Abstract
To date, research on Christians’ attitudes toward Muslims has generally focused on the cognitive components of these attitudes. Based on Eagly and Chaiken’s conceptualization of the ABC Model of Attitudes, this study extended that research by considering multiple evaluative aspects of attitudes. Relevant questionnaires from a larger study of 209 self-identified Christians were used to predict attitudes toward American Muslims. It was hypothesized that adding a behavioral component beyond the cognitive would significantly increase prediction of attitudes toward Muslims in the United States. Drawing on research on related prejudices, it was also predicted that the often-neglected affect of disgust would significantly increase the prediction of attitudes toward Muslims beyond only combining behavioral and cognitive factors. In this study, behavioral factors included engagement with people of different faiths, while cognitive factors included dogmatism and endorsement of Christian nationalism. A hierarchical regression analysis indicated that combining behavioral with cognitive components significantly increased the prediction of American Christians’ attitudes toward American Muslims. Yet, further adding the affective component of physical disgust beyond the cognitive and behavioral components did not significantly increase predictive power. Implications of these findings for the Christian church and American society at large are addressed in the light of the important social implications of understanding attitudes toward Muslims. The need for future research to emphasize the multifaceted nature of anti-Muslim attitudes is also discussed.
The history of relations between Christians and Muslims ranges from congenial coexistence to mutual distrust to violent conflict (Thomas, 2017). Specifically, within the United States, there is a long-standing record of animosity among some Christian groups toward American Muslims despite current efforts at reconciliation (Merino, 2010; Różycka-Tran, 2017). A better understanding of the sources of this animosity is important given the potential destructive social consequences. This is especially relevant given the significant increase in hate crimes committed against Arab-Americans since the 2001 attack on the World Trade Center and the Pentagon (Anderson, 2002; Kishi, 2017).
Studies of Christian–Muslim relations in the United States have considered a broad range of factors related to distrust of the other. These include sociocultural and situational factors (Schlueter et al., 2020; Shaver et al., 2017). Another key contributor to the nature of Christian–Muslim relations are attitudes, which are evaluative and include ratings such as “good,” “bad, “trustworthy,” and “dishonest” (Ajzen & Fishbein, 2000; Bravo López, 2011; Ciftci, 2012). To date, efforts to understand American Christians’ attitudes toward Muslims in the United States have been hindered in part because researchers have often focused on a more general conceptual understanding of attitudes (e.g., “Islamophobia”). That approach emphasizes primarily the cognitive (thoughts/beliefs) aspect of attitudes regarding Muslims and underemphasizes the multifaceted nature of attitudes (Bravo López, 2011; Ciftci, 2012; Tapias et al., 2007). In the present study, I use Eagly and Chaiken’s (2007) more comprehensive elaboration of the so-called ABC theory of attitudes, which proposes that attitudes are evaluative tendencies that encompass affective, behavioral, and cognitive facets as three types of evaluative responding toward an object. Specifically, this study explores how incorporating these three evaluative components of attitudes can together better predict American Christians’ attitudes toward American Muslims than can any of these attitude facets alone.
First, I discuss the relevance of attitudes toward Muslims in terms of the implications of Christian–Muslim relations in the United States and gaps in the current research. Second, I present the conceptual aspects of Eagly and Chaiken’s (1998, 2007) elaboration of the ABC theory of attitudes and its relevance for addressing research in this area. Then, I review the findings regarding how combining the aforementioned aspects of attitudes better predicts American Christians’ attitudes toward American Muslims. Finally, I discuss implications of these findings in relation to strategies for reconciliation efforts between Christians and Muslims in the United States.
American Christian–Muslim Attitudes and Relations
Overall, compared to other religious groups in the United States, Muslims report experiencing the highest level of religious discrimination by all respondents, regardless of faith identity (Mogahed & Mahmood, 2019). Common perceptions include believing that American Muslims are more prone to violence and are more hostile to the United States. Pew Research Center (2017) noted, in their analysis of the Federal Bureau of Investigation (FBI) hate crimes statistics that assaults on Muslims in the United States increased significantly between 2015 and 2016, surpassing the 2001 statistics. These include physical assaults as well as vandalism, intimidation, and other forms of discrimination. Of course not all hateful acts against Muslims are committed by Christians, but over 71% of Americans profess a Christian faith, and Islam is the fastest growing religion in the United States. Thus, a better understanding of the relationship and attitudes between the two groups is a worthwhile pursuit.
Research on American Christians’ attitudes toward American Muslims suggests a significant amount of anti-Muslim sentiment in general (Bridge Initiative, 2016; Różycka-Tran, 2017), though with mixed results. For example, research on denominational affiliation has yet to identify a consistent difference between Protestants’ and Catholics’ views of Muslims (Bridge Initiative, 2016). Researchers also generally neglect to consider the attitudes of a significant number of Orthodox Christians (Sharp, 2012). Race and ethnicity also interact with Christian faith and are associated with differing perceptions of Muslims. For example, Sherkat and Lehman (2018) found that among Christians, non-White Catholics reported the most positive views toward Muslim immigrants. Research on the effects of conservatism of religion also produces mixed results. For example, some studies (e.g., Jung, 2012) found level of conservatism related to higher levels of anti-Muslim attitudes, while other studies (e.g., S. D. Johnson, 2006; Ribberink et al., 2017) found that conservative, orthodox Christian groups are more tolerant of Muslims. Political identification may also be related to Christians’ attitudes toward Muslims and perhaps confound results about effects of religion (Mogahed & Mahmood, 2019).
Gaps in the Research
Given the multifaceted nature of attitudes, the above-noted inconsistency in the findings may result from assessing different components of attitudes. Most measures tend to focus on cognitive perceptions (e.g., whether Muslims are trustworthy; Raiya et al., 2008; Różycka-Tran, 2017). Some measures (e.g., Pew Research Center, 2017) focus almost exclusively on general feelings toward Muslims or neglect important affective factors beyond fear (Kunst et al., 2013). Other measures consider anti-Muslim prejudice, sentiment, and conspiracy beliefs (e.g., Uenal et al., 2021), but not behavior.
Some research attempts to provide a more inclusive assessment of the multiple aspects of attitudes by using the Islamophobia Scale (Lee et al., 2009). Yet, the only affective component that scale measures is fear. In addition, the affective and behavioral aspects are combined in the items such that the two are not readily distinguishable from each other as they often are in real life (Eagly & Chaiken, 1998). In the present study, I address these gaps in research by exploring the predictive power of combining affective, behavioral, and cognitive processes (the so-called ABC’s of attitudes; Eagly & Chaiken, 1998, 2007) to assess American Christians’ attitude toward American Muslims.
Research on American Christians’ attitudes toward Muslims does not often distinguish between American Muslims and Muslims in general. This is because Muslims are most often perceived as a monolithic group rather than a diverse group culturally and ideologically (Abu-Ras & Suarez, 2009; Nyang, 1999, as cited in Khan & Ecklund, 2013). In other words, attitudes toward Muslims in general are believed to predict attitudes toward specific Muslim groups. In the present study, I assessed attitudes toward American Muslims specifically.
The ABC’s of Attitudes
Researchers (e.g., Allport, 1954; Breckler, 1984) have long assumed that attitudes are complex phenomena consisting of motivational, emotional, behavioral, perceptual, and cognitive components. The ABC model of attitudes suggests that attitudes are comprised of affective, behavioral, and cognitive components that are distinct, but not mutually exclusive, dimensions. Eagly and Chaiken’s (2007) inclusive model of attitudes specifically proposes that attitudes are evaluative tendencies toward an attitude object. An attitude object refers to the circumscribed entity toward which the attitude is directed. The attitude object can be abstract (e.g., conservatism) or concrete (e.g., the Sears Tower; Eagly & Chaiken, 1998). According to Eagly and Chaiken (2007), the attitude object is an essential feature because it distinguishes the concept of attitude from other psychological concepts such as moods. While moods also involve evaluative reactions, the reactions are often more diffuse rather than directed toward a specific entity. In the present study, the specific attitude object is the social group American Muslims.
According to Eagly and Chaiken (2007), attitudes are evaluative tendencies rather than relatively fixed personality dispositions even though attitudes differ in strength. The evaluative tendencies of attitudes encompass all classes of evaluative responding, including feelings and emotions, overt behavior, and beliefs and thoughts (the so-called ABC’s of attitudes). Eagly and Chaiken propose that these evaluative tendencies could be overt or covert and are not necessarily conscious.
In general, there is consistency among the affective, behavioral, and cognitive evaluative responses of attitudes (Eagly & Chaiken, 1998). But there can also be ambivalence and incongruence at times. Because people tend to prefer consistency in their evaluative attitudes, they will often impose it when it is lacking to avoid cognitive dissonance (Cooper, 2019; Festinger, 1957). For example, one may feel fearful of a dissimilar other (affect), but be intentional to befriend them (behavior) due to thoughts that racism is unethical (cognitions). Consistent befriending of dissimilar others may in turn reduce negative feelings about that social group and affirm one’s thoughts regarding the ethical treatment of others. In contrast, whether realistic or not, perceived fear that another group is dangerous often results in greater belief that the other group is untrustworthy as well as behaviors that seek to avoid members of that group (Ciftci, 2012). The three evaluative responses of attitudes thus often influence one another in a variety of reciprocal ways to achieve consistency among them (Eagly & Chaiken, 1998, 2007).
The three evaluative responses are also used to inform oneself regarding the nature of one’s attitudes. For example, if one is not sure how they feel toward the issue of climate change, they may arrive at an answer by recalling their past behaviors that are relevant to that issue (e.g., whether or not they recycle regularly). Similarly, if one is assessing their attitude toward a social group, they may recall their previous feelings or behaviors toward that group.
Eagly and Chaiken’s (2007) model of attitudes specifically focuses on the evaluative nature of attitudes, not on the strength of attitudes or on the various other functions (e.g., ego defense) of attitudes. From that view, all three evaluative aspects of attitudes are considered an integral part of any evaluative response. Thus, in the present study, American Christians’ attitudes toward American Muslims are examined in terms of its multifaceted evaluative structure, not its origins, strength, or other purpose(s).
Affective component
The affective evaluative tendencies of attitudes consist of emotional responses prompted by a person, object, event, and so on. These affective features often consist of a sympathetic nervous system response and cover a wide range of positive and negative feelings (e.g., liking or disliking, anger, warmth, fear). They can also include ambivalent feelings about the same attitude object (Eagly & Chaiken, 1998, 2007). Research regarding the affective component of Christians’ attitudes toward Muslims has generally focused on fear and its strong role in the perceived threat that Muslim individuals and groups pose to one’s own group (Altheide David, 2017; Andersen & Mayerl, 2018).
Haner et al. (2019) found that highly religious Christians reported a significant amount of fear regarding both American and foreign Muslims as potential terrorists as well as support of more stringent immigration policies (Dunwoody & McFarland, 2018; Zimmerman, 2008). Raiya et al. (2008) additionally found that Christians who feared that Muslims would “desecrate” Christianity had higher anti-Muslim attitudes than those who did not have this fear. Fear of American Muslims has also been associated with issues related to perceived contamination of the purity and integrity of Christianity (Haidt et al., 1994, 1997; Hodson et al., 2014; Taylor, 2007).
The primary focus on fear neglects other potentially relevant emotions including disgust, which is often associated with prejudice toward dissimilar others, including Muslims (Hodson et al., 2014; Petersen, 2017; Stephan et al., 2009; van Leeuwen & Petersen, 2018). Disgust refers to feelings of revulsion that are often accompanied by behavioral withdrawal due to fear of contamination (W. I. Miller, 1997; Rozin et al., 1999). The present study expands beyond the historical focus on fear as the primary emotion in anti-Muslim attitudes by using a measure of general physical disgust to assess its relation to American Christians’ attitudes toward Muslims. Some research suggests that negative attitudes toward American Muslims could be related not just to disgust toward Muslims specifically, but also to general disgust sensitivity. Hodson et al. (2014), for example, found that disgust sensitivity and disease concerns were significantly related to specific intergroup disgust sensitivity.
Given the relationship between general physical disgust and disgust toward specific perceived outgroups, I predicted that physical disgust will be significantly related to the cognitive and behavioral measures described in the following sections. This prediction is in accordance with Eagly and Chaiken’s (2007) model, which proposes that all three are important evaluative responses toward Muslims. In addition, I assessed whether disgust added incrementally to the prediction of attitudes toward American Muslims over and above that predicted by the cognitive and behavioral measures.
Behavioral component
Eagly and Chaiken (2007) argue that in addition to affect, behaviors constitute another integral evaluative response elicited by attitude objects. This is especially true given the far-reaching impact of both prosocial and antisocial behaviors. Behavioral responses associated with attitudes can be overt actions toward the attitude object as well as intentions to act that are not expressed in covert behavior. They include a wide array of possible actions such as befriending, avoiding, and discriminating against. From this view, behaviors are arguably the most explicit evaluative aspect of the attitude.
Research on the behavioral aspects of anti-Muslim attitudes has generally focused on negative, discriminatory actions such as support for a travel ban on Muslims or discrimination in hiring (Ali et al., 2015; Foundation for Ethnic Understanding, 2019). In the current study, the behavioral measure instead assessed the extent of more positive behaviors such as openness to religious others as demonstrated by active efforts to learn about and engage with them (K. A. Johnson et al., 2018). This measure provided by the archival data set did not assess behavior toward Muslims specifically. Nevertheless, research has demonstrated that just as a generalized prejudice can be targeted across specific targets (Akrami et al., 2011), general tolerance toward religious perceived outgroups is significantly related to one’s propensity toward openness to engage with specific religious groups, including Muslims (Brewer & Pierce, 2005; Mummendey & Wenzel, 1999).
Based on Eagly and Chaiken’s (2007) model of attitudes, I thus expected that the behavioral measure would be significantly related to both the affective and cognitive evaluative measures. This prediction is based on the integral importance of the behavioral evaluative responses toward attitude objects. In addition, I predicted that the behavioral measure would significantly add to the prediction of attitudes toward Muslims over and above the cognitive measures alone, which are the most often assessed.
Cognitive component
The cognitive component of attitudes includes specific beliefs, associations, or linkages that one establishes between the attitude object and various attributes that one ascribes to that object (Eagly & Chaiken, 1998). Research that has emphasized the cognitive aspects of Christians’ anti-Muslim attitudes has focused on different cognitive-motivational factors such as authoritarianism (Altemeyer, 1998; Duckitt, 2015) and social-dominance orientation (Sidanius & Pratto, 1993, 1999). Such cognitions are often associated with anti-Muslim attitudes and discriminatory actions (Duckitt, 2001; S. D. Johnson, 2006). This study focused on the following two important cognitions which have also been related to anti-Muslim attitudes.
Christian nationalism
Christian nationalism includes beliefs regarding America having a unique, God-appointed role in the world. This unique role is conceptualized especially with regard to the parallels between the United States and ancient Israel. From this view, both nations were commanded by God to preserve their cultural and blood purity, even if this requires war and separatism. Appeals to Christian nationalism are often quite explicitly evangelical and often exclusionary with regard to other religious faiths (including Islam) or cultures (Whitehead & Perry, 2020; Shortle & Gaddie, 2015).
Dogmatism
Dogmatism refers to a rigid cognitive style characterized by a high degree of certainty regarding one’s beliefs. This certainty is relatively impervious to challenge or change and is often accompanied by intolerance toward others who disagree or are otherwise perceived as belonging to outgroups (Rokeach, 1948, 1951, 1954, 1960). In his original conceptualization of the term, Rokeach (1960) noted that a generally dogmatic style often stems from belief in an ultimate authority. Later research (e.g., Bronstein et al., 2019; Jelen & Wilcox, 1991) suggests a significant relationship between general dogmatism and religious dogmatism. That research refers to dogmatism as a cognitive style in which religious ideas can be considered a specific type of general dogmatism.
Both religious and general dogmatism are characterized by certitude and a tendency to dismiss challenges to one’s beliefs regarding outgroups. These beliefs are associated with a negative attitude toward outgroup members (e.g., Muslims). For example, research (e.g., Wright & Khoo, 2019) suggests that both religious dogmatists and atheists (who demonstrate a more general dogmatism) tend to have equally negative views toward Muslims. Thus, the measure used in this study assessed the more general form of dogmatism (Altemeyer, 1996) that was available in the archival data set.
Rigid, absolutist, and exclusivist thinking are characteristic of both Christian nationalism and dogmatism. These two cognitive styles are hallmark facets of outgroup prejudice. Thus, consistent with Eagly and Chaiken’s (2007) ABC model of attitudes, I predicted that they would together significantly predict attitudes toward Muslims. Furthermore, the cognitive measures should be significantly correlated to the affective and behavioral measures.
Rationale and Current Study
To review, the present study assessed American Christians’ attitudes toward American Muslims using Eagly and Chaiken’s (2007) conceptualization of the ABC theory of attitudes. Most studies to date focus on the cognitive (beliefs) aspects of attitudes toward Muslims. Those studies emphasize the thoughts one has about Muslims (e.g., “All Arabs are Muslims” or “Male Muslims are all misogynist”). Other studies include one of the other two evaluative facets of attitudes. For example, studies that focus on the affective component measure the way one feels towards Muslims (fearful, warm, etc.). Studies that focus on the behavioral component of attitudes emphasize ways in which one interacts with Muslims (e.g., avoiding, embracing, aggressing against, etc.). In the current study, I sought to assess all three elements to provide a more comprehensive assessment of attitudes toward Muslims. In addition, to date, most research that emphasizes the affective component focuses on fear as the primary affective component. In this study, I instead assessed general physical disgust, which is another important affective variable related to attitudes toward specific perceived outgroup members.
I hypothesized in the present study that the affective, behavioral, and cognitive measures would be significantly correlated with one another, as each is an integral part of the evaluative nature of attitudes (Eagly & Chaiken, 1998, 2007). I further predicted that adding the behavioral component (engagement with religious others) to the cognitive factors (Christian nationalism and dogmatism) would significantly increase prediction of attitudes toward Muslims. In addition, it was hypothesized that physical disgust would add incremental predictive power beyond that of the cognitive and behavioral factors alone. Hierarchical multiple regression analysis was used in accordance with the prediction that acknowledging the multifaceted nature of attitudes allows for examination of the individual as well as the collaborative contributions of each component for the prediction of attitudes toward Muslims.
Method
Participants
The sample for this study was obtained from an archival data set related to various aspects of religiosity. The data had been gathered by other researchers in collaboration with the Foundation for Religious Diplomacy. Participants had been pre-screened via Amazon’s MTurk for religious affiliation. Only those who self-identified as Christian were admitted into the online Qualtrics study.
This sample consisted of 209 participants, 99 (47.4%) females and 110 (52.6%) males. Participants ranged in age from 20 to 70 years old (M = 36, SD = 10). Over 41% of the participants self-identified as Catholic, just over 17% as Evangelical, and 41.1% as other Protestant affiliations. Collectively, these individuals self-reported a moderate amount of religious engagement, as noted in the Religious Commitment Inventory (RCI-10; M = 30.5; SD = 11). In addition, 58% reported attending services at their place of worship at least once per month or more. Slightly more than half also reported attending other religious events such as Bible studies or prayer groups at least once per month. Participants were 78% Euro-American, 10.5% African American/Black, 5.8% Asian or Asian American, 4.8% Hispanic, and 1% multiple ethnicity.
Measures
Participants completed a large number of surveys for a broader cross-sectional survey study. Only those questionnaires relevant to the aims of the current study were included in these analyses.
RCI-10
This 10-item scale measures the extent of one’s commitment to religious beliefs, values, and practices and the degree to which this is implemented into daily living (Worthington et al., 2003). It was used for this study as an additional verification of participants’ self-professed Christianity. The measure uses a 5-point Likert-type scale where 1 = not at all true of me and 5 = totally true of me. The measure also has two subscales: interpersonal (four questions; example: “I enjoy spending time with others of my religious affiliation”) and intrapersonal (six questions; example: “My religious beliefs lie behind my whole approach to life”). The subscale scores were combined for a total score. Previous research (e.g., Worthington et al., 2003) has demonstrated high convergent and discriminant validity for the RCI-10. The Cronbach’s alpha (α = .93) for the scale in this study indicates high reliability.
Disgust Scale
This was the primary affective measure used in this study (Haidt et al., 1994). It has two 10-item subscales. One subscale is related to level of disgust in relation to physical things (e.g., “stepping into dog poop”; touching a stranger’s feet). The other subscale is related to disgust about moral issues (e.g., stealing from a neighbor; marital infidelity). For each item, participants are asked to rate how disgusting certain situations are using a Likert-type scale from 1 = not at all disgusting to 7 = extremely disgusting. For this study, the focus was on the physical disgust subscale. As noted earlier, the rationale for this is that the propensity for physical disgust more generally is related to disgust for members of perceived outgroups, including Muslims (Hodson et al., 2014). Previous research (Olatunji et al., 2007) has demonstrated high construct and discriminant validity for the physical disgust subscale. In the present study, the Cronbach’s alpha (α = .88) additionally showed high reliability for this subscale.
Religious Diversity Scale
This five-item scale was adapted by K. A. Johnson et al. (2018) from the Cultural Diversity Scale (Kottke, 2011). It was the primary behavioral measure. It assesses engagement with and active learning about people of other religious groups. This scale did not assess engagement with Muslims specifically. Yet, as discussed previously, research suggests that a more general willingness to engage with religious others is significantly related to openness to engage with specific religious groups, including Muslims (Brewer & Pierce, 2005). Participants were asked to respond using a Likert-type scale from 1 = strongly disagree to 7 = strongly agree. Sample items include, “I often have contact with people from other religious groups” and “I am familiar with the teachings or philosophies of other religious groups.” Higher scores on this scale indicate more interreligious engagement. In previous research, the scale has demonstrated good internal consistency and convergent validity, as it is significantly related to scales assessing positive attitude toward religiously dissimilar others (K. A. Johnson et al., 2018). In the present study, Cronbach’s alpha (α = .85) also showed high reliability for this scale.
Christian Nationalism Scale
This four-item brief questionnaire was used as one of the cognitive measures (Shortle & Gaddie, 2015). It measures a combination of patriotic beliefs that emphasize the prominent role of America in the world and the integral role that Christianity plays in America and God’s special blessing on it. The scale has Likert-type items with five options that range from strongly agree to strongly disagree. The items included “America holds a special place in God’s plan,” “God has chosen this nation to lead the world,” “The United States was founded as a Christian nation,” and “It is important to preserve the nation’s religious heritage.” This scale has demonstrated high convergent validity, correlating significantly with negative views of Muslims (Shortle & Gaddie, 2015). In the present study, the scale also showed high reliability (α = .88).
Dogmatism Scale
This was the second cognitive scale used in this study (Altemeyer, 1996). It assesses general rigid thinking characterized by a closed perspective that is relatively impervious to challenge or change. As noted earlier, general dogmatic thinking is significantly related to religious dogmatism (Jelen & Wilcox, 1991; Rokeach, 1960). To reduce participant fatigue, only 7 items from a 12-item shortened version of Altemeyer’s (1996) Dogmatism Scale were administered in the original study. The selection of these seven items is consistent with previous confirmatory factor analyses that found that the original dogmatism scale is unidimensional and that various shortened versions retain good internal consistency and construct validity (Crowson, 2009; Shearman & Levine, 2006). Participants responded on a Likert-type scale from 1 = does not describe my feelings to 5 = clearly describes my feelings. A sample item is “I am so sure I am right about the important things in life; there is no evidence that could convince me otherwise.” Three of the items were reversed-scored so that higher overall scores indicate higher levels of dogmatism. Cronbach’s alpha was high (α = .84), indicating good internal consistency.
Attitudes toward religious groups
This novel five-item scale provided by the data set was used as the dependent measure. It asks participants to rate their beliefs regarding the integrity, trustworthiness, and benevolence of specific religious groups. It also measures the participant’s feelings toward that group, including level of fear as well as their overall warm or cold feelings toward people in that group. In this study, only the results specifically pertaining to American Muslims were used. Sample items include, “To what extent do you think the religious group has good morals?” and “. . . is likely to cause harm.” Total scores on this scale provide an estimate of an individual’s summary attitude regarding American Muslims. Questions consisted of a Likert-type scale from 1 to 7. Several items were reverse scored so that higher scores indicate more negative attitudes toward Muslims. As noted in Table 1, evidence of the convergent validity of this novel scale is seen in the significant positive correlations with various scales generally associated with negative attitudes toward Muslims, including the cognitive measures (dogmatism and Christian nationalism) and the affective measure (physical disgust). Evidence of convergent validity of the scale is seen in its negative correlation with the Religious Diversity Scale, which measures openness to engaging with religious others. In addition, Cronbach’s alpha (α = .89) indicates good internal consistency for this novel scale.
Descriptive Statistics and Pearson Correlations for All Continuous Variables.
SD: standard deviation.
p < .010.
Procedure
Participants followed a link via MechanicalTurk, the online platform, to the study that was available through Qualtrics. First, they were presented with a prescreening question regarding religious affiliation. Only those who identified as Christian were able to continue the survey. The remaining eligible participants were directed to the Informed Consent page, which indicated that this was a study about beliefs and attitudes that would take between 20 and 30 minutes to complete. Participants were also informed of the small ($1.50) compensation they would receive for taking part in the study. Next, the RCI-10 was presented. The order of the remaining two blocks of questionnaires, including those used in this study, was then randomly counterbalanced using Qualtric’s randomization procedure. Within each block, the order of the questionnaires as well as the order of the items within the questionnaires were also randomized evenly across participants. The first block included the Disgust Scale as well as the shortened Dogmatism and Christian Nationalism Scales. The second block contained the Religious Diversity Scale and Attitudes Toward Religious Groups Scale. Finally, participants were presented with a brief demographic questionnaire and a final consent to allow their data to be used. A total of 42 participants were excluded due to conflicting self-reports of their faith (e.g., claiming to be Christian, yet later self-reporting that they did not believe God exists) or for failing attention checks. These exclusions were made prior to conducting any data analyses. No other data management techniques were used.
Results
Means, standard deviations, and correlations for all the continuous variables, that is, questionnaires used in the regression models, are shown in Table 1. Collectively, the skewness of all the measures (–.6 to .5) was well within the recommended range. In addition, the kurtosis for all the scales was under 1. Observations of histograms and Q–Q plots also revealed roughly normal distributions for the standardized residuals. Thus, it can be concluded that the data for all the variables reasonably met guidelines for univariate normality (Weston & Gore, 2006).
Table 1 reveals significant low to moderate correlations overall. Note that the highest correlations, in descending order, were between the cognitive components measured (Christian Nationalism and dogmatism). The remaining highest correlations included one of these two cognitive factors, that is, Christian Nationalism and physical disgust, dogmatism and attitudes toward American Muslims, and finally Christian Nationalism and attitudes toward American Muslims.
Hierarchical multiple regression
Before conducting the hierarchical multiple regression analysis, tests of its assumptions were conducted. As noted above, univariate normality was established. Questionnaire scores were converted to standardized z scores to assess for univariate outliers. Any z score that was 1.5 times beyond the interquartile range, that is, z score greater than ±2.68 was considered an outlier. No outliers were detected.
Multivariate outlier detection was performed using Mahalanobis’ distance (DM2), and this indicated no outlying values (all p values > .001; Cain et al., 2017). Levene’s test indicated homogeneity of error variance across the dependent variables. Multicollinearity was additionally ruled out, as tolerance values were all greater than 0.1 (VIF < 2) for all measures.
There were no leverage values greater than 0.2, nor values for Cook’s distance above 1 (all Cook’s d < .2). There was also evidence of homoscedasticity, as assessed by visual inspection of a plot of studentized residuals versus unstandardized predicted values. In addition, bivariate scatterplots among all the questionnaire scores indicated that the linearity assumption was met. There was independence of residuals (Durbin–Watson statistic = 2.028).
In the first step of the hierarchical multiple regression, the two predictors related to the cognitive aspect of attitudes, that is, Dogmatism and Christian Nationalism, were entered. The results indicated that these two factors significantly predicted attitudes toward American Muslims, R2 = .121, F(2, 208) = 14.116, p < .0005. The addition of the behavioral measure regarding engagement with other religious groups in Model 2 led to a statistically significant increase in prediction of attitudes toward American Muslims, R2 increase of .065, F(1, 205) = 16.386, p < .0005. Contrary to prediction, in Model 3, the addition of the affective measure of physical disgust did not further significantly increase the prediction of attitudes toward American Muslims above and beyond that accounted for by the cognitive and behavioral measures, R2 increase of .010; F(1, 204) = 2.633, p < .106.
The full model including Christian Nationalism, dogmatism, behavioral engagement with different religions, and physical disgust to predict attitudes toward Muslims (Model 3) was statistically significant (adjusted R2 = .196), F(4, 204) = 12.4333, p < .0005. Yet, in the final adjusted model, all of the predictor variables except for physical disgust were statistically significant, with behavioral engagement the best predictor (B = –.367; β = –.264; p < .0005), followed by dogmatism (B = .195; β = .177; p < .015), and then Christian Nationalism (B = .171; β = .151; p < .05), and finally physical disgust (B = .066; β = .111; p < .106). When the effects of the other variables were controlled for in Model 3, it was the behavioral component that had the greatest relationship with attitudes toward Muslims (partial r = –.280; p < .0005).
Discussion
This study had three primary predictions based on Eagly and Chaiken’s (2007) elaboration of the ABC model of attitudes. First, given the multifaceted nature of attitudes, I predicted that there would be significant correlations among the affective, behavioral, and cognitive measures. This hypothesis was supported. Second, I predicted that adding the behavioral to the cognitive component would significantly increase power to predict attitudes toward Muslims. This expectation was supported and highlights the importance of considering multiple facets of attitudes. Interestingly, behavior was actually the strongest predictor of attitudes toward Muslims. This is consistent with research that finds that interacting with and befriending Muslims significantly decreases anti-Muslim attitudes, even if one begins with negative cognitions about them (e.g., Mogahed & Mahmood, 2019).
The third main prediction was that adding the affective component of physical disgust would further increase the power for predicting attitudes toward Muslims beyond the combination of cognitive and behavioral components. This hypothesis was not supported in a hierarchical multiple regression. The correlation between physical disgust and attitudes toward Muslims was significant but low. When controlling for the influence of the cognitive and behavioral aspects, the partial correlation was even lower.
The simplest interpretation of the lack of significance of the contribution of physical disgust is that for this Christian sample, physical disgust is not a significant contributor to attitudes toward Muslims. This would be analogous to the finding that negative attitudes toward atheists are “not the result of disgust, but of mistrust” (Gervais et al., 2011). From that view, and in light of previous findings regarding fear and attitudes toward Muslims, it is the concern/fear evoked by mistrust, rather than disgust per se, that contribute to negative attitudes. The data in this sample may indirectly corroborate this interpretation. Alternatively, the lack of significance regarding the contribution of physical disgust may be the result of the measure used, as noted below.
Limitations of the current study and future research implications
Validity of the scales used
It is plausible that the lack of significant findings for disgust may be related to the construct validity of the physical disgust measure as used in the context of this study. Specifically, the physical disgust measure (Haidt et al., 1994), though a highly validated scale, did not assess disgust toward Muslims in particular. Although research suggests that general physical disgust is significantly related to disgust toward specific groups including Muslims (Hodson et al., 2014), the scale used may not have been precise enough to detect Muslim-specific disgust. This possible lack of connection between general proclivity toward physical disgust and disgust toward Muslims specifically may have underestimated the role of disgust in attitudes toward American Muslims in this study. Of note, though physical disgust did not add significantly toward the prediction of attitudes toward Muslims beyond the cognitive and behavioral measures, it did correlate significantly (though at a low level) with the attitudes measure in this study. It is not possible to know for certain whether the significant correlation is a statistical artifact given the sample size. Future research should investigate more contextual measures of disgust toward Muslims to more accurately assess the connection between physical disgust and racism and prejudice (Choma et al., 2016; Goplen & Plant, 2015; Hodson & Costello, 2007).
The measure used to assess attitudes toward American Muslims, while having high internal consistency, may not have been sufficiently indicative of the various aspects related to attitudes toward Muslims. In this study, it is possible that the cognitive and behavioral aspects were in themselves strong enough predictors to overcome this potential limitation of the attitudes questionnaire. Nevertheless, a more representative measure such as the Islamophobia Scale (Lee et al., 2009) could be helpful in future research to help detect a possible relationship between attitudes toward Muslims and physical disgust.
Another concern related to the construct validity of the scales used involves the Religious Diversity Scale. This behavioral scale assessed general openness to engagement with religious others rather than with Muslims specifically. Despite the general consistency between openness to engagement with dissimilar others in general and openness to engagement with specific groups (Brewer & Pierce, 2005), future research should assess engagement with Muslims specifically to see how this behavioral component of attitudes relates to the affective and cognitive aspects. With regard to the Religious Diversity Scale used in this study, it is also not possible to know to what degree these self-reported behaviors reflect true behavior versus social desirability bias for some individuals in this sample. Future research should include more implicit measures of attitudes as well as direct behavioral observations/interviews.
Shortening the full Dogmatism Scale for this study could also have compromised the construct validity of the scale. Although it had good internal consistency, it may not have been sensitive enough to detect Muslim-specific attitudes. The full 20-item Dogmatism Scale (Altemeyer, 1996) may provide a more valid assessment of the relationship of general dogmatism with the affective and behavioral components of attitudes toward American Muslims. Furthermore, it could be helpful for future studies to use religious dogmatism-specific scales to measure Christians’ attitudes toward Muslims.
Limitations to the generalizability of the results
In this study, there were insufficient numbers of participants from any of the denominations to analyze between-group differences. As a result, potentially significant results both within as well as between groups may have been missed. For instance, the overall results may obscure the effect of physical disgust on attitudes toward Muslims among specific Christian groups as compared to the collective Christian group. Further research should include sufficient sample sizes of denominational groups, including Orthodox Christians, so that both within- and between-group comparisons can be made.
Another limitation to the generalizability of the results is that the sample was comprised of Christians within the United States. It is thus not clear whether the ABC model of attitudes studied here is equally applicable to the attitudes toward Muslims of Christians across the world. Future research could assess the degree to which the same three components of attitudes toward Muslims are also predictive for non-American Christians. In addition, the hierarchical regression analysis conducted in this study does not consider other important factors that may relate to attitudes toward Muslims. Future research could consider additional factors such as environmental conditions that may mediate the role of the affective, behavioral, and cognitive aspects of attitudes toward Muslims.
The cross-sectional design of this study also does not allow for examining what happens to the relationship among affect, behavior, and cognition in relation to Christians’ attitudes toward Muslims over time. Thus, it is not clear to what extent the current results regarding the interplay of the various evaluative facets of attitudes can be generalized across Christians as they move along their journey in relating to Muslims. Future longitudinal research would be very helpful in this regard.
Notwithstanding the limitations of the current study, the results suggest that attitudes toward Muslims are a multifaceted phenomenon. Specifically, the results are consistent with Eagly and Chaiken’s (2007) conceptualization of the ABC theory regarding the multifaceted evaluative nature of attitudes. In this study, physical disgust as an affective component did not add significantly to the prediction of attitudes toward Muslims in America beyond the cognitive and behavioral measures. Yet, the results of the present study, in conjunction with the existing research regarding the power of other affective components such as fear (e.g., Dunwoody & McFarland, 2018) seem to corroborate the ABC model of attitudes.
Future directions
The multifaceted aspect of attitudes has significant implications for the church in the United States and American society in general. As we continue to address negative attitudes and work toward reconciliation among groups, the ABC model of attitudes alerts us to not assume a one- or two-dimensional approach to reconciliation efforts between American Christians and Muslims in the United States. Closer examination of the specific needs of a group with regard to the various aspects of attitudes would be more productive in helping to facilitate positive attitudes toward Muslims. For example, efforts that emphasize only behavior (actions) and cognition (beliefs) toward Muslims may omit the central concern for a specific Christian group whose primary difficulty may be the affective component (e.g., fear or disgust of Muslims). Reconciliation efforts that focus on one or two of the facets or do not recognize the comprehensive nature of attitudes may unwittingly impede progress overall.
Christian church communities often play an important role in modifying a congregation’s multifaceted attitudes toward overall social justice (Todd & Allen, 2011) and Muslims specifically. Research by Houston-Kolnik and Todd (2013) is relevant to how small groups and adult education classes within the church could help challenge false beliefs (cognitions) regarding Muslims as well as increase participation (behavior) in social justice initiatives in the community. Regarding the affective component of attitudes toward Muslims, church leaders may play an especially important role. As church leaders establish the values and mission for their church, they can encourage more positive feelings (e.g., compassion) toward Muslims based on the benevolence of a loving God (Miller & Polson, 2019). Indeed, the Biblical command to love one’s neighbor (including Muslims) encompasses all the facets of attitudes.
Church reconciliation leaders should consider which aspects of attitudes are most relevant for various members or groups within their congregation. A Christian could, for example, feel fright or disgust when seeing a Muslim person. This Christian could then deliberately engage more positive cognitions regarding respect for all humanity. Subsequently, they could exhibit positive behavior toward the Muslim person. In this example, positive behavior and cognitions can occur despite the negative affective response to the other and may actually begin to alter that initial negative affect. In addition, collaboration with community agencies that have expertise in Christian–Muslim relations can help modify attitudes (Todd, 2017).
Attention to the ABC’s of attitudes can help foster more positive attitudes toward American Muslims not only among Christians but also in the broader American society. Attempts to improve intergroup relations should incorporate the reality that attitudes are the result of affective, behavioral, and cognitive factors that often are present regardless of one’s religious affiliation. Rather than focus on Christian faith as a causal agent in anti-Muslim attitudes, it may be more effective to combat anti-Muslim prejudice among everyone by fostering open-mindedness, compassion, and interaction with Muslim individuals as fellow members in the community.
Footnotes
Acknowledgements
The author wishes to thank John W. Morehead, of the Foundation for Religious Diplomacy, for the original idea for this study, and Kathryn Johnson, PhD, associate research professor of psychology at Arizona State University, for collecting the data and providing feedback on an earlier version of this article. The author also wishes to thank Anna Edwards, psychology student at Bethel University, for help with the original data cleaning and analysis.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by a grant from the Louisville Institute for the Multifaith Matters Collaborative Inquiry Team, under the leadership of John W. Morehead.
