our results suggest that the answer
is no.9 Contrary to our expectation, we find no evidence that respondents systematically differentiate
Muslims from Muslim-Americans. To the degree that Americans attribute negative traits to Muslims, they see
those traits as embodied in Muslim-Americans as well.10
How do stereotypes of Muslims and Muslim-Americans compare to those of these other ethnic
groups? As we have seen, global evaluations of Muslims are generally less favorable than those of many other
groups. However, Figure 2 shows that evaluations vary across stereotype dimension. White respondents do
consider both Muslims and Muslim-Americans significantly less trustworthy and more violent than any other
group. (Blacks are considered nearly as violent, on average.) With regard to competence, however, white
respondents do not always evaluate Muslims and Muslim-Americans more negatively than the other groups
we examine.11 Blacks are perceived more as lazy than as hardworking and indeed are the only group whose
mean evaluation is on the “lazy” side of the scale. This finding confirms a variety of previous work (Kinder
and Sanders 1996; Sniderman and Piazza 1993). Both blacks and Hispanics are perceived as slightly less
intelligent than Muslims and Muslim-Americans (although in case of Muslim-Americans the differences are
not significant).
A final question is whether these four trait dimensions tap the underlying stereotype dimensions of
competence and warmth. A plausible alternative is that these trait dimensions merely capture a single
underlying dimension such as generalized affect toward Muslims or Muslim-Americans, which would then
9 Other surveys that have employed a split sample design have found that Muslims are rated less favorably than Muslim-Americans. For example, in a March 2002 Pew survey, 55% said their overall opinion of Muslim-Americans was “favorable.” The comparable figure for opinions of Muslims was 47% (the difference is significant at p<.001). 10 It is possible that respondents who were asked about “Muslims” were thinking of Muslims in the US. The contrast would perhaps have been clearer if we had asked about “Muslims outside the United States.” If so, then we are likely underestimating the prevalence of negative stereotypes of “Muslims.” Regardless, evaluations of Muslim-Americans are not particularly favorable. 11 White respondents evaluate Muslims and Muslim-Americans more negatively than whites and Asian- Americans on both traits reflecting competence and more negatively than Hispanic-Americans on the “hardworking” dimension.
14
suggest that people do not differentiate either group in terms of warmth and competence. To answer this
question, we estimated separate confirmatory factor analytic models for Muslims and Muslim-Americans, in
one case specifying a single underlying dimension and in the other case specifying two dimensions, with the
trait dimensions associated with warmth and competence loading on separate dimensions. For evaluations of
both Muslims and Muslim-Americans, the two-dimensional model was a significantly better fit to the data
than the one-dimensional model.12 The correlations among the two dimensions was 0.75 for evaluations of
Muslims and 0.72 for evaluations of Muslims—suggesting, naturally, that these evaluations, while reflecting
distinct dimensions, are nevertheless strongly related.
In sum, we have uncovered three key findings. First, stereotypes of Muslims and Muslim-Americans
have specific content: they involve the sense that Muslims, whether in the United States or elsewhere, tend to
be violent and untrustworthy, but not lazy or unintelligent. A general coolness toward Muslims—evident in
feeling thermometers—does not imply negative evaluations on every stereotype dimension, either in the
absolute or relative to other groups. As a consequence, one cannot easily generalize about whether Muslims
are evaluated more or less favorably than other groups in the United States. The answer depends on the
dimension of evaluation. Second, Americans consider Muslim-Americans and Muslims similar on most
dimensions. Despite the sympathetic treatment that Muslim-Americans received in post-September 11 media
accounts, many Americans do not differentiate them from Muslims on these dimensions. Finally, and most
importantly, negative stereotypes of Muslims and Muslim-Americans on the warmth dimension are prevalent.
Large pluralities and even majorities of Americans describe both groups as “violent” and “untrustworthy.”
With regard to Muslims and Muslim-Americans, some stereotypes are commonplace.
12 For stereotypes of Muslims, the fit statistics of the one-dimensional model were: chi-squared 116.6 (p<0.001); RMSEA=0.30; TLI=0.66; CFI=0.89). The comparable statistics for the two-dimension model were: 0.928 (p=0.34), 0.0001, 1.0, and 1.0. (In these models, a “good” fit is often defined as an insignificant chi-squared statistic, an RMSEA value below 0.05 and ideally close to 0, and TLI and CFI values above 0.9 and ideally close to 1.) For stereotypes of Muslim-Americans, the results were similar (here comparing the one vs. two dimensional models): chi-squared of 150.9 (p<.001) vs. 1.1 (p=0.29); RMSEA=0.34 vs. 0.01; TLA=0.58 vs. 1.0; and CFI=0.86 vs. 0.99.
15
Group-centrism and the War on Terror
The next question is whether evaluations of Muslims—both overall and in terms of warmth and
competence—affect policy preferences regarding the War on Terror. At least since Converse’s (1964)
seminal work, scholars have recognized that attitudes toward groups can structure public opinion. Although
Converse found little evidence that Americans’ beliefs were structured by ideology, he found considerable
evidence that the public evaluates issues in terms of their association with “visible social groups.” Group-
centric reasoning allows citizens to make political decisions without much detailed information or more
sophisticated abstract reasoning. Of course, as Converse recognized, groups matter only when citizens grasp
the relationship between the group and the policy: “The individual must be endowed with some cognitions of
the group as an entity and with some interstitial ‘linking’ information indicating why a given party or policy is
relevant to the group” (Converse 1964: 237). This suggests that issues can be more or less group-centric and
that the influence of groups can be heightened or diminished depending on the information provided by
elites or the media. Particular issues, specific events, and political debates can make attitudes toward a
specific group central to policy preferences.
Converse’s insight has received much empirical support in studies of American domestic policy. For
example, whites’ opinions about racial policies and crime policies depend on their views of black Americans
(Hurwitz and Peffley 2002; Kinder and Sanders 1996; Kinder and Winter 2001). Tolerance for free
expression depends on evaluations of the group demanding speech rights (Sullivan, Piereson, and Marcus
1982). Opinion on gay rights is influenced by attitudes toward gay people (Brewer 2003). Issue framing
accentuates the importance of group evaluations (Hurwitz and Peffley 2005; Mendelberg 2001; Nelson and
Kinder 1996; Winter 2008). In sum, when group cues are salient and clear, beliefs and feelings about social
groups are important ingredients in policy choices (Conover 1988; Nelson and Kinder 1996).
Previous research on international and foreign policy attitudes also suggests the importance of social
group identification. This was true during World War II: in 1939, American attitudes toward Germans and
Italians were related to support for intervention in Europe (Berinsky 2009). This was also true during the
Cold War: the more Americans perceived the Soviet Union as threatening and untrustworthy, the more they
16
favored a more aggressively militaristic foreign policy and containment of the Soviet Union (Hurwitz and
Peffley 1990; see also Holsti 1962). Similarly, Berinsky (2009) also demonstrates that favorable attitudes
toward blacks were associated with support for sanctions on South Africa in the 1980s.
Less well-understood is the connection between group-centrism and policy preferences in the War
on Terror. One perspective is Kam and Kinder (2007), who suggest that ethnocentrism—“prejudice broadly
conceived” (322)—helps explain support for the War on Terror. Kam and Kinder argue that those inclined
to see the world in us-them terms should be especially likely to support policies targeting this “strange and
shadowy enemy” (321), and find that ethnocentrism is associated with a variety of attitudes related to the War
on Terror (see also Kinder and Kam 2009).
We suspect, however, that attitudes toward the War on Terror depend less on a generalized
ethnocentrism than on evaluations of a more specific group: Muslims. The conditions laid out by Converse
for group-centrism to affect policy attitudes are certainly met in this case. For one, as we have already shown,
citizens do have “some cognitions of the group as an entity.” That is, they have generalized and often
negative views of both Muslims and Muslim-Americans. Second, the necessary “interstitial ‘linking’
information” is also present. The “enemy” in the War on Terror is not entirely shadowy and is instead
routinely depicted as extremist Muslims, whether embodied in Osama bin Laden himself, various al-Qaeda
leaders, or nameless others depicted in news stories of violence in the Middle East and elsewhere—even as
political leaders sought to separate this enemy from the whole of Islam (see Schildkraut 2002: 520-522).
Thus, our first expectation is that those with negative views of Muslims should be more likely to
support efforts to fight terrorism, the Iraq War, and perhaps George W. Bush himself, as his presidency has
been so defined by these wars. The only research that connects attitudes toward Muslims and the War on
Terror (Nisbet, Ostman, and Shanahan 2007, Schildkraut 2002), finds that unfavorable views of Islam were
associated with increased support for various measures that would subject Muslims within the United States
to additional legal restrictions or police scrutiny.
Our analysis goes further, however, by linking particular dimensions of stereotypes to a broader
range of policy preferences. Previous research on group-centrism and policy preferences has sometimes
17
focused on omnibus measures of affect—e.g., feeling thermometer scores, racism scales, and so on.
However, there is good reason to think that specific beliefs about groups, rather than summary attitudes, are
consequential. For example, perceptions that blacks are lazy are much more strongly associated with attitudes
toward government assistance for blacks, welfare, and affirmative action than are perceptions that blacks are
violent (Gilens 1999; Peffley, Hurwitz, and Sniderman 1997; Sniderman and Piazza 1993). Thus, it is
important not only to go beyond a generalized ethnocentrism to investigate attitudes toward Muslims, it is
important to develop expectations that speak to the relative impact of warmth and competence stereotypes of
both Muslims and Muslim-Americans.
We suggest two possible hypotheses about how warmth and competence stereotypes of Muslims will
affect attitudes toward the War on Terror. One is that warmth will be a more important predictor than
competence: the more someone stereotypes Muslims as violent and untrustworthy, the more they will
support the War on Terror, but assessments of Muslims as lazy or unintelligent will not have similar effects.
Previous research has found that warmth judgments are more important in person perception than anything
else (Asch 1946). Other studies have also found that warmth judgments are more accessible (Wojciszke,
Bazinska, and Jaworski 1998; Fiske, Cuddy, and Glick 2007). One reason is that warmth judgments more
directly reflect assessments about the threat posed by a group, illuminating the potential costs of interacting
with that group; “people who are not friendly are more dangerous to others than are people who are not
competent” (Cuddy, Fiske, and Glick 2007: 634). This accessibility should give warmth judgments greater
potency. Warmth judgments are also particularly like to drive “active” behaviors, which are “those that are
conducted with directed effort to overtly affect the target group” (Cuddy, Fiske, and Glick 2007: 633).
“Passive” behaviors—“those conducted with less directed effort” and “a less deliberate or obvious intention
on the part of the actor to bring about a specific outcome” (633)—are facilitated more by competence
stereotypes. Our focus is on policies that directly affect Muslim and Muslim-American populations, rather
than policies analogous to passive behaviors such as avoidance or neglect.
Finally, the influence of stereotypes matters much more when there is a link between the particular
content of the policy and the stereotype. For example, stereotypes of blacks as lazy drive attitudes toward
18
welfare, but stereotypes of blacks as hostile drive attitudes toward criminal justice policies (Peffley, Hurwitz,
and Sniderman 1997). Thus, those who see Muslims as violent and untrustworthy should be most likely to
support the War on Terror because these policies directly implicate concerns about violent intentions and
reflect a desire to protect Americans from actual violence. For all of these reasons, warmth stereotypes may
be more important predictors of attitudes toward the War on Terror than competence stereotypes.
Another second hypothesis is that the relevance of competence judgments depends on warmth
judgments. That is, there should be an interactive effect, such that people who perceive Muslims negatively
on the warmth dimension but positively on the competence dimension should be the most supportive of the
War on Terror. The logic is straightforward: a group is more threatening if it is both violent and competent;
here, competence implies “a higher efficiency in wrongdoing” (Wojciszke 2005: 170). Cuddy, Fiske, and
Glick (2007: 79) make a similar point: “competent behavior is particularly diagnostic when the other person is
perceived as immoral-unsociable; the competence of an enemy potentially has greater consequences than the
competence of a friend.” Thus, competence stereotypes may affect support for the War on Terror only
among those who tend to see Muslims as violent and untrustworthy.
We also expect stereotypes of Muslims to be a more potent explanatory factor than stereotypes of
Muslim-Americans. Although the overall level of derogation is similar for both groups (see Figure 2), there is
good reason to believe that the consequences of derogation may differ. In particular, the “linking
information” needed for people to tie the group to the policy in question is simply less prevalent for Muslim-
Americans. Rhetoric in the War on Terror arguably centers more on Muslims than Muslim-Americans, and
thus we expect Muslims to constitute a more salient threat in the minds of Americans with respect to the
policies in question than do Muslim-Americans. The possible exception concerns efforts directly targeted at
potential terrorists living in the United States, such as monitoring phone calls and email. Here, stereotyping
of Muslim-Americans may be more important, as they could be the targets of such policies.
To summarize, our analysis of the consequences of attitudes toward Muslims and Muslim-Americans
is guided by four expectations. First, overall attitudes toward Muslims will be associated with support for the
War on Terror, net of any effects of ethnocentrism. Second, when attitudes are distinguished in terms of
19
warmth and competence stereotypes, warmth stereotypes will be more strongly and consistently associated
with support for the War on Terror than competence stereotypes. Those who see Muslims as violent and
untrustworthy should be most likely to support policies that are part of the War on Terror because these
policies directly implicate concerns about violent intentions and reflect a desire to protect Americans from
actual violence. Third, competence stereotypes may matter in combination with warmth stereotypes, in
which case those who stereotype Muslims as both competent and violent or untrustworthy will more strongly
support the War on Terror. Finally, stereotypes of Muslims should have a stronger relationship to policy
preferences than stereotypes of Muslim-Americans.
The Consequences of Stereotyping
To test these expectations, we draw on both the 2004 ANES and 2006-2007 CCES. This helps
ensure that the apparent influence of Muslim stereotypes is not an artifact of a particular sample, survey
instrument, or election campaign. We begin by describing the variables employed in our analysis of the
CCES data.
To capture attitudes toward the War on Terror, we draw on a number of items that figure in previous
research (e.g., Kam and Kinder 2007). Assessments of Iraq policy are based on two questions: whether
respondents considered the war in Iraq a mistake and whether they supported a troop withdrawal. A majority
of respondents in the combined 2006-2007 CCES samples said that the war was a mistake (59%) and favored
withdrawal (60% in 2006, the only year this question was asked). Six additional items tap views on the
broader War on Terror. Three involve government spending: whether federal spending for defense, foreign
aid, and the War on Terror should be increased, decreased, or kept the same. In the CCES, the majority
favored a constant or increased level of spending on defense (35 and 40%, respectively) and the War on
Terror (30 and 34%). The majority of respondents wanted to decrease spending on foreign aid (69%).13 A
fourth item asked whether respondents felt security or the preservation of civil liberties was more
13 The foreign aid spending question was asked only in 2006.
20
important.14 In the combined sample, 62% favored preserving civil liberties. Two other items, present only
in the 2007 survey, concerned specific measures intended to prevent domestic terrorist attacks: allowing
government agencies to monitor the phone and email of ordinary Americans and allowing courts to authorize
secret searches of homes.15 The majority of respondents opposed both monitoring phone and email (77%)
and secret searches of homes (85%) consistent with the finding that respondents favor the preservation of
civil liberties over security. Finally, because of the close association between the War on Terror and the Bush
administration, we include overall presidential approval.16 The majority of the sample disapproved of the
President in both 2006 (56%) and 2007 (62%)—figures close to the findings of other surveys from these
periods.
The key independent variables in our analyses are stereotypes of Muslims and Muslim-Americans on
the dimensions of warmth and competence. We measure stereotyping as the deviation between assessments
of Muslims or Muslim Americans and assessments of the respondent’s ingroup, averaging across the traits for
warmth (violent and untrustworthy) and the traits for competence (intelligent and hardworking). These
measures capture the comparative assessments that are central social identity theory (Tajfel and Tuner 1979)
and parallel the conceptualization and operationalization of ethnocentrism (Kam and Kinder 2007). In
constructing these measures, we focus only on non-Muslim white, black and Hispanic respondents.17 In
principle, the measures can range between 1 and -1. Positive numbers reflect negative stereotypes of Muslims
or Muslim-Americans vis-à-vis the ingroup—e.g., in the case of warmth, respondents scoring 1 see their own
14 This item is drawn from Davis and Silver (2004). The precise wording was “Which of these two statements you agree with most? ‘In order to curb terrorism in this country, it will be necessary to give up some civil liberties’; or ‘We should preserve our freedoms above all, even if there remains some risk of terrorism.’” 15 The exact wording was “Please tell me if you would favor or oppose each of the following as a way to prevent terrorist attacks in the United States: Allow government agencies to monitor the telephone calls and email of ordinary Americans on a regular basis” and “Allow courts to authorize federal law enforcement agents to conduct secret searches of Americans’ homes without informing the occupants for an unspecified period of time.” 16 The response options change slightly between the 2006 and 2007 CCES. In 2006 approval was measured with a four-category scale ranging from strongly approve to strongly disapprove; in 2007 a middle neutral category was added. We combined these measures by coding those without an opinion as a middle category. Estimating models of approval separately by year produces similar findings. 17 In other words, stereotypes of Muslims on the warmth dimension are computed for a white respondent by taking the difference between her evaluations of whites and Muslims on the trust and violence questions and then computing the average of these deviations.
21
group as very peaceful and trustworthy and see Muslims or Muslim-Americans as very violent and
untrustworthy. Those at the midpoint on the scale regard their own group as indistinguishable from Muslims.
In line with the results in Figure 2, stereotypes of warmth are more unfavorable than stereotypes of
competence for both Muslims and Muslim-Americans.18 By comparing the effects of warmth and
competence stereotypes, we test our expectation that warmth stereotypes matter more in these policy
domains.
For the CCES data, we employed two different specifications for each dependent variable, one using
warmth and competence stereotyping of Muslims and the other using warmth and competence stereotyping
of Muslim-Americans. (Recall that respondents were randomized to be asked about one or the other group.)
Comparing these paired models allows us to examine whether the effects of views on Muslims and Muslim-
Americans differ. The models also included ideological, attitudinal, and demographic factors known to affect
opinion in these domains. The first is ethnocentrism. Including ethnocentrism helps ensure that the estimated
effects of stereotypes are not picking up a more general hostility to outgroups. Following Kam and Kinder
(2007), we use the stereotype measures related to whites, blacks, Asians, and Hispanics to compute
ethnocentrism scores; these are the deviation between assessments of a given outgroup and assessments of
the ingroup averaged across the three outgroups and four traits.19 The measure is scaled between -1 and 1,
with positive numbers indicating more negative evaluations of the three outgroups relative to the ingroup.
Other predispositions in the model include authoritarianism and ideology. We measured
authoritarianism with four questions that ask about the most important values for parents to emphasize in
raising their children (see also Feldman and Stenner 1997). Respondents were presented with pairs of values
relating to the authority of the parents as compared to the autonomy of the child: independence or respect
18 In the pooled CCES sample, we find very modest levels of negative stereotyping in terms of competence (for both Muslims and Muslim-Americans, a mean of .08). By contrast we find much more negative stereotyping on the warmth dimension (for Muslims, a mean of .21; for Muslim-Americans, a mean of .20). 19 For example, for a white respondent, we first compute the difference between her assessments of whites as a group and her assessments of each of the three outgroups (Hispanics, Asians and blacks) averaged across the four traits. Then we average these three differences to compute the ethnocentrism scale. In the pooled CCES, the overall mean of this variable—0.06 on the -1/1 scale—indicates a modest degree of ethnocentrism, which is what Kam and Kinder also found. The mean is slightly higher for whites (0.07) than Hispanics (0.05) or blacks (0.02).
22
for elders, curiosity or good manners, obedience or self-reliance, and being considerate or well-behaved. An
additive index combining these items allows us to capture authoritarian tendencies.20 We measured ideology
by combining responses to four different issue questions: support for a ban on partial-birth abortion, for
funding stem cell research, for increasing the minimum wage, and for extending capital gains tax cuts. Higher
values on this measure indicate conservative positions on these issues.21 We expect both of these measures
to be positively correlated with support for the war in Iraq, the War on Terror, and President Bush (Davis
and Silver 2004; Huddy et al. 2005; Kam and Kinder 2007).
Perceived threat is measured with a question that asks respondents how likely they think it is that the
United States will suffer another terrorist attack in the next 12 months. Forty-seven percent of the combined
CCES respondents said that it was somewhat or very likely—a figure somewhat lower than polls in earlier
time periods. Support for policies to combat terrorism is stronger among those who consider further
terrorist attacks likely (Herrmann, Tetlock, and Visser 1999; Huddy et al. 2005). We also include party
identification, with higher values indicating identification with the Republican Party. Two measures capture
religious practice and belief. One is religious service attendance, with those who do not identify with a religion
coded as not attending a place of worship. The other is a dichotomous variable capturing whether the
respondent identifies as born again, which is the only available proxy in the CCES for more conservative or
evangelical strains of Christianity. The other independent variables capture whether the respondent is female;
and whether the respondent identifies as non-white. The models for presidential approval also include
evaluations of the national economy, measured by whether respondents believed the nation’s economy had gotten
better or worse, or stayed the same, in the previous year. Finally, as our models are based on data from the
pooled 2006 and 2007 CCES when we have the same dependent variable in both surveys, we include when
necessary a variable representing the year of the survey.
Our models in the 2004 ANES are similar. The 2004 ANES also contains a number of questions on
policies implicated in the War on Terror: spending on the War on Terror, border security, defense, and
20 In the CCES sample, the mean of this 0-1 scale, where 1 indicates the highest level of authoritarianism, is 0.44. Its reliability, measured with Cronbach’s alpha, is 0.64. 21 The mean of this scale is .39, and its reliability is 0.77
23
foreign aid; whether the wars in Iraq and Afghanistan were worth it; whether the war in Iraq has decreased
the threat of terrorism; overall approval of President Bush; approval of President Bush with regard to Iraq,
terrorism, and foreign affairs; and vote choice in 2004. Our models include the same independent variables
as in the CCES models, with these exceptions. Most importantly, the ANES did not include stereotype
measures for Muslims, so instead we rely on the Muslims feeling thermometer (see Figure 1). We use the
difference between a respondent’s feeling thermometer rating of Muslims and her rating of her own racial or
ethnic group to create a measure of derogation of Muslims. While this measure does not allow us to investigate
the dimensions of stereotyping, it does provide individuals’ assessments of Muslims relative to their own
ingroup. Our measurement of ethnocentrism is also slightly different, in that it is based on only three trait
judgments (hardworking-lazy, trustworthy-untrustworthy, and intelligent-unintelligent). We also rely on a
different proxy for conservative or evangelical Christianity: belief in the inerrancy of the Bible. Previous
research demonstrates the efficacy of this measure (Alwin et al. 2006), which we operationalize as two
dummy variables: one that captures those who believe the Bible is the literal word of God (“biblical literalist”)
and those who believe that the Bible is the word of God but not to be taken literally (“biblical believer, not
literalist”). The omitted category is those who believe the Bible is not the word of God.
In both sets of models, all independent variables with the exception of ethnocentrism and
stereotypes are coded between 0 and 1. All dependent variables are also coded between 0 and 1, such that 1
represents, for example, the belief that the Iraq war was not a mistake, opposition to withdrawing troops
from Iraq, support for President Bush, a preference for security over civil liberties, support for monitoring
phone and email, support for allowing courts to authorize secret searches of homes, a preference for
increased spending on the War on Terror and defense, and a preference for decreased spending on foreign
aid. All models are estimated with logit or ordered probit, as appropriate.
Results: 2004 ANES
We begin by presenting the results from the 2004 ANES, which allow us to test the first hypothesis
discussed earlier: that unfavorable attitudes toward Muslims should be associated with greater support for the
24
War on Terror. Appendix Table A-1 presents the coefficients and standard errors from our models. We
focus on graphical presentation of the key results: the marginal effect of attitudes toward Muslims on the
predicted probability of supporting the War on Terror, the Iraq War, and the president; with all other
variables are held at their means or modal values. Figure 3 presents the marginal effects of the Muslims
derogation measure on the change in the predicted probability of a given response for each dependent
variable.22 We sort these effects in descending order by size.
[insert Figure 3 about here]
Feelings toward Muslims are indeed associated with attitudes toward policies related to the War on
Terror. Those who view Muslims less favorably than their own ethnic group are more likely to support
increased spending on the war on terror, defense, and border security. They are more likely to support
decreased spending on foreign aid. They are more likely to approve of how President Bush is handling the
War on Terror and to believe the war in Iraq has decreased the threat of terrorism. They are more likely to
support the war in Afghanistan. Each of these effects is also substantively significant. For example, the
marginal effect of attitudes toward Muslims on the probability of support for anti-terrorism spending is 0.17,
which is roughly comparable to the marginal effects of party identification (0.20) and conservatism (0.16).
The effects of feelings toward Muslims are comparably sized in other models—e.g., an effect of 0.18 in the
model of Bush approval with regard to terrorism and an effect of 0.15 in the model of whether Iraq has
reduced the threat of terrorism.
Views of Muslims more strongly affect attitudes toward the War on Terror than attitudes toward the
war in Iraq or President Bush. Notably, however, we do find effects in the case of Iraq and presidential
approval when the survey questions explicitly mention terrorism. Spending on policy areas like border
security and defense, the war in Afghanistan, the relationship of the war in Iraq to the threat of terrorism, and
President Bush’s performance on the issue of terrorism are, in this sense, of a piece. Meanwhile, views on
Iraq and approval of Bush on dimensions other than terrorism are not significantly related to feelings about
22 The marginal effects were calculated using the “mfx” command in Stata 10. The magnitude of these effects is roughly equal to a shift from one-half standard deviation below the mean to one-half standard deviation above the mean.
25
Muslims. The effects of the other variables are also in line with previous research—e.g., more support for the
War on Terror among both Republican and conservatives. However, we find less evidence of an effect for
ethnocentrism, contrary to previous research (Kam and Kinder 2007). Ultimately, the chief finding here
confirms our first hypothesis: attitudes toward Muslims clearly structure attitudes toward the War on Terror.
Results: 2006-2007 CCES
Do stereotypes of Muslims and Muslim-Americans matter? Appendix Tables A-2, A-3, and A-4
present the coefficients and standard errors from our models. Again we focus graphical presentation of the
key marginal effects: those of warmth and competence stereotypes of Muslims and Muslim Americans.
Figure 4 presents the marginal effects of stereotypes of Muslims on change in the predicted probability of a
given response for each dependent variable; Figure 5 presents the effect of negative warmth and competence
stereotypes of Muslim-Americans. In both figures, the marginal effects are sorted by the size of the effect for
warmth stereotypes of Muslims (the dimension and group for which we think stereotypes will be most
strongly associated with attitudes toward the War on Terror).
[insert Figures 4-5 about here]
Our results suggest that stereotypes matter, but that particular stereotypes of a particular group are
most important: negative stereotypes of Muslims on the warmth dimension—that is stereotypes of Muslims
as violent and untrustworthy. Stereotypes of Muslims’ competence are less important, as are stereotypes of
Muslim-Americans generally. Each of these findings corresponds to our expectations. Furthermore, as in the
analysis of the ANES data, the effects of stereotypes are most evident on attitudes connected to the broader
War on Terror, rather than attitudes specific to either the Iraq War or President Bush. Indeed, in all sets of
models in the ANES and CCES, overall attitudes toward Muslims and warmth stereotypes of Muslims are
associated with many of the same policy domains, including spending on defense, foreign aid, and the War on
Terror.
In the CCES models, stereotypes of Muslims as violent and untrustworthy are associated at a
statistically significant level with increased spending for the War on Terror, increased spending on defense,
26
decreased spending on foreign aid, and a willingness to sacrifice civil liberties for security. The magnitudes of
these effects are substantively significant. For example, the marginal effect of negative warmth stereotypes of
Muslims on the probability of supporting increased spending for the War on Terror is 0.31; by comparison,
the marginal effects of party identification and ideology are 0.22 and 0.32, respectively. The marginal effect
of warmth stereotypes on the other three measures is roughly comparable, both in relative and absolute
magnitude—e.g., its effect on the probability of supporting increased spending on defense is 0.24 and its
effect on the probability of preferring security over civil liberties is 0.27. By contrast, negative warmth
stereotypes of Muslims are not significantly associated with support for specific policies designed to fight
domestic terrorism—monitoring phone and email and allowing secret searches—or with attitudes toward the
Iraq War or President Bush. Again, this mirrors the ANES findings, when feelings about Muslims were
associated with attitudes toward the Iraq War or President Bush only in the explicitly context of terrorism.
Do the effects of warmth stereotypes of Muslims exceed those of competence stereotypes of
Muslims? This hypothesis receives some support. Competence stereotypes of Muslims are rarely associated
with attitudes toward the War on Terror, the Iraq War, or President Bush. Only once do competence
stereotypes have a statistically significant effect: those who see Muslims as unintelligent and lazy are more
likely to support increased spending on defense. Moreover, the estimated marginal effects of negative
competence stereotypes are just as likely to be positive as negative. These null findings emerge even if we
estimate models that include competence stereotypes but not warmth stereotypes. The effects of warmth
stereotypes are substantively similar regardless of whether competence stereotypes are included.23 On the
other hand, formal tests of the equivalence of the effects of warmth and competence stereotypes suggest that
those effects are different at the 0.05 level in the model of spending on the War on Terror and different at
p=0.08 in the model of spending on foreign aid—but not statistically distinguishable in the models of defense
23 These results are available on request. Given the correlation between warmth and competence stereotypes of Muslims (r=0.69), we ran a battery of collinearity diagnostics. These did not indicate any problematic level of collinearity. The variance inflation factors were never higher than 3.00, well below the cutoff of 10 that some texts identify as problematic (e.g., Kennedy 1992: 183).
27
spending or civil liberties vs. security. 24 We think that the preponderance of evidence supports the primacy
of warmth stereotypes, but obviously this conclusion is tentative.
Perhaps, however, the effects of competence stereotypes are conditional on warmth stereotypes.
The hypothesis delineated earlier is that competence stereotypes may matter more among those who have
negative warmth stereotypes. We tested this hypothesis in a subsequent series of models that included an
interaction between competence and warmth stereotypes of Muslims. Those models are presented in Table
A-3. In no model is the interaction term statistically significant. Instead, the results continue to suggest that
warmth stereotypes are more strongly associated with attitudes than are competence stereotypes. Similar
results obtain in the models including stereotypes of Muslim-Americans (data not shown). Thus, competence
stereotypes appear to have little direct or indirect effect on policy preferences.
The final hypothesis concerned the effects of stereotypes of Muslim-Americans, which we expected
would be weaker than the effects of stereotypes of Muslims except perhaps in domestic policy domains
within the War on Terror (e.g., government monitoring of phone conversations or email within the United
States). The results, displayed in Figure 5, tend to confirm this hypothesis. Stereotypes of Muslim-Americans
on either dimension, warmth or competence, are associated with attitudes toward the War on Terror in only
one instance: support for increased spending on defense, which is significantly associated with both
stereotype dimensions.25 But contrary to expectations, warmth stereotypes of Muslim-Americans are not
associated with policies focused on homeland security, such as eavesdropping and secret searches. Thus,
stereotypes of Muslims appear more influential than those of Muslim-Americans, particularly on the warmth
dimension. However, formal tests of the equivalence of the effects of Muslim and Muslim-American warmth
stereotypes suggest statistically significant differences in only one case, spending on the War on Terror. Thus,
although substantial evidence suggests that stereotypes of Muslims are more influential than stereotypes of
Muslim-Americans, that inference is not definitive.
24 The test of equivalence in the model of spending on the War on Terror generated a chi-squared statistic of 5.7 (p=0.01, one-tailed). The comparable statistics in several other models were: spending on defense (χ2=0.1; p=0.79); spending on foreign aid (χ2=1.98; p=0.08); and civil liberties vs. security (χ2=1.12; p=0.14). 25 Tests for the equality of the coefficients for warmth and competence stereotypes of Muslims-Americans demonstrate no significant differences between the effects of the two dimensions.
28
With regard to the other predictors of opinion, our models of the CCES data generally confirm
previous research. Partisanship and conservatism have strong and consistent effects across most of the
dependent variables we examine. Authoritarians are more willing to spend money on defense, less willing to
send money to foreign countries, more willing to sacrifice civil liberties in the name of security, less likely to
see Iraq as a mistake, and more likely to approve of the job the President is doing. Perceptions of threat also
matter: those who believe another terrorist attack is likely are more likely to support aggressive action. All of
these findings confirm previous research (e.g., Huddy et al. 2005). However, once we account for Muslim
stereotypes, we find that ethnocentrism has few statistically significant effects on attitudes (see Tables A2 and
A4). The War on Terror clearly has a group-centric basis, but it is attitudes toward the particular group most
implicated that matter. We discuss ethnocentrism in more detail below.
Our findings suggest four conclusions regarding the consequences of stereotyping. First, we find
that both overall derogation of Muslims and specific stereotypes of Muslims are an important component of
attitudes toward the War on Terror. Second, our results tend to confirm our expectation that warmth
stereotypes, not competence stereotypes, are more strongly associated with attitudes toward the War on
Terror.26 This latter result suggests that the influence of stereotypes may depend on not only the link
between the policy and the group, but also the link between the policy and the particular stereotype of that
group. Third, the pattern of findings suggests that attitudes toward Muslims are more important than
attitudes toward Muslim-Americans, which confirms our expectation that Muslims would more salient in the
public’s mind.27
26 It is possible that our models underestimate the total effect of Muslim warmth stereotypes. Muslim warmth stereotypes are significantly associated with perceptions of the likelihood of future attack (data not shown). By contrast, competence stereotypes of Muslims and both competence and warmth of Muslim- Americans have no effect on perceptions of threat. This is consistent with literature showing the connection between stereotypes and perceptions of threat (e.g., Stephan et al. 2002). Thus, there may be indirect effects of warmth, through perceived threat, as well as the direct effects of warmth shown in Table A-2. 27 A relevant question is whether the effects of attitudes toward Muslims are conditional on other individual level factors. Although a thorough investigation is beyond the scope of this paper, we estimated models including interactions between stereotype dimensions and political sophistication, religious service attendance, and the perceived threat of a terrorist attack. We did not find any consistent evidence of interactions between stereotypes and these factors. Clearly, however, this is an arena for future study.
29
Finally, the effects of attitudes toward Muslims and stereotypes of Muslims do not extend from the
War on Terror to the War in Iraq or President Bush himself, except when Iraq and President Bush are
specifically connected to the idea of terrorism—i.e., whether the War in Iraq has reduced the threat of
terrorism, and Bush’s handling of terrorism. This appears to suggest that merely mentioning Iraq or Bush
does not provide the “linking information,” to use Converse’s term, that enables people to connect attitudes
toward Muslims to their policy preferences. That linkage needs to be made explicit. This may also suggest
that, despite President Bush’s repeated assertion that the war in Iraq is part of the broader War on Terror, the
public does not necessarily connect the two—at least at the time our surveys were conducted. Thus, whereas
notions such as “terror” and “defense” and “security,” which form part of the rhetoric of the War on Terror,
implicate Muslim populations, the war in Iraq may not do so, even though many of the factions in Iraq are
themselves Muslims. In fact, politics within Iraq—which entail tensions among Muslim sects and factions,
many of whom the United States even seeks or has sought as allies—may mean that Americans do not see
the enemy in Iraq as an undifferentiated Muslim population.
Why Stereotypes and Not Ethnocentrism?
Our analysis focuses on public opinion in the fall of 2006 and fall of 2007—more than five years
after the September 11 attacks and almost four years into the war in Iraq—and in the fall of 2004, when the
September 11 “rally effect” was over and attitudes toward President Bush and his policies, including the War
on Terror, had polarized along partisan lines. A first question is how our findings compare to those of
researchers who investigated earlier time periods. Huddy and colleagues (2005) examine support for anti-
terrorism policy in the year immediately following the September 11 attacks. Kam and Kinder (2007)
examine support for the War on Terror in the fall of 2002, during the lead-up to the Iraq War but after the
fall of the Taliban in Afghanistan. Although our respective analyses concern three very different periods,
there are important continuities. At each point in time, partisan and authoritarian predispositions, as well as
perceptions of threat, affected support for efforts to combat terrorism.
30
We also find some differences, particularly with respect to the nature of group-centric thinking. Kam
and Kinder (2007) demonstrate persuasively that the initiation of the War on Terror activated ethnocentrism,
which helped produce support for the War on Terror in the fall of 2002. We find little evidence that
ethnocentrism affects attitudes toward the War on Terror. 28 In 2004, 2006, and 2007, such attitudes
depended on a different species of group-centrism—one that denigrates a particular “enemy,” Muslims, as
opposed to one that denigrates outgroups generally. Kam and Kinder (2007: 336) speculate that views on
Muslims would likely be significant predictors of support for the War on Terror but that ethnocentrism
would still contribute to opinion as well. Our results suggest that attitudes toward Muslims matter more.
One possible explanation for this difference is that the effects of ethnocentrism on policy attitudes
derive from the effects of stereotypes of Muslims. That is, ethnocentrism leads people to hold negative
stereotypes of Muslims and these stereotypes then predict policy attitudes. Indeed, ethnocentrism is strongly
related to both stereotypes of Muslims and Muslim-Americans and overall affect toward Muslims, controlling
for authoritarianism, religiosity, education, and several other factors (see Table II in the auxiliary materials).
This is unsurprising, as ethnocentrism is associated with negative attitudes toward a variety of outgroups,
including gays and lesbians, illegal immigrants, and foreigners (Kinder and Kam 2009). However, its effects
on policy attitudes with respect to the War on Terror are mostly accounted for by stereotypes of Muslims. If
we exclude attitudes toward Muslims from of our models, we find more consistent statistically significant
positive effects of ethnocentrism. However, excluding ethnocentrism from the models tends not to affect the
estimates effects of Muslim stereotypes. 29 These results are provided in Tables III-IV of the auxiliary
28 We do find significant positive effects of ethnocentrism on measures of presidential performance in the 2004 NES, but not in the 2006 or 2007 CCES. In 2004, ethnocentrism was associated with presidential approval generally and with regard to foreign relations—all questions on which attitudes toward Muslims had no effect. In the CCES data ethnocentrism has a positive effect on support for decreasing spending on foreign aid in the model with Muslim American stereotypes, but not in model with Muslim stereotypes. 29 If we reestimate the ANES models in Tables A-1 without the Muslim derogation measure, ethnocentrism becomes statistically significant in the models of spending on border security and foreign aid and in the model of Bush’s handling of the Iraq War, but not in any other models beyond those in which its effect is already significant in Table A-1. If we reestimate the CCES models in Table A-2 without the Muslims stereotype measures, ethnocentrism becomes a substantively larger and statistically significant predictor of spending on the War on Terror and defense. By contrast, our estimates of the effect of Muslim stereotypes are less sensitive to the exclusion of ethnocentrism. If this measure is excluded the effects of Muslim stereotypes are substantively similar in the 2006-2007 CCES models. In the 2004 ANES models, its effects
31
materials. Our findings therefore suggest this role for ethnocentrism: it strongly affects people’s tendency to
hold negative stereotypes of Muslims and Muslim-Americans, but it is these specific stereotypes, not
ethnocentrism itself, that primarily drive attitudes towards the War on Terror.
Of course we do not know if the inclusion of a measure of Muslim stereotyping in models based on
earlier survey data would have reduce the impact of ethnocentrism. Possibly, just as ethnocentrism is
activated under certain conditions, it is also deactivated. It may be that early on—in the face of a new kind of
war, with an unclear enemy—Americans were more likely to think about the world in general “us versus
them” terms. As time went on, the enemy may have become better identified by its religious affiliation in
news stories and political rhetoric. Once the abstract enemy became more concrete, citizens may have relied
less on a general predisposition toward outgroups and more on assessments of this better-defined enemy.
Accounting for Endogeneity
An obvious rejoinder to the analysis thus far is this: perhaps stereotypes of Muslims, rather than
influencing attitudes toward the War on Terror, have been created or strengthened by it (see Berinsky 2009:
288, fn.13). If so, then our estimates of its effects are biased. However, we doubt that stereotypes of
Muslims are simply creations or rationalizations of support for policies in the War on Terror. Two sets of
evidence, one mostly qualitative and one quantitative, support this contention
In Berinsky’s (2009) account of how views of ethnic groups shape support for war, he notes that the
reverse could be true. In his case study, World War II could shape attitudes toward Japanese-Americans,
rather than these attitudes’ driving support for the war. However, he suggests, historical accounts
demonstrate that ethnic hostilities predated the war. For example, prejudice against nationals from one axis
power, Italy, had long existed. There were well-established stereotypes, such as “dark skin” and “proclivity
for crime” (146). Thus, Americans “could easily map their feelings toward ethnic groups onto the
international scene” (147).
are also similar when ethnocentrism is excluded, except that its effect on overall approval of Bush becomes larger, doubling in magnitude and becoming statistically significant.
32
The same is true with regard to Muslims. Violent events involving groups of Muslims preceded
September 11, 2001, by years if not decades or centuries. Many such events also received substantial
coverage in the American media, including the killing of Israeli athletes and coaches during the 1972
Olympics, the Iran hostage crisis, the attack on the Marine barracks in Beirut, the hijackings of the Achille
Lauro and various airplanes, the downing of Pan Am flight 103, ongoing conflicts between Israelis and
Palestinians, the World Trade Center bombing in 1993, the embassy bombings in Kenya and Tanzania, and
the attack on the USS Cole. Some scholars of the media have decried for decades the commonplace portrayal
of Muslims as violent villains (Said 1997; Shaheen 1984). Survey evidence also suggests longstanding negative
impressions: in a 1980 survey, 44% of respondents believed that Muslims had “contempt” for Christianity
and 58% believed that Muslims had contempt for Judaism. Large proportions of respondents were willing to
characterize “Arabs” as “barbaric, cruel” (44%) and “warlike, bloodthirsty” (50%)—suggesting that similar
characteristics would be imputed to Muslims (Slade 1981). Moreover, even if attitudes toward Muslims
became more negative in the immediate aftermath of September 11, the War on Terror took shape more
gradually over the ensuing months and years. It is unlikely that people withheld judgments about Muslims
until after they had formulated attitudes toward the War on Terror, and only then constructed a portrait of
Muslims consonant with those attitudes.
There is also little evidence that people’s impressions of Muslims or Muslim-Americans became less
favorable after September 11. The percentage of Americans with a “favorable” impression of Muslim-
Americans was nearly identical before September 11 (55% in a September 2000 Pew survey) and several years
thereafter (50% in a July 2005 Pew survey). A similar question about Muslims was asked in a March 1993
Zogby survey, conducted three weeks after the first World Trade Center bombing. In this survey, 23% said
their impression of Muslims was favorable, 36% said it was unfavorable, and 41% provided no opinion. In
an August 2007 Pew survey, 43% said their impression of Muslims was favorable, while 35% said their
impression was unfavorable and 22% had no opinion. While fewer Americans had no opinion of Muslims in
2007 as compared to 1993, if anything the average opinion was more favorable in 2007 than in 1993. These
results dispute the notion that the events of September 11 gave rise to a new and more derogatory opinion of
33
Muslims and Muslim-Americans, and that stereotypes of Muslims derive from support for anti-terrorism
policies rather than the other way around.
The second set of evidence comes from a more traditional quantitative attempt to circumvent
endogeneity: an instrumental variables regression model. These models depend on finding suitable
instruments for the endogenous variable, ones correlated with it but not with the error term of the original
model. Fortunately, such instruments exist in the 2004 ANES, although not in the 2006-2007 CCES. We
can take advantage of this fact: people who are prejudiced toward one group tend to be prejudiced toward
others (Sniderman and Piazza 1993). Specifically, attitudes toward Muslims are strongly associated with
attitudes toward other “cultural outgroups” (Kalkan, Layman, and Uslaner 2009), including gays and lesbians,
people on welfare, and feminists. However, there is little theoretical reason to expect attitudes toward these
groups to be associated with support for the War on Terror, which has little to do with gay rights, welfare, or
feminism. The ANES included feeling thermometers for each of these three groups, which can serve as
plausible instruments for affect towards Muslims. We calculate affect towards these ougroups as we do affect
towards Muslims, by subtracting the feeling thermometer score for the ethnic or racial ingroup from the score
for each outgroup.
For each dependent variable that was significantly associated with attitudes toward Muslims—six in
total (see Figure 3)—we estimated a instrumental variables (IV) model with these instruments as well as an
analogous ordinary least squares (OLS) regression model that includes the original measure of affect for
Muslims.30 The models were otherwise specified as in Table A-1. We carried out a series of diagnostic tests
and robustness checks, which are detailed in Section II and Table V of the auxiliary materials. All of these
suggest that the instrumental variables model performs relatively well: the instruments are suitably strong
predictors of the endogenous variable and the instruments appear uncorrelated with the residuals, which
serve as a proxy for the error term. Both conditions for a valid instrumental variables model appear met.
30 We also estimated two-stage discrete choice models “by hand”: using the instruments to created predicted attitudes toward Muslims, and then regressing the dependent variables on these values and the other covariates (using logit or ordered probit, as in Table A-1). Traditional standard errors for the coefficients of the predicted values will be incorrect, so we calculate standard errors by bootstrapping. The results are quite similar to those we report in the text. See Table V in the auxiliary materials for more information.
34
To summarize the results, Figure 6 presents both the OLS and IV coefficients and 95% confidence
intervals for the effect of attitudes toward Muslims on these six measures of support for the War on Terror.
The confidence intervals for the IV coefficients are wider—a natural consequences of the instrumental
variables model, which typically trades efficiency in order to reduce bias. However, little bias is even evident.
The estimated effects are comparable across both the IV and OLS models. In fact, in two models—for
spending on border security and foreign aid—the IV estimate is actually somewhat larger than the OLS
estimate. As discussed Section II of the auxiliary materials, a similar story emerges when we estimate discrete
choice models and when we employ a robust inference procedure. Taken together, this evidence is certainly
not definitive. However, it is far more consistent with the notion that attitudes toward Muslims drive
attitudes toward the War on Terror than with the reverse.
Conclusion
The attacks of September 11 shifted the priorities of the United States. The resulting War on Terror
does not involve a singular “enemy,” but it does entail an interaction, and at times a direct confrontation, with
Muslim populations. Yet our knowledge of how Americans view Muslims both at home and abroad, as well
as the consequences of these views, is minimal. Thus, one major contribution of this study is empirical: it
expands our knowledge by delineating the stereotypes that Americans have of both Muslims and Muslim-
Americans, comparing these stereotypes to those of other ethnic groups, and showing how stereotypes
influence support for the War on Terror, a domain of American foreign policy relatively unexplored in the
public opinion literature.
One central finding is that Americans tend to see both Muslims and Muslim-Americans as violent
and untrustworthy—a finding that dovetails with psychological theories of stereotype content and with
depictions of Muslims in the news and entertainment media. Muslims and Muslim-Americans are denigrated
more strongly on the warmth dimension than the other ethnic groups we examined. A second finding is that
Americans do not differentiate Muslim-Americans from Muslims on these dimensions. On average, Muslim-
Americans are considered just as violent and untrustworthy as Muslims. In short, despite rhetorical attempts
35
by political leaders to distinguish Muslims who commit violent acts from the vast majority of peaceful
Muslims, and despite favorable depictions of Muslim-Americans in the wake of September 11, many
Americans have generalized and derogatory views of Muslims generally and Muslims in the United States.
Even more importantly, attitudes toward Muslims and, in particular, stereotypes of Muslims
influence support for the War on Terror. Our findings clarify the group-centric nature of public opinion
about the War on Terror in important ways. First, we show that attitudes toward Muslims, not simply a
generalized ethnocentrism, are central. This suggests that, when thinking about the War on Terror,
Americans may not be envisioning an ill-defined enemy but one clearly identified by religion. Second, we
show that specific stereotypes affect attitudes, and in specific ways. Although the group-centric basis of
policy preferences is commonly understood in terms of overall affect or attitudes—and certainly our results
show that overall affect toward Muslims does matter—particular stereotypes appear especially consequential.
Perceptions of Muslims as violent and untrustworthy—warmth stereotypes—are a key ingredient in support
for several aspects of the War on Terror. They are not associated, however, with attitudes toward the War in
Iraq or President Bush, although our findings that rely on overall attitudes toward Muslims suggest that such
attitudes do matter when the War in Iraq and Bush are situated in the context of terrorism. In short, we find
that, however normatively problematic is citizens’ use of stereotypes, they nevertheless use them in nuanced
ways: stereotypes matter most when the group is clearly implicated and there is a close correspondence
between the stereotype dimension and the policy in question.
This correspondence points to this study’s theoretical contribution. Scholars have long known that
stereotyping of social groups is ubiquitous and that stereotypes serve as convenient heuristics in decision-
making. Equally well-known, thanks in part to Converse’s insights more than forty years ago, is that citizens
often rely on their views of social groups to form political opinion. Our study builds on these findings by
showing how theories of stereotype content generate more refined expectations about (1) the relevant
dimensions on which groups are likely to be stereotyped and (2) which stereotype dimensions are likely to
affect political opinions. In particular, distinguishing between the warmth and competence dimensions of
stereotypes helps to illuminate the nature of outgroup stereotyping as well as how stereotypes influence policy
36
preferences. What is crucial for understanding the relationship between stereotypes and preferences is how
policy and policy debates intersect with particular beliefs about particular groups.
At this point, relatively little research in political science speaks to this intersection (Gilens 1999,
Hurwitz and Peffley 1990; Peffley, Hurwitz and Sniderman 1997). Future research might begin by
investigating the differential effects of warmth and competence stereotypes. We find that warmth
competence were more important in the policy domains we examine, although our evidence is not definitive.
In particular, scholars could engage the prediction that warmth stereotypes are associated with support for
“active” behaviors and policies that more directly address the group, whereas competence behaviors are
associated with “passive” behaviors. Scholars might theorize more carefully about where policies relevant to
particular social groups fall on the active-passive continuum and then systematically test how active and
passive policies are associated with warmth and competence stereotypes.
Finally, it is important to engage the normative implications of our findings. These findings speak to
the vigorous debate over how much prejudice exists and whether its very nature has changed—a debate that
has centered on the nature of prejudice toward African-Americans (see, e.g., Kinder and Sanders 1996;
Sniderman and Piazza 1993). A fuller account of contemporary prejudice should incorporate multiple groups,
particularly those who are newly prominent in American politics. Muslims and Muslim-Americans are two
such groups. Moreover, an account of contemporary prejudice must distinguish among different
manifestations of prejudice by identifying the particular stereotypes pertaining to social groups. On average,
overall attitudes toward Muslims are neutral and even favorable, depending on the omnibus measure
employed, but yet attitudes are much more negative when Americans evaluate Muslims’ proclivity for
violence and their trustworthiness. These particular stereotypes demonstrate that prejudice is alive and well
with respect to Muslims and Muslim-Americans. Moreover, it is consequential.
A second implication concerns the lack of differentiation between Muslims and Muslim-Americans
in the stereotypes ascribed to these two groups. Survey data show that Muslim-Americans tend to share
traditional American values, endorse assimilation into American culture, and reject Islamic extremism (Pew
Center 2007). Nevertheless, negative stereotypes of Muslim-Americans are just as prevalent as those of
37
Muslims. One explanation is that Muslim-Americans are believed to be outside of the mainstream of
American society and to share the same traits that are imputed to Muslims elsewhere. The danger is that such
sentiments could be mobilized in support of restrictions on the civil liberties of Muslims in the United States.
Although we did not find that negative stereotypes of Muslims or Muslim-Americans increased support for
monitoring the phone calls and email of “ordinary Americans” or to conduct secret searches of “American’s
homes” in 2007, other data suggest that people may be more willing to curtail the civil liberties of Muslims-
Americans in particular. According to a 2003 Cambridge Survey Research Associates poll, about 40 percent
of Americans are willing to empower government to monitor “Muslims legally living in the United States”
more than other groups (cf. Moore 2002). The nature and consequences of attitudes about Muslim-
Americans are promising areas for future research.
A question arising from these findings is how stereotypes could be changed, or at least how attitudes
toward the War on Terror could be decoupled from stereotypes of Muslims and Muslim-Americans. We see
both possibilities as unlikely. For one, the tendency to stereotype originates in relatively stable values,
predispositions, and cognitive styles. These factors are unlikely to change dramatically over a person’s
lifespan. There will always be those who stereotype. Second, current events suggest no end to the elite
messages and media coverage that (even inadvertently) frame the War on Terror in ways that enhance the
predictive power of negative stereotypes and reinforce the content of these stereotypes. If the messages that
the public hears are different and if there is not another major terrorist attack by Muslims on U.S. soil, then
perhaps the content and importance of Muslim stereotypes could change over time. But given the longevity
of these stereotypes, as well as the ongoing violence in Israel, Iraq, Afghanistan, Pakistan, India, and
elsewhere, the prevailing depiction of the Muslim world will likely continue to emphasize violence and the
threat that Muslims pose to Americans. It will thus be difficult for many Americans to think of Muslims as
anything but enemies.
38
Figure 1. Mean Feeling Thermometer Scores of Muslims and Other Groups
Illegal immigrants
Gays and lesbians
Muslims
Christian Fundamentalists
Hispanic-Americans
Asian-Americans
Jews
Catholics
Blacks
Whites
30 40 50 60 70 80
The data are based on white non-Muslim respondents only. Data points are weighted means, with bars representing 95 percent confidence intervals. The underlying scales run from 0-100, where 100 indicates the most favorable response. The vertical line indicates the midpoint of the scale. Source: 2004 ANES.
39
Figure 2. Means of Stereotype Items for Muslims and Other Groups
Whites
Blacks
Hispanic-Americans
Asian-Americans
Muslim-Americans
Muslims
3 4 5
Peaceful – Violent
Whites
Blacks
Hispanic-Americans
Asian-Americans
Muslim-Americans
Muslims
3 4 5
Trustworthy – Untrustworthy
Whites
Blacks
Hispanic-Americans
Asian-Americans
Muslim-Americans
Muslims
3 4 5
Hardworking – Lazy
Whites
Blacks
Hispanic-Americans
Asian-Americans
Muslim-Americans
Muslims
3 4 5
Intelligent – Unintelligent
The data are based on white non-Muslim respondents only. Data points are weighted means, with bars representing 95 percent confidence intervals. The underlying scales run from 1-7, where 7 indicates the most unfavorable response. The vertical lines indicate the midpoint of the scale. Source: 2006 and 2007 CCES.
40
Figure 3. Attitudes toward the War on Terror and Feelings toward Muslims (2004 ANES)
Decrease spending on foreign aid
Strongly approve of Bush (War on Terror)
Increase spending on border security
Increase spending on War on Terror
Increase spending on defense
Iraq decreased threat of terrorism
War in Afghanistan worth it
Strongly approve of Bush
Voted for Bush
Strongly approve of Bush (Iraq)
War in Iraq worth it
Strongly approve of Bush (foreign affairs)
-1 -.5 0 .5 1
Change in Predicted Probability
These graphs depict marginal effects of feelings toward Muslims on various measures of attitudes toward the War on Terror. These effects are derived from the models in Table A-1, and include 90 percent confidence intervals. These effects are calculated with all other variables at their means. Source: 2004 ANES.
41
Figure 4. Attitudes toward the War on Terror and Stereotypes of Muslims
Increase spending on War on Terror
Decrease spending on foreign aid
Support security over civil liberties
Increase spending on defense
Favor monitoring phone and email
Somewhat approve of Bush
Oppose withdrawal from Iraq
Iraq War was not a mistake
Favor secret search of homes
-1 -.5 0 .5 1 Change in Predicted Probability
Effect of Negative Warmth Stereotypes
of Muslims
-1 -.5 0 .5 1 Change in Predicted Probability
Effect of Negative Competence Stereotypes
of Muslims
These graphs depict marginal effects of stereotypes of Muslims on various measures of attitudes toward the War on Terror. These effects are derived from the models in Table A-2, and include 90 percent confidence intervals. These effects are calculated with all other variables at their means. Source: 2006-2007 CCES.
42
Figure 5. Attitudes toward the War on Terror and Stereotypes of Muslim-Americans
Increase spending on War on Terror
Decrease spending on foreign aid
Support security over civil liberties
Increase spending on defense
Favor monitoring phone and email
Somewhat approve of Bush
Favor withdrawal from Iraq
Iraq War was not a mistake
Favor secret search of homes
-1 -.5 0 .5 1 Change in Predicted Probability
Effect of Negative Warmth Stereotypes of Muslim-Americans
-1 -.5 0 .5 1 Change in Predicted Probability
Effect of Negative Competence Stereotypes
of Muslim-Americanss
These graphs depict marginal effects of stereotypes of Muslim-Americans on various measures of attitudes toward the War on Terror. These effects are derived from the models in Table A-4, and include 90 percent confidence intervals. These effects are calculated with all other variables at their means. Source: 2006-2007 CCES.
43
Figure 6. OLS and Instrumental Variables Estimates of the Effects of Attitudes toward Muslims
Spending on War on Terror
Spending on defense
Spending on border security
Spending on foreign aid
Iraq decreased threat of terror
Bush approval (War on Terror)
-.2 0 .2 .4 Unstandardized Coefficient
OLS
IV
This graph depicts ordinary least squares (OLS) and instrumental variables (IV) coefficients, with 95 percent confidence intervals. See Table V of the auxiliary materials. Source: 2004 ANES.
44
Table A-1. Models of Attitudes toward the War on Terror (2004 ANES)
Increase spending on war
on terror
Increase spending
on border security
Increase spending
on defense
Decrease spending
on foreign
aid
War in Afghanis
tan worth it
War in Iraq
worth it
War in Iraq
decrease d threat
of terrorism
Approve of Bush
Approve of Bush’s handling of Iraq
Approve of Bush’s handling of war
on terror
Approve of Bush’s handling
of foreign
relations Voted
for Bush Derogation of Muslims 0.433* 0.550* 0.594* 0.523* 0.655* 0.085 0.523* 0.162 0.094 0.515* -0.051 0.162 [0.172] [0.196] [0.166] [0.173] [0.391] [0.391] [0.174] [0.192] [0.196] [0.194] [0.195] [0.616] Ethnocentrism 0.103 0.601 -0.571* 0.4 -0.036 -0.279 -0.031 1.053* 0.536 0.145 0.976* 1.351 [0.302] [0.367] [0.299] [0.307] [0.683] [0.689] [0.311] [0.353] [0.357] [0.350] [0.360] [1.065] Party Identification 0.466* -0.024 0.779* -0.250* 2.499* 4.085* 1.175* 2.261* 2.056* 1.724* 1.919* 5.521* [0.145] [0.164] [0.139] [0.145] [0.332] [0.353] [0.146] [0.177] [0.176] [0.171] [0.174] [0.558] Conservatism 0.360* 0.956* 0.812* 0.778* -0.557 1.041* 0.570* 0.596* 0.398 0.681* 0.882* 2.413* [0.214] [0.238] [0.205] [0.214] [0.471] [0.472] [0.214] [0.237] [0.243] [0.234] [0.241] [0.787] Authoritarianism 0.035 0.351* 0.1 0.032 -0.291 0.254 0.334* 0.332* 0.289* 0.117 0.147 0.298 [0.138] [0.155] [0.130] [0.139] [0.315] [0.307] [0.139] [0.154] [0.155] [0.155] [0.159] [0.499] Religious service attendance -0.172 -0.256* -0.292* -0.152 0.43 -0.009 0.061 -0.22 -0.175 -0.063 0.063 0.562 [0.129] [0.146] [0.123] [0.129] [0.293] [0.290] [0.130] [0.145] [0.145] [0.144] [0.146] [0.465] Biblical literalist 0.352* 0.360* 0.474* 0.119 0.115 0.527 0.186 0.444* 0.440* 0.288* 0.321* -0.094 [0.147] [0.163] [0.140] [0.148] [0.326] [0.338] [0.150] [0.164] [0.165] [0.163] [0.169] [0.529] Biblical believer, not literalist 0.147 0.310* 0.220* 0.253* 0.109 0.211 0.151 0.19 0.114 0.300* 0.124 -0.464 [0.122] [0.132] [0.114] [0.124] [0.267] [0.291] [0.128] [0.138] [0.141] [0.136] [0.143] [0.427] Female -0.276* 0.015 -0.187* -0.103 -1.122* -0.141 -0.045 0.067 -0.083 -0.314* -0.167* 0.102 [0.082] [0.092] [0.078] [0.083] [0.192] [0.186] [0.084] [0.092] [0.093] [0.091] [0.094] [0.296] Nonwhite -0.165 -0.498* -0.005 -0.371* -0.951* 0.058 -0.03 -0.027 -0.269* -0.03 -0.16 -0.692* [0.108] [0.117] [0.104] [0.108] [0.220] [0.251] [0.110] [0.121] [0.130] [0.121] [0.125] [0.387] National economic evaluation 2.040* 1.702* 1.480* 1.672* 3.004* [0.191] [0.191] [0.183] [0.190] [0.588] Observations 840 840 764 838 837 832 841 827 830 824 816 654
Models were estimated including only non-Muslim white, black, and Hispanic respondents. Cell entries are ordered probit or logit coefficients, with estimated standard errors in parentheses. Estimates of cutpoints are not displayed. The feeling thermometer difference is constructed such that higher values signify less favorable evaluations of Muslims, relative to the ingroup. Source: 2004 ANES. *p<0.05 (one-tailed).
45
Table A-2. Models of Attitudes toward the War on Terror, Including Stereotypes of Muslims
Increase Spending on War
on Terror
Increase Spending
on Defense
Decrease Spending
on Foreign
Aid
Support Security
over Civil Liberties
Favor Monitoring Phone &
Favor Secret
Searches of Homes
Iraq Was Not a
Mistake
Oppose Troop
Withdrawal Bush Job Approval
Warmth of Muslims 0.932* 0.639* 0.878* 1.311* 0.906 -0.685 -0.159 0.133 0.296 [0.280] [0.289] [0.529] [0.572] [0.711] [0.693] [0.746] [1.107] [0.291] Competence of Muslims -0.364 0.786* -0.599 0.149 -0.898 0.842 -0.712 1.591 0.278 [0.378] [0.389] [0.720] [0.782] [0.938] [0.905] [1.026] [1.530] [0.392] Ethnocentrism 0.128 -0.758 -0.114 -1.11 1.104 0.253 0.595 -2.215 -0.379 [0.556] [0.571] [1.074] [1.184] [1.285] [1.227] [1.542] [2.482] [0.568] Party Identification 0.663* 0.961* 0.808* 1.603* 2.018* 1.205* 4.607* 3.964* 2.059* [0.195] [0.198] [0.374] [0.411] [0.505] [0.491] [0.600] [0.881] [0.227] Conservatism 0.947* 0.737* 0.237 1.693* 0.526 0.553 3.135* 3.539* 1.185* [0.206] [0.208] [0.400] [0.418] [0.497] [0.492] [0.561] [0.870] [0.222] Likelihood of attack 0.796* 0.571* 0.238 1.166* 1.230* 1.158* 1.310* 0.859 0.449* [0.176] [0.180] [0.316] [0.376] [0.455] [0.446] [0.520] [0.742] [0.190] Authoritarianism 0.346* 0.836* 0.965* 0.987* 0.563 0.502 1.835* 0.245 0.719* [0.174] [0.177] [0.316] [0.390] [0.450] [0.438] [0.500] [0.768] [0.194] Religious service attendance 0.198 0.003 -0.38 0.216 0.441 0.675* -0.352 -0.169 0.032 [0.136] [0.138] [0.258] [0.288] [0.336] [0.326] [0.386] [0.600] [0.146] Born again -0.108 -0.011 0.164 -0.105 0.133 0.153 0.279 0.418 0.290* [0.124] [0.126] [0.227] [0.259] [0.306] [0.292] [0.353] [0.537] [0.130] Female -0.16 0.243* -0.053 0.891* 0.264 -0.132 0.396 -0.654 0.094 [0.106] [0.106] [0.187] [0.237] [0.273] [0.266] [0.313] [0.464] [0.116] Nonwhite 0.0001 -0.077 -0.614* 0.278 0.450 0.193 0.281 -0.459 0.066 [0.157] [0.154] [0.284] [0.350] [0.391] [0.391] [0.471] [0.719] [0.183] Evaluation of Economy 1.917* [0.263] Year 2007 -0.221* 0.051 0.345 -0.678* 0.058 [0.104] [0.105] [0.230] [0.312] [0.126] Observations 563 566 222 538 338 336 518 218 567
Models were estimated including only non-Muslim white, black, and Hispanic respondents. Cell entries are ordered probit or logit coefficients, with estimated standard errors in parentheses. Estimates of cut points are not displayed. Warmth and competence of Muslims coded such that higher numbers represent negative evaluations, relative to the ingroup. Source: 2006 and 2007 CCES. *p<.05 (one-tailed).
46
Table A-3. Models of Attitudes toward the War on Terror, Including Stereotypes of Muslim-Americans
Increase Spending on War
on Terror
Increase Spending
on Defense
Decrease Spending
on Foreign
Aid
Support Security
over Civil Liberties
Favor Monitoring Phone &
Favor Secret
Searches of Homes
Iraq Was Not a
Mistake
Oppose Troop
Withdrawal Bush Job Approval
Warmth × competence -0.533 -0.728 0.403 -0.524 0.61 0.486 -0.465 2.995 0.103 [0.664] [0.691] [1.328] [1.537] [1.095] [1.020] [1.618] [3.312] [0.638] Warmth of Muslims 0.991* 0.719* 0.841 1.373* 0.85 -0.733 -0.094 -0.15 0.283 [0.291] [0.301] [0.542] [0.601] [0.713] [0.697] [0.784] [1.155] [0.302] Competence of Muslims -0.116 1.115* -0.799 0.413 -1.139 0.652 -0.476 0.05 0.225 [0.495] [0.508] [0.976] [1.103] [1.016] [0.975] [1.329] [2.241] [0.513] Ethnocentrism 0.121 -0.794 -0.144 -1.118 1.073 0.259 0.584 -2.335 -0.369 [0.564] [0.582] [1.077] [1.190] [1.258] [1.203] [1.560] [2.520] [0.569] Party Identification 0.651* 0.946* 0.825* 1.588* 2.024* 1.208* 4.590* 4.099* 2.061* [0.196] [0.198] [0.378] [0.413] [0.505] [0.491] [0.602] [0.901] [0.228] Conservatism 0.946* 0.738* 0.226 1.693* 0.536 0.563 3.134* 3.508* 1.186* [0.206] [0.208] [0.401] [0.418] [0.498] [0.493] [0.561] [0.874] [0.222] Likelihood of attack 0.797* 0.573* 0.243 1.165* 1.240* 1.166* 1.309* 0.895 0.450* [0.176] [0.180] [0.316] [0.376] [0.456] [0.447] [0.520] [0.744] [0.190] Authoritarianism 0.332* 0.815* 0.975* 0.974* 0.58 0.513 1.814* 0.308 0.723* [0.175] [0.178] [0.318] [0.392] [0.452] [0.439] [0.504] [0.781] [0.195] Religious service attendance 0.193 -0.003 -0.376 0.211 0.458 0.689* -0.357 -0.158 0.033 [0.136] [0.138] [0.258] [0.288] [0.338] [0.328] [0.386] [0.607] [0.147] Born again -0.104 -0.005 0.155 -0.1 0.136 0.156 0.281 0.357 0.289* [0.124] [0.126] [0.229] [0.259] [0.307] [0.292] [0.353] [0.543] [0.130] Female -0.16 0.243* -0.051 0.892* 0.266 -0.13 0.394 -0.6 0.095 [0.106] [0.106] [0.187] [0.237] [0.274] [0.266] [0.313] [0.467] [0.116] Nonwhite -0.01 -0.087 -0.605* 0.268 0.471 0.21 0.273 -0.406 0.067 [0.157] [0.154] [0.286] [0.351] [0.394] [0.393] [0.472] [0.723] [0.183] Evaluation of Economy 1.918* [0.263] Year 2007 -0.219* 0.052 0.348 -0.675* 0.058 [0.105] [0.105] [0.230] [0.312] [0.126] Observations 563 566 222 538 338 336 518 218 567
Models were estimated including only non-Muslim white, black, and Hispanic respondents. Cell entries are ordered probit or logit coefficients, with estimated standard errors in parentheses. Estimates of cut points are not displayed. Warmth and competence of Muslims coded such that higher numbers represent negative evaluations, relative to the ingroup. Source: 2006 and 2007 CCES. *p<.05 (one-tailed).
47
Table A-4. Models of Attitudes toward the War on Terror, Including Stereotypes of Muslim-Americans
Increase Spending on War
on Terror
Increase Spending
on Defense
Decrease Spending
on Foreign
Aid
Support Security
over Civil Liberties
Favor Monitoring Phone &
Favor Secret
Searches of Homes
Iraq Was Not a
Mistake
Oppose Troop
Withdrawal Bush Job Approval
Warmth of Muslim-Americans 0.093 0.553* 0.107 0.935 0.074 0.326 0.608 0.355 0.033 [0.282] [0.285] [0.577] [0.577] [0.706] [0.688] [0.777] [1.545] [0.297] Competence of Muslim-Americans 0.452 1.036* -1.29 0.766 1.576* 0.142 0.788 -2.033 0.202 [0.344] [0.358] [0.811] [0.723] [0.862] [0.781] [1.046] [2.178] [0.359] Ethnocentrism 0.294 -1.084* 2.030* -0.158 0.253 -0.336 -1.284 1.264 -0.404 [0.539] [0.545] [1.174] [1.095] [1.235] [1.219] [1.455] [2.970] [0.559] Party Identification 0.654* 0.631* 0.836* 1.762* 1.057* 1.648* 3.241* 4.062* 1.485* [0.191] [0.190] [0.338] [0.406] [0.488] [0.505] [0.522] [1.027] [0.216] Conservatism 1.071* 1.064* 1.197* 1.527* 0.907* 0.218 4.009* 4.699* 1.362* [0.198] [0.202] [0.397] [0.411] [0.511] [0.489] [0.560] [0.985] [0.223] Likelihood of attack 1.136* 0.918* -0.144 0.982* 1.171* 1.084* 1.405* 3.021* 0.725* [0.178] [0.179] [0.292] [0.378] [0.476] [0.480] [0.484] [0.967] [0.193] Authoritarianism 0.157 0.551* 0.932* 1.409* 1.538* 0.556 1.831* -0.976 0.699* [0.177] [0.176] [0.315] [0.383] [0.469] [0.455] [0.526] [1.010] [0.195] Religious attendance -0.005 -0.288* -0.622* 0.397 0.713* 0.407 -0.191 0.712 0.073 [0.135] [0.136] [0.250] [0.284] [0.345] [0.352] [0.381] [0.701] [0.144] Born again 0.172 0.235* 0.533* -0.143 -0.418 -0.094 0.025 1.028 0.263* [0.121] [0.121] [0.246] [0.256] [0.313] [0.307] [0.343] [0.647] [0.129] Female -0.027 0.13 -0.300* 0.068 0.137 -0.814* 0.267 -0.836* 0.147 [0.104] [0.103] [0.179] [0.223] [0.273] [0.271] [0.294] [0.507] [0.112] Nonwhite 0.092 0.082 -0.622* 1.078* 0.322 0.472 -1.413* -3.326* -0.368* [0.161] [0.158] [0.270] [0.347] [0.404] [0.393] [0.538] [1.212] [0.187] Evaluation of Economy 1.576* [0.258] Year 2007 -0.055 0.155 -0.113 -0.118 0.177 [0.103] [0.102] [0.223] [0.293] [0.125] Observations 564 569 250 540 317 321 522 230 566
Models were estimated including only non-Muslim white, black, and Hispanic respondents. Cell entries are ordered probit or logit coefficients, with estimated standard errors in parentheses. Estimates of cut points are not displayed. Warmth and competence of Muslims coded such that higher numbers represent more negative evaluations, relative to the ingroup. Source: 2006 and 2007 CCES. *p<.05 (one-tailed).
48
Bibliography
Alexander, Michele G., Marilynn B. Brewer, and Richard K. Herrmann. 1999. “Images and Affect: A
Functional Analysis of Out-Group Stereotypes.” Journal of Personality and Social Psychology 77(1): 78-93.
Alsultany, Evelyn. 2008. “The Prime Time Plight of the Arab Muslim American after 9/11.” In Amaney
Jamal and Nadine Naber (eds.), Race and Arab Americans Before and After 9/11. Syracuse: Syracuse
University Press. pp. 204-228.
Alwin, Duane F., Jacob L. Felson, Edward T. Walker, and Paula a Tufi. 2006. “Measuring Religious
Identities in Surveys.” Public Opinion Quarterly 70(4): 530-564.
Asch, Solomon E. 1946. “Forming Impressions of Personality.” Journal of Abnormal Social Psychology 42: 258-
290.
Bales, Robert F. 1950. “A Set of Categories for the Analysis of Small Group Interaction.” American
Sociological Review 15(2): 257-263.
Berinsky, Adam. 2004. Silent Voices: Opinion Polls and Political Representation in America. Princeton: Princeton
University Press.
Berinsky, Adam. 2009. In Time of War: Understanding American Public Opinion from World War II to Iraq. Chicago:
University of Chicago Press.
Brewer, Paul R. 2003. “The Shifting Foundations of Public Opinion about Gay Rights.” Journal of Politics 65(4)
1208-1220.
Conover, Pamela J. 1988. “The Role of Social Groups in Political Thinking.” British Journal of Political Science
18: 51-76.
Converse, Phillip E. 1964. “The Nature of Belief Systems in Mass Publics.” In David E. Apter (ed.), Ideology
and Discontent. New York: Free Press.
Cuddy, Amy J.C., Susan T. Fiske, and Peter Glick. 2007. “The BIAS Map: Behaviors from Intergroup Affect
and Stereotypes.” Journal of Personality and Social Psychology 92(4): 631-648.
Davis, Darren W. 2007. Negative Liberty: Public Opinion and the Terrorist Attacks on America. New York: Russell
Sage Foundation.
49
Davis, Darren W. and Brian D. Silver. 2004. “Civil Liberties vs. Security: Public Opinion in the Context of
the Terrorist Attacks on America.” American Journal of Political Science 48(1): 28-46.
Feldman, Stanley and Karen Stenner. 1997. “Perceived Threat and Authoritarianism.” Political Psychology 18(4):
741-770.
Fiske, Susan T., Amy J.C. Cuddy, Peter Glick, and Jun Xu. 2002. “A Model of (Often Mixed) Stereotype
Content: Competence and Warmth Respectively Follow from Perceived Status and Competition.” Journal
of Personality and Social Psychology 82(6): 878-902.
Fiske, Susan T., Amy J.C. Cuddy, and Peter Glick. 2007. “Universal Dimension of Social Cognition: Warmth
and Competence.” Trends in Cognitive Sciences 11(2): 77-83.
Gilens, Martin. 1999. Why Americans Hate Welfare: Race, Media, and the Politics of Antipoverty Policy. Chicago:
University of Chicago Press.
Herrmann, Richard, Philip E. Tetlock, and Penny S. Visser. 1999. “Mass Public Decisions to Go to War: A
Cognitive-Interactionist Framework.” American Political Science Review 93(3): 553-573.
Holsti, Ole R. 1962. “The Belief System and National Images: A Case Study.” Journal of Conflict Resolution 6:
244-52.
Hurwitz, Jon, and Mark Peffley. 1990. “Public Images of the Soviet Union: The Impact of Foreign Policy
Attitudes.” Journal of Politics 52(1): 3-28.
Hurwitz, Jon, and Mark Peffley. 2002. “The Racial Components of ‘Race-Neutral’ Crime Policy Attitudes.”
Political Psychology 23(1): 59-75.
Hurwitz, Jon, and Mark Peffley. 2005. “Playing the Race Card in the Post-Willie Horton Era: The Impact of
Racialized Code Words on Support for Punitive Crime Policy.” Public Opinion Quarterly 69(1): 99-112.
Huddy, Leonie, Stanley Feldman, Charles Taber, and Gallya Lahav. 2005. “Threat, Anxiety, and Support of
Antiterrorism Policies.” American Journal of Political Science 49(3): 593-608.
Kalkan, Kerem Ozan, Geoffrey C. Layman, and Eric M. Uslaner. “‘Band of Others’? Attitudes toward
Muslims in Contemporary American Society.” Journal of Politics 71:847-872.
50
Kam, Cindy D., and Donald R. Kinder. 2007. “Terror and Ethnocentrism: Foundations of American
Support for the War on Terror.” Journal of Politics 69(2): 320-338.
Karim, Karim H. 2003. Islamic Peril: Media and Global Violence. Montreal: Black Rose Books.
Kennedy, Peter. 1992. A Guide to Econometrics (3rd edition). Cambridge, MA: MIT Press.
Kinder Donald R., and Cindy D. Kam. 2009. Us Against Them: Ethnocentric Foundations of American Opinion.
Chicago: University of Chicago Press.
Kinder, Donald R. and Lynn M. Sanders. 1996. Divided By Color: Racial Politics and Democratic Ideals. Chicago:
University of Chicago Press.
Kinder, Donald R. and Nicholas Winter. 2001. “Exploring the Racial Divide: Blacks, Whites, and Opinion on
National Policy.” American Journal of Political Science 45(2) 439-453.
Lippman, Walter. 1922. Public Opinion. New York: Free Press
Mendelberg, Tali. 2001. The Race Card. Princeton: Princeton University Press.
Moore, Kathleen. 2002. “A Part of US or Apart from US?: Post September 11 Attitudes toward Muslims
and Civil Liberties.” Middle East Report 224:32-35.
Nacos, Brigitte L., and Oscar Torres-Reyna. 2003. “Framing Muslim-Americans Before and After 9/11.” In
Pippa Norris, Montague Kern, and Marion Just (eds.), Framing Terrorism: The News Media, the Government,
and the Public. New York: Routledge.
Nacos, Brigette L.., and Oscar Torres-Reyna. 2007. Fueling Our Fears: Stereotyping, Media Coverage and Public
Opinion of Muslim Americans. Lanham: Rowman & Littlefield Publishers.
Nelson, Thomas E. and Donald R. Kinder. 1996. “Issue Frames and Group-Centrism in American Public
Opinion. Journal of Politics 58(4):1055-78.
Nisbet, Erik C., Ronald Ostman, and James Shanahan. 2007. “Public Opinion toward Muslim Americans:
Civil Liberties and the Role of Religion, Ideology, and Media Use.” In Abdulkader Sinno (ed.), Muslims in
Western Politics. Bloomington, IN: Indiana University Press.
Panagopoulos, Costas. 2006. “The Polls—Trends: Arab and Muslim Americans and Islam in the Aftermath
of 9/11.” Public Opinion Quarterly 70(4): 608-624.
51
Peffley, Mark, Jon Hurwitz, and Paul Sniderman. 1997. “Racial Stereotypes and Whites’ Political Views of
Blacks in the Context of Welfare and Crime.” American Journal of Political Science 41(1): 30-60.
Peretz, Martin. 1996. “Israel Bonds.” New Republic, 5 August, pp.12-16.
Pew Center for People and the Press. 2007. “Muslim Americans: Middle Class and Mostly Mainstream.”
http://pewresearch.org/assets/pdf/muslim-americans.pdf.
Phalet, Karen, and Edwin Poppe. 1997. “Competence and Morality Dimensions of National and Ethnic
Stereotypes: A Study in Six Eastern European Countries.” European Journal of Social Psychology 27: 703-723.
Poole, Elizabeth. 2002. Reporting Islam: Media Representations of British Muslims. London: I.B. Tauris & Co.
Rosenberg, S., C. Nelson, and P.S. Vivekananthan. 1968. “A Multidimensional Approach to the Structure of
Personality Impressions.” Journal of Personality and Social Psychology 9: 283-294.
Said, Edward W. 1997. Covering Islam: How the Media and the Experts Determime How We See the Rest of the World.
New York: Vintage.
Schildkraut, Deborah J. 2002. “The More Things Change…American Identity and Mass and Elite
Responses to 9/11.” Political Psychology 23(3): 511-535.
Shaheen, Jack. 2009. Reel Bad Arabs: How Hollywood Vilifies a People. Northampton, MA: Olive Branch Press.
Sheikh, Kashif Z., Vincent Price, and Hayg Oshagan. 1995. “Press Treatment of Islam: What Kind of
Picture Do the Media Paint?” International Communication Gazette 56: 139-154.
Slade, Shelley. 1981. “The Image of the Arab in America: Analysis of a Poll on American Attitudes.” Middle
East Journal 35 (2): 143-162
Sniderman, Paul M., and Thomas Piazza. 1993. The Scar of Race. Cambridge: Harvard University Press.
Stephan, Walter G., et al. 2002. “The Role of Threats in the Racial Attitudes of Blacks and Whites.”
Personality and Social Psychology Bulletin 28(9): 1242-54.
Sullivan, John L., James E. Piereson, and George E Marcus. 1982. Political Tolerance and American Democracy.
Chicago: University of Chicago Press.
Tajfel, Henri, and John Turner. 1979. “An Integrative Theory of Group Conflict.” In W.G. Austin and
S. Worchel (eds.), The Social Psychology of Intergroup Relations. Monterey, CA: Brooks/Cole. pp. 33-48.
52
Tourangeau, Roger, Mick P. Couper, and Frederick Conrad. 2007. “Color, Labels, and Interpretive
Heuristics for Response Scales.” Public Opinion Quarterly 71(1): 91-112.
Traugott, Michael et al. 2002. “How Americans Responded: A Study of Public Reactions to 9/11/01.” PS:
Political Science and Politics 35(3): 511-516.
Viorst, Milton. 1994. Sandcastles: The Arabs in Search of the Modern World. New York: Alfred A. Knopf.
Weston, Mary Ann. 2003. “Post 9/11 Arab American Coverage Avoids Stereotypes.” Newspaper Research
Journal 24 (1): 92-106.
Wojciszke, Bogdan. 2005. “Morality and Competence in Person- and Self-Perception.” European Review of
Social Psychology 16: 155-188.
Wojciszke, Bogdan, Roza Bazinska, and Marcin Jaworski. 1998. “On the Dominance of Moral Categories in
Impression Formation.” Personality and Social Psychology Bulletin 24: 1251-1263.
Winter, Nicholas J.G. 2008. Dangerous Frames. Chicago: University of Chicago Press.
Ybarra, O., W.G. Stephan, and L.A. Schaberg. 2000. “Misanthropic Memory for the Behavior of Group
Members.” Personality and Social Psychology Bulletin 26: 1515-1525.