Mapping American Attitudes Towards Refugees and Immigrants: Insights into Anti-Refugee and Anti-Muslim Sentiments

Abstract

The number of refugees across the world is increasing rapidly, as is the prevalence of Islamophobia. This flow of people and changing perceptions of them usually result in negative attitudes and hostility toward Muslims and refugees since they are perceived as the “outgroup”. This globally prevalent trend is usually attributable especially to refugees being perceived as a social, economic, or security threat. The goal of this paper is to understand Americans’ perception of outgroups and how they are related to anti-refugee and anti-Muslim sentiment by using a data-driven approach. We used latent class analysis (LCA) to determine whether there were any latent classes concerning outgroup attitudes in the US. Our results showed that people fall into four different categories regarding how they perceive Muslims, refugees, and minorities. At the same time, there were five different latent classes regarding how they perceive immigrants. This paper aims to explore this complex issue and help to reduce prejudice and intergroup conflict, ameliorate negative attitudes, and provide these groups with a stable social support system.

1. Introduction

The refugee crisis is a global issue that many countries are facing. Most refugees originate from and hosted by Muslim-majority countries (Darussalam et al. 2021). Yet, the number of Muslim refugees accepted to the United States has decreased over time, especially since the Syrian Civil War started in 2011. According to the PEW Research Center (2020), the US has admitted more Christian refugees than Muslim refugees in recent years. Their results showed that 79% of refugees in the country were Christians. In contrast, in 2016, 46% of refugees accepted to the US were Muslims. Furthermore, according to Amos (2016), there was a 98% decrease in the number of Syrian refugees admitted to the US between January and April 2018. Thus, in the present study, we sought to understand the attitudes of US voters toward Syrian and Muslim refugees, as well as towards other outgroups—immigrants and people of color. In a 2018 Pew survey, almost half of Americans (43%) stated that the US is not responsible for accepting refugees. However, it is difficult to generalize this finding to the country because the numbers fluctuate when political opinions are considered. The same study showed that 74% of Democrats thought that the US had a responsibility to accept refugees (Krogstad 2020). These differences imply that many factors affect how people perceive outgroup members.

In light of the Pew Research Center findings, the present study aims to examine how outgroup members are regarded in the US in three different outgroups: refugees and Muslims, immigrants, and people of color. The present study is critical because its data were collected right before and after the 2016 elections, which mark the start of a sharp reversal in the immigrant and refugee policies in the US. This remains relevant and timely since the Trump administration is back in office after a four-year-long gap. This timeframe is critical because it represents the beginning of crucial policy changes. There were changes in the form of stricter immigration laws, political polarization, and shifts in public attitudes toward minorities. While analyzing current trends with new datasets gives valuable insights, this dataset may serve as an essential reference point to better understand shifts in the context of US policy.

Thus, we aim to examine differences among Americans’ perceptions of different outgroups. Our primary purpose is to understand anti-refugee and, relatedly, anti-Muslim sentiment. However, most researchers approach this situation as if attitudes toward outgroup fall under only two categories: for outgroup members or against outgroup members. In contrast, we are interested in exploring the possibility that these attitudes are not binary (“for” versus “against”) and that they are, in fact, much more nuanced and varied. Therefore, we expanded our investigation to examine attitudes toward immigrants and people of color so that we can understand different latent classes among Americans and work with them on this issue accordingly.

To achieve this goal, we used a data-driven approach, latent class analysis, to uncover latent classes and test the heterogeneity of groups. Usually, ingroup members are perceived as unique individuals, and outgroup members are viewed as more homogenous. However, the outgroup homogeneity effect also extends to viewing different minority groups as more similar (Tepper and Gilovich 2024). This effect also causes people to attribute more stereotypic traits to outgroup members than ingroup members. We believe that perceptions surrounding outgroup members are more complex than simply being high or low on xenophobia, the “attitudinal orientation of hostility against non-natives in a given population” (UN Office of the High Commissioner for Human Rights (OHCHR) and Discrimination 2005) on the one hand and allophilia, positive attitudes toward outgroup members (Pittinsky et al. 2011), on the other.

Therefore, latent class analysis (LCA) was used in the current study because of its ability to identify unobserved subgroups within the population. Since many different factors can influence perceptions of refugees, Muslims, and immigrants, traditional statistical methods might not fully capture the complexity and heterogeneity of these attitudes. LCA provides a more nuanced understanding of how different segments of the American population perceive these outgroups. By exploring these latent classes, this study gives a better understanding of the attitudinal landscape in the US. It also highlights where policy and social support efforts can be most effectively focused.

1.1. Anti-Immigrant and Anti-Muslim Sentiment

Anti-Muslim sentiment in the US has been increasing, especially since 9/11. It is more complicated than simply disliking Muslims, as it is about linking Muslims to violence and terrorism (Ciftci 2012). Ciftci (2012) examined the determinants of anti-Muslim sentiment in the West and attitudes that link Muslims to terrorism and violence. His study showed that perceived realistic and symbolic threat is the most significant source of Islamophobic attitudes in countries like the US. His results demonstrated that citizens of those countries are more likely to link Muslims to terrorism if they feel threatened by their existence, either physically or culturally.

Muslims in the US are members of an outgroup of general American society ethnically, culturally, and religiously. Ciftci (2012) found that individuals who were higher on anti-Muslim attitudes felt that Muslims symbolically or generally threatened their culture or well-being. These results are significant, as most Syrian refugees identify themselves as Muslims, which might leave this population more vulnerable to societal pressure in the US. It is important to better understand anti-Muslim sentiment to better serve the needs of Muslim refugees in the US.

According to a mixed-methods study conducted by Utržan et al. (2019), refugee resettlement organizations that surveyed and resettled half of all Syrian refugees in the US between 2013 and 2016 and were expected to resettle more Syrian refugees in 2016 than in the previous year (i.e., 2015). After experiencing war trauma, having to go through all stages of the resettlement programs, and being accepted to be resettled in the US, Muslim refugees also face increasing prejudice and discrimination in their new homes. Utržan et al. (2019) noted how misinformation plays a role in increasing prejudice and discrimination against Syrian refugees in the country. This not only leads to anti-refugee sentiment but also increases fear. This important finding shows us the importance of public education and awareness.

1.2. Xenophobia and Intergroup Conflict

Many disciplines, including social psychology, have focused on solving problems related to xenophobia and intergroup conflict from different perspectives (Suleman et al. 2018). Even though they approach the issue from different perspectives, most researchers agree that some problematic views about outgroup members are related either to lack of information or to misinformation (Genschow et al. 2022). This is also observed for refugees from different parts of the world.

Most refugees are perceived as coming from lower socio-economic backgrounds with lower education levels (Atrooz et al. 2024). They are also perceived as an economic threat, as they enter the job market in their new destination (Riek et al. 2006). In contrast with people’s assumptions, most refugees in the US come from high-class and well-educated backgrounds, as they can afford the smugglers’ cost to be transported to Western countries such as Greece (Zunes 2017). Immigrants and refugees usually become productive citizens, start businesses, and put more money into the system than they received (Zunes 2017). Moreover, refugees who became permanent immigrants in North America and Europe are less likely than members of the native population to commit a crime (Abolarin et al. 2023). For example, the only Syrian asylum-seeker to have been arrested for terrorism was denounced to German authorities by other Syrian refugees (Zunes 2017).

Nonetheless, media outlets might contribute to the paranoia about Muslims being terrorists, and cognitive images of certain groups change behaviors toward them. Hence, we must focus on those stereotypical images constructed in people’s minds. However, it is not enough to focus on eliminating the negative stereotypes. We also must focus on the positive spectrum (i.e., allophilia) with regard to attitudes toward outgroup members.

1.3. Allophilia

We are also interested in the positive attitudes of US voters towards outgroup members. There is a bias in the literature on intergroup relations, as most researchers pay more attention to the negative spectrum of attitudes than to the positive spectrum. However, Pittinsky et al. (2011) found that positive attitudes toward minorities predict positive behaviors and vice versa. Thus, it is important to investigate the positive spectrum to improve intergroup relations and promote positive attitudes. It is critical to focus on promoting these positive attitudes, as they lead to positive behaviors toward that particular group and reduce negative attitudes. According to the psychological theory of valence congruence, positive behaviors were predicted better by how much people like members of outgroups than by how little they dislike them (Pittinsky et al. 2011).

Terms such as “tolerance”, “acceptance”, and “respect” are used widely in the literature, but these do not fully cover truly positive attitudes toward specific outgroup members. Still, efforts to promote tolerance are the backbone of work by many leaders who aim to resolve intergroup conflict. This is essential but not enough, as not hating the outgroup is not the ultimate goal; liking them is. Consequently, Pittinksy et al. (2011) created a validated scale to measure allophilia. It includes factors such as having positive feelings toward the other group, feeling comfortable around them, believing that one has a close connection with them, seeking interaction with them, and feeling impressed and inspired by them. As is apparent from these five factors, terms such as “tolerance” or “acceptance” are not sufficient to describe allophilia.

In addition, emotions motivate certain behaviors, such as increased generosity and helpfulness (Pittinsky and Montoya 2009). It is crucial to note that allophilia is directed toward a specific group; we used it because each of our items focuses on a specific group, not on perceptions of outgroups in general. Therefore, we used allophilia as a label in our analysis to investigate positive attitudes toward specific groups in the US. We specifically wanted to use this term as a label because some of our items are about positive attitudes towards outgroups, such as “Muslims are patriotic”, “Immigrants are generally good for America’s economy”, and “Black people are hardworking”.

A study conducted by Brown (2015) recruited students who took service-learning courses working with elders with dementia and/or physical limitations and used this scale. Even though students initially expressed negative emotions such as fear, they exhibited all components of the allophilia framework: affection, engagement, kinship, creativity, enthusiasm, and comfort. At the end of the course, participants mentioned positive emotions like empathy. This is an evidence-based approach to allophilia and how it influences people’s learning and behavior. It shows us that positive attitudes lead to positive behaviors (Fredrickson 1998). Consequently, a change from negative to positive attitudes will give us better results and provide a holistic perspective on the problem. This is especially important because positive and negative attitudes are not exactly opposites and should be examined individually. Also, research has demonstrated that positive attitudes foster proactive and prosocial behaviors (Pittinsky et al. 2011). This shows us how important it is to reduce negative attitudes and promote positive ones.

2. Method

Data for this paper were derived from the American National Election Studies (2016) dataset, which is publicly available on ICPSR. ANES has been collecting data on Americans’ social backgrounds, political dispositions, values, perceptions, evaluations, and opinions about groups, candidates, and ongoing discussions during the years of presidential elections since 1948. The 2016 ANES study had a total sample size of 4271 interviews, 1181 of which were gathered through traditional face-to-face interviewing by trained interviewers using professional interviewing software, while the rest were conducted on the internet (Allen-West et al. 2017). For some sensitive questions on the survey, ANES provided participants with a tablet to minimize the social desirability bias that could result from the direct interview of participants by the research team. Lastly, the following demographic variables were recorded: respondent age, education level, political affiliation, race/ethnicity, marital status, and family composition. Based upon 4150 valid cases out of 4271 total cases, the mean age of participants was 49.58 (range = 18–90, SD = 17.58).

We used the items below to investigate latent classes. We identified 18 items that met our criteria and grouped them under three categories: immigrants, minorities, and others. We then created models based on those variables (

Table 1. Items.

Refugees-Muslims
V161214X	Allow Syrian Refugees
V162295X	Favor/oppose US government torture of suspected terrorists
V162355	Muslims are patriotic
V162353	Muslims are violent
Immigrants
V161196X	Build wall with Mexico
V161192	Policy toward unauthorized immigrants
V161195	Children brought illegally should be sent back
V162157	What should immigration levels be
V162270	Immigrants increase crime rates in the US
V162268	Immigrants are generally good for America’s economy
V162158	How likely immigrants will take away jobs
V162269	America’s culture is generally harmed by immigrants
Minority
V162266	Minorities should adapt to customs and traditions of US
V162317	Whites unable to find job because employers hire minorities
V161197	How important to speak English in the US
V162318	Federal government treats blacks or whites better
V162350	Black people are violent
V162346	Black people are hardworking

).

3. Analytic Approach

Firstly, we selected specific variables we were interested in and determined each item’s kurtosis, skewness, and other descriptive statistics (i.e., mean, SD, max, min, and range). The variables were chosen for their capacity to indicate attitudes towards and stereotypes about other groups.

Generally speaking, stereotypes are characterized as beliefs or generalizations about social groups, and prejudice can be defined as negative responses to members of those groups (Madva and Brownstein 2018). According to Lee and Fiske (2006), research shows that stereotypes portray immigrants as incompetent and untrustworthy. According to Fiske’s Stereotype Content Model (Fiske et al. 2002), the dimensions of perception are competence and warmth, the dynamics of which are related to the immigrants’ perceived status and the sense of competition within society. However, research has shown that immigrants’ national origin influences the perceptions of majority members, as each nation has unique circumstances and features (Lee and Fiske 2006). We included items related to different immigrant and refugee populations to elucidate these differences and heterogeneity in people’s perceptions of them. Those items included topics such as allowing Syrian refugees, building a wall with Mexico, policies toward unauthorized immigrants, children who have been brought to the country illegally, and how immigrants, in general, influence America’s economy.

In addition, we included items about different minority groups (i.e., black people) in the US; for instance, African immigrants are not perceived in the same way as African Americans are (Lee and Fiske 2006). Thus, items such as “black people are violent” or “black people are hardworking” would be essential to distinguish African immigrants from African Americans, as these are generally stereotypes that are used to stigmatize African Americans. This variation within and between the groups aligns well with our idea that there is significant heterogeneity in outgroup perceptions in the United States. Results found by Lee and Fiske (2006) support this idea, as they found that most immigrant groups were viewed through ambivalent stereotypes about their implied socioeconomic status or nationality. According to the SCM, the stereotypical image of an immigrant would be of a low-competence, warm person. However, once more details and information are given, heterogeneity is observed.

Furthermore, according to Berry (1984), most members’ preference for acculturation style (integration, separation, assimilation, and marginalization) depends on whether the group is valued or not. Therefore, we included items such as “Immigrants generally harm America’s culture”, “minorities should adapt to customs and traditions of the US”, and “how important it is to speak English in the US” to distinguish between the acculturation styles.

After selecting 18 variables based on the stereotype literature, we categorized them and reviewed the frequencies of each response category. Variables with fewer than 5 response categories were treated as categorical. Variables with 5 or more response categories and skewness/kurtosis within the acceptable range (slight deviations from 0) were treated as continuous.

If the skewness or kurtosis was extreme, the variables were treated as categorical and we recoded them into fewer response categories. We combined their responses so that all items in each set had the same number of responses for ease of interpretation. By doing so, we aimed to reduce the options to a scale from 1 to 3. For categorical variables in latent class analysis, we estimated thresholds. We combined responses to estimate fewer thresholds so that that model estimation would be more straightforward and model non-convergence might be avoided. In addition, we wanted to make the interpretation and presentation of results easier by using fewer thresholds.

Lastly, we identified continuous variables with problematic and extreme skewness/kurtosis and recoded them into fewer response categories. We recorded the items as 1, indicating more negative attitudes toward outgroups, and 3, meaning more positive attitudes toward outgroups (1 = xenophobia, 3 = allophilia). Other continuous items were not recoded into fewer response categories.

After cleaning the dataset and fully recoding items, we ran the analysis, choosing models with the lowest BIC scores and creating graphs for each model to present response probabilities.

4. Results

We used 18 items to determine class membership in three different domains. We then provided descriptive labels for different classes of xenophobia and allophilia. It is important to note that the classes in each category do not contain the same participants. For instance, Class 1 in the Immigrants category differs from Class 1 in the Minority category.

We labeled and ordered the groups from the one highest in xenophobia to the one highest in allophilia (negative attitudes toward outgroups—positive attitudes toward outgroups). Other groups in between are labeled as showing moderate xenophobia, moderate allophilia, or neutral. Although there is a greater focus on negative attitudes toward outgroups and the aim of reducing prejudice, it is important to focus on positive attitudes toward outgroups (allophilia) and seek ways to increase them (Pittinsky et al. 2011).

Table 2. Best-fit models.

Model	Number of Classes	BIC	AIC	Shares
A4 (Refugee/Muslim)	4	BIC(4): 27,045.26	26,720.91	0.0878 0.2729 0.4426 0.1968
B5 (Immigrants)	5	BIC(5): 52,206.94	51,386.52	0.1171 0.2265 0.1857 0.3793 0.0915
C4 (Minority)	4	BIC(4): 35,890.76	35,388.33	0.3088 0.1337 0.4956 0.0619

describes the best-fit models for each category.

We ran a latent class analysis in Rstudio 4.0.4 with different numbers of classes and aimed to choose the best fit to the data relative to the model with one class fewer. A lower Bayesian Information Criterion (BIC) score indicates improved model fit and is the primary indicator for deciding the number of classes (Nylund et al. 2007). We selected four-class and five-class models as our final models because they had the lowest BIC scores compared to others in their categories (

Table 3. Population shares and membership of chosen models.

Model	Estimated Class Population Shares	Predicted Class Memberships (by Modal Posterior Prob.)
A4	0.0821 0.2087 0.4369 0.2724	0.0623 0.1653 0.4843 0.2882
B5	0.1734 0.1605 0.2849 0.2469 0.1343	0.1543 0.1545 0.3123 0.2505 0.1285
C4	0.3085 0.4969 0.1302 0.0645	0.291 0.5517 0.1025 0.0548

).

Generally, attitudes towards outgroups can be described as positive or negative. However, when we look at a set of items, attitudes are more complex than “pro-outgroup” or “anti-outgroup”. For example, we might find pro-outgroup responses for most items but anti-outgroup responses for others; responses are not always all “anti” or all “pro” across all items. Attitudes are likely more complex than being pro- or anti-outgroup. To uncover potential gradations and nuances, we used LCA, a data-driven approach, to test the heterogeneity of groups and how they go beyond what we observed as we uncovered latent classes after the analysis. Based on this analysis, we examined the emerging profiles to see if they look similar within and between categories. Another strength of this approach is that it is a model-based and probabilistic approach, which makes it easier to replicate.

Refugees and Muslims

The four-class model had the lowest BIC value (27,045.26) among all class solutions and was therefore considered for selection. We had four items in the Refugees and Muslims category, and we highlighted results that stood out for their high probabilities. As lower numbers on the scale meant a more favorable attitude toward outgroups and respondents in Class 4 had a higher probability of responding with 1 and 2, the results show that this class had the most negative perceptions of Muslims and refugees. Respondents in Class 1 had high probabilities of selecting responses 2 and 3, so they were the class with the most favorable views. Thus, Class 4 was the class that had the greatest tendency toward xenophobia against Muslims and refugees in the US and Class 1 was the highest in allophilia.

The profile plot for the four-class model is displayed in Figure 1. This figure represents the probability of endorsing response 3, which indicates the least favorable attitude toward outgroups. Class 1 primarily included participants with the lowest probability of perceiving Muslims as patriotic and non-violent. Class 1 had the highest probability of endorsing response 3 on those items when compared to the other three classes, which are similar to each other on those two items. Thus, Class 1 was the highest in xenophobia and was labeled accordingly. We saw some discrepancies across classes for the item related to allowing Syrian refugees into the country. However, overall, Class 3 and Class 4 had very similar profiles. Class 4 was characterized by having a high probability of accepting Syrian refugees and seeing Muslims as patriotic or non-violent. That is why this class was labeled as being the highest in allophilia. Also, Class 2 was found to be moderate in xenophobia, whereas Class 3 was moderate in allophilia. The populations shares can be found in Figure 2.

• Immigrants

Given that we had eight items under the immigrant category, we determined individual percentages for each item. We focused on the probabilities associated with each item and checked where the highest probability fell. Overall, Class 2 was the most welcoming towards immigrants (highest in allophilia) and Class 1 was identified as the class highest in xenophobia. The five-class model displayed the lowest BIC value (52,206.94) among all class solutions and was therefore considered for selection (see Figure 3).

The profile plot for the five-class model shows that classes have relatively similar patterns compared to the model for the Refugee−Muslim category. Interestingly, especially for the item about whether children brought illegally should be sent back, respondents in all classes were most likely to select response 3, which could be considered to endorse not sending them back. Furthermore, we see that all classes thought that immigrants were good for the economy but were also taking away jobs in the country. Thus, we can infer that informing people about the benefits of immigrants to America’s economy is not enough.

Class 2 included participants with the highest probability of accepting immigrants and being more welcoming toward them (Figure 3). Thus, they were the highest in allophilia and were labeled accordingly. Class 1 was characterized by having the lowest probability of accepting immigrants and welcoming them and the highest probability of thinking they bring more harm than good to society. That is why this class was labeled as being the highest in xenophobia.

B.

Minority

The results for the minority category were straightforward. Our results indicated that Class 1 had the most favorable attitude towards minorities and the highest allophilia score, while Class 4 had the least favorable approach to them and the highest xenophobia score. The four-class model yielded the lowest BIC value (35,890.76) among all class solutions and was, therefore, considered for selection.

The profile plot for the five-class model showed that Class 1 primarily included individuals with the highest probabilities of allophilia towards minority members in the US. Thus, they had the highest tolerance and empathy and were labeled accordingly. Class 4 was characterized by having an extremely low probability of accepting minority members as equals and a high probability of believing they should learn the language and adapt to the customs. That is why this class was labeled as being the highest in xenophobia. It was also interesting to see that the item about the importance of speaking English in the US could not be used to distinguish between any of the classes. On the other hand, the items about Black people being non-violent or hardworking distinguished Class 4, the class highest in xenophobia, from the others (Figure 4).

5. Discussion

Our results suggest that attitudes toward refugees, immigrants, and other outgroup members in the US are more complex than simply being pro- or anti-outgroup. Using a data-driven approach, we shed light on potential latent classes, identified the classes that emerged, and observed whether they looked similar. We identified and addressed a gap within the literature, as not many latent class analyses have been conducted as exploratory studies to understand perceptions of outgroup members in the US. This is a good starting point for researchers, as this study aimed to increase understanding of the heterogeneity of outgroup-related perceptions, especially of perceptions of refugees, using a latent class analysis approach.

Our findings are significant because they also challenge the idea of the “out-group homogeneity effect”. Judd and Park (1988) described the different conceptualizations of ingroup and outgroup members held by adults by using this term from the field of social psychology. They suggested that ingroup members are considered unique individuals, while outgroup members are conceived of as more homogenous. This difference also leads people to attribute more stereotypic traits to outgroup members than to compared to ingroup members. Thus, increasing perceived group variability might reduce prejudice and discrimination, even as intergroup biases generate them (Shilo et al. 2019). Prior research has also indicated that perceived outgroup homogeneity increases ingroup favoritism over the outgroup and increases depersonalization, leading to even more stereotyping (Simon et al. 1990). Therefore, by identifying those subgroups and going beyond a narrow focus, our contribution was to disentangle the heterogeneity within the US and explain those differences. Also, we provided projections for future studies, as the election in 2016 was a milestone for many minority members in the US.

Our findings indicate that there are identifiable subgroups in the US in terms of how they perceive outgroup members, such as immigrants and refugees. Overall, results indicated that there are gradations of attitudes toward outgroup members in the US, and things are not black-and-white as it is mainly assumed. A study by Yalcinkaya et al. (2018) showed that public support for accepting refugees into Western countries depends on perceived cultural malleability (the possibility of adaptation and cultural change). For instance, the perception that child refugees have more potential to adapt predicted support for policies in favor of accepting refugees. This study was replicated with German and American samples. It also aligns well with our idea about the heterogeneity of opinions about outgroup members such as immigrants and refugees. However, the situation became more complex after the 2016 elections due to the political shift in the country.

The political climate in the country changed abruptly after the 2016 elections. For instance, a new Public Charge rule proposed by the Trump administration on 15 October 2019 aims at preventing legal immigrants who use government benefits they are legally entitled to, such as food assistance or services such as Medicaid, from receiving green cards/permanent residency (Khafagy 2019). As a result, lower-income migrants would have more difficulty obtaining US citizenship. A draft of the proposal was first published in September 2019 and was widely criticized. The attacks on immigration, which also include the goal of redefining asylum, might have become a stress factor for immigrants. Even though refugees and asylum seekers were exempted from the above rule, there are several similar legal actions, such as Executive Order 13769 (Protecting the Nation from Foreign Terrorist Entry into the United States, also known as the Muslim ban/travel ban), and these might cause Muslim refugees especially to feel overwhelmed. One of the countries whose citizens were banned from entering the US was Syria, which has the highest number of refugees in the world, as we discussed earlier. Furthermore, it is critical to note that on 26 September 2019, the White House announced that 2020 refugee admissions to the country would be limited to 18,000 (Hartig 2020). Furthermore, Executive Order 13888 (Enhancing State and Local Involvement in Refugee Resettlement) gave state and local authorities the power to deny refugee resettlement in their area.

After Donald Trump’s presidency, Joe Biden won the 2020 election, and there was a shift in immigration policies as Biden began to reverse many of Trump’s policies. In 2022, Trump announced his candidacy for the 2024 presidential election, and after winning the election, he began his second term in January 2025. His administration announced its intent to resume border wall construction and reinstate strict asylum policies.

Our results indicate that perceptions of outgroup members in the US (i.e., minorities, immigrants, and refugees) are not black and white. There are gray areas that need to be explored by researchers, and policy-makers need to consider them. For instance, for groups of people such as minorities, refugees, and Muslims, we found four different categories of attitudes (high allophilia, moderate allophilia, moderate xenophobia, and high xenophobia).

For immigrants, we found five distinct latent classes of attitudes (high allophilia, moderate allophilia, neutral, moderate xenophobia, and high xenophobia). This result demonstrates that attitude towards outgroups is not simply about accepting or rejecting outgroup members. There are at least four or five different classes in the US with regard to their perceptions of minorities, immigrants, and refugees in the country. This is a significant finding, as the people in two or three of these latent classes are more likely than those in the highly xenophobic class to change their opinions. This is possible with education or other tools, such as intergroup contact.

6. Limitations and Future Directions

Our framework for understanding perceptions of outgroup members focuses on motivations behind attitudes toward minorities, refugees, and immigrants. Attitudes toward these groups are motivated by several factors, such as security or economic concerns. That is why we also included the item about terrorism and how it relates to the stereotypical image of a “Muslim”; the item was intended to help us understand the role of security concerns in the acceptance of Syrian refugees by the public.

Future research should also include items about employment to see how economic concerns play a role in people’s perceptions. One reason to examine economic concerns is that individuals who perceive immigrants as competing for economic resources (e.g., “They are stealing our jobs”) are more likely to be motivated to remove this competition. This idea aligns well with the Realistic Conflict Theory. It suggests that when the resources are limited, competition leads people to discriminate against outgroup members and increases anti-refugee sentiment. According to Mangum (2019), Americans tend not to be concerned about immigrants’ cultural impact but clearly perceive them as an economic threat. When they examined different racial groups, Mangum (2019) found that White Americans have both cultural and economic concerns about immigrants.

These perceptions and policy changes mentioned above also influence public opinion, which could be a future area in which researchers could observe differences. The results might differ between this study, conducted with the ANES 2016 dataset, and a study using the ANES 2020 and 2024 data, which will be available after the elections. We selected ANES 2016 to explore the immediate shift in societal dynamics.

Another point worthy of attention is the biased sampling of the ANES 2016 study. It is difficult to say that the ANES 2016 dataset represents Americans well, as 71.1% of participants were white, 9.3% were black, and 10.5% were Hispanic. Hence, it is not a nationally representative sample, and it is difficult to generalize the findings to the American public.

7. Implications for Researchers

The most important implication of this study was that we cannot categorize people as simply discriminatory or non-discriminatory. Our findings reveal that there are at least four and/or five groups of people, or patterns of variables, in terms of perceptions of outgroup members. Therefore, researchers and policy-makers should be aware of this spectrum to better serve people in the US.

As previously stated, competition for economic resources is a strong motivator for people to remove their competition. It is likely to influence people’s voting behavior, resulting in more votes for the candidates that promise to decrease the competitiveness of the outgroup (Wright and Esses 2019). Consequently, addressing people’s economic, security, and cultural concerns about groups and accurately informing them using an evidence-based approach is necessary. A high level of education is found to be a predictor of reduced negative sentiment toward Muslims (Ciftci 2012).

Islamophobia has reached unparalleled levels in the United States. The latent class analysis in this article determined that feelings toward outgroup members are more multifaceted than we think. There is significant variation in people’s attitudes, which are shaped by different factors, such as perceiving refugees as a threat or the discourse that right-wing politicians have been using to feed xenophobia in the country. Hence, increasing knowledge of Islam and fact-checking are essential actions that can be used to reduce misperceptions and misinformation about Muslims (Ciftci 2012).