Perceived Group Deprivation and Intergroup Solidarity: Muslims’ Attitudes towards Other Minorities in the United States

: What is the relationship between the sense of perceived discrimination among members of a marginalized racial, ethnic, or religious group and their political attitudes towards other marginalized groups within their society? Research on minority groups in politics has established that the feeling that one’s own group is socially deprived and discriminated against is generally associated with an increase in within-group solidarity, observable in members’ stronger expressions of collective identity—also called “group consciousness” or “linked fate”—as well as their robust support for political parties and policies seen as directly beneﬁtting members of their in group. Yet an underappreciated strand of this same research suggests that a strong sense of in-group deprivation may also lead to greater empathy and political support for other marginalized minorities, a phenomenon we refer to as intergroup solidarity . In this paper, we use the case of Muslim Americans to test the hypothesis that perceptions of group deprivation can lead to increased intergroup solidarity with other socially marginalized racial, ethnic, and religious minorities. We ﬁnd that Muslims who feel that they have been discriminated against and / or who believe Muslims as a group are a target of discrimination are more likely to embrace the struggles of other groups and recognize the marginalization of other groups. Our ﬁndings suggest that in-group political consciousness raising may be a ﬁrst step toward intergroup coalition building among those who su ﬀ er from discrimination and marginalization.


Introduction
Political scientists and sociologists studying racial and ethnic minorities in American political life have repeatedly found links between a strong sense of group deprivation or disadvantage and a higher level of solidarity and political commitment among members of the in group (Dawson 1994;Espiritu 1992;Garcia Bedolla 2005;Gurin et al. 1980;Harris 1999;Masuoka 2006;Miller et al. 1981;Sanchez 2006;Tate 1993;Verba and Nie 1972). This strain of the literature on minority group politics has generally focused on two outcomes consistently predicted by perceived group deprivation and indicating a strong sense of in-group solidarity: increased feelings of shared group identity-also referred to as "group consciousness" (Verba and Nie 1972) or "linked fate" (Dawson 1994)-and political support for in-group-focused policies, parties, and candidates (Barreto and Woods 2005;Dawson 1994;Espiritu 1992;Padilla 1985;Tate 1993). As such, the extant literature on the politics of African Americans, Latinx Americans, Asian Americans, and, more recently, American Muslims primarily tells the story of how members of these groups who perceive themselves as members of a deprived minority are more likely to articulate a greater sense of political and social solidarity focused on their own in group. Many of these same studies, however, include traces of evidence suggesting the possibility of a link between perceived group discrimination within these minorities and an emergent sense of positive social sentiment and supportive political attitudes extended towards other socially marginalized groups. A brief review of this literature will both demonstrate the extent to which previous studies have focused on the effect of perceived discrimination on increasing in-group solidarity while also highlighting how this same research suggests the possibility that group deprivation may also drive a stronger sense of intergroup solidarity, or empathy and support for the struggles of other social minorities.

African Americans, Perceived Group Deprivation, and Resulting Solidarities
African Americans are by far the most extensively researched racial group on questions of perceived group deprivation and its connection to greater levels of in-group solidarity (Dawson 1994;Gurin et al. 1980;Harris 1999;Jackman and Jackman 1973;Miller et al. 1981;Tate 1993;Verba and Nie 1972). In fact, it was through seminal studies of Black politics in the US that political scientists and sociologists established the importance of perceived group deprivation as a variable within objectively marginalized ethnic and racial groups in American society, and one that could therefore be used to examine the links between the sense of group deprivation among specific members of racial and ethnic communities and their political attitudes and behavior (Garcia Bedolla 2005;McClain and Stewart 2006;Miller et al. 1981;Tate 1993). As mentioned above, apart from political participation (Leighley and Vedlitz 1999;Miller et al. 1981;Verba and Nie 1972), the two outcomes of greatest interest to those studying the effects of perceived group deprivation on African Americans have been measures of solidarity focused on the in group-greater feelings of collective identity (often called "group consciousness" or "linked fate") on the one hand, and/or a high level of support for policies or political candidates directing social aid towards African Americans on the other.
Both earlier and later studies of Black politics in the United States found a strong link between the historical and ongoing experiences of discrimination of African Americans-and specifically the strong sense of perceived group deprivation this produced among most members-and a powerful feeling of social closeness to and political unity with other Black Americans (Dawson 1994;Gurin et al. 1980;Miller et al. 1981;Tate 1993;Verba and Nie 1972). As the central figures in the later wave of these studies, Tate (1993) and Dawson (1994) went on to show that, for most Black Americans, this sense of race-linked group deprivation was also tied to a robust support by African Americans-especially those with strong feelings of group deprivation-for "race-specific programs" (Tate 1993, p. 40) and "governmental racial policies" (Dawson 1994, p. 194). As such, the central finding of those who have examined the relationship between perceived deprivation among US Blacks and political outcomes is that this sense of group marginalization among African Americans produces social and political manifestations of in-group solidarity (Tate 1993, p. 45).
Yet evidence from some of these same studies, and other adjacent ones, suggests that the same sense of shared group marginalization that drives African Americans to turn their social and political energies inward may, in some cases, push them to express a broader, outwardly focused, intergroup solidarity. For example, Gurin, Miller, and Gurin's landmark 1980 study is most often cited as evidence of a strong in-group political solidarity among African Americans. But their evidence also reveals that Black Americans who had the strongest sense of shared deprivation were not only more likely to support pro-Black policies, but were also more likely to articulate political support for ameliorating the lower-status situations of women and poor people (Gurin et al. 1980, pp. 37, 40). Similar evidence of a potential "spill-over" effect of the in-group solidarity generated from shared group deprivation is present in Tate (1993) and Dawson (1994), who both find that deprivation-informed strong racial identities are linked to support for government policies and politicians aimed at helping minority and discrimination became motivated to advocate for political and social change aimed at bettering the status of Asians in the US. Along these same lines, and edging towards the central concerns of this paper, evidence in select studies demonstrates that feelings of group deprivation among Asian Americans are connected to the desire of members of this marginalized group to find common ground with other racial minorities and support politics and policies that benefit African Americans and Latinx people (Cain et al. 1991;Espiritu 1992;Uhlaner 1991). It is this intergroup solidarity driven by experiences of discrimination targeted at one's own in group that we seek to investigate in the case of Muslim Americans.
In summary, the existing research on the relationship between perceived group deprivation and forms of solidarity among the most "racialized" (Bonilla-Silva 2001;Omi and Winant 1986) minority groups in American life-Blacks, Latinx, and Asian Americans-has found that, with some variation across groups, a strong sense of perceived group disadvantage predicts a heightened sense of in-group solidarity, expressed as intensified feelings of collective identity and greater levels of support for politics and policies aimed at addressing concerns of the particular in group. In addition, our review of this literature has revealed signs that a sense of group deprivation may also be linked to a greater sense of intergroup solidarity, feelings of concern and support for members of other socially marginalized groups. As the next section will show, a growing literature on Muslim Americans and politics reveals a link between perceived discrimination and greater in-group solidarity quite similar to the pattern observed among the more commonly studied racial groups discussed above. Like this broader literature, though, research focused on Muslim Americans politics has largely marginalized the question of whether perceived group deprivation might lead members of racialized minority groups to develop a greater sense of intergroup solidarity, measured as concern and support for marginalized, racialized out groups. After reviewing the relevant literature on Muslim American politics, we will directly take up this empirical question.

American Muslims, Perceived Group Deprivation, and Solidarity
The relationship between perceived discrimination and solidarity among Muslim Americans has mostly been investigated using qualitative and historical methods, but the results are strikingly similar to those we have described from studies of the other three US minority groups discussed above-African Americans, Latinx Americans, and Asian Americans. First, as with those groups, research on Muslim Americans demonstrates a clear link between Muslims' strong sense of perceived group deprivation and a stronger sense of collective identity. Indeed, this relationship is key to the work on "racialization" of Muslim Americans (Selod 2015;Selod 2018), which, while acknowledging variation within Muslim Americans in terms of their personal experiences of discrimination, asserts that, in general, aggregated experiences of social marginalization among Muslims has led to a stronger identification with "Muslim American" as a collective and political identity among most members of this group (Cainkar 2002;Chouhoud et al. 2019;Naber 2005). 1 Second, the growing literature on Muslim Americans reveals a link between perceived group deprivation and another dimension of in-group solidarity-an increased support for policies, parties, and politicians seen as pro-Muslim rights. For example, political scientists have used both interviews and surveys to find that increased antipathy towards Muslims domestically and abroad by Republican governments drove Muslim Americans towards stronger support for the Democratic party, who they generally understood to be more supportive of US Muslim civil rights and liberties (Calfano and Lajevardi 2019;Jalalzai 2009;1 As with African American, Latinx American, and Asian American, the category "Muslim American" encompasses a highly complex internal diversity yet has come to be seen and treated by the state, the media, and a significant portion of the American public as a singular and uniform racial group due to the social, legal, and political processes of "racialization" (Bonilla-Silva 2001;Lajevardi and Oskooii 2018;Omi and Winant 1986;Selod 2015). Also, like members of these other groups, Muslim Americans have-to varying degrees-undergone experiences of social marginalization and exclusion that have caused them to consider their own group to be a consistent target of discrimination and social deprivation (Braunstein 2019;Dahab and Omori 2019;Lajevardi and Oskooii 2018;Selod 2015).
Religions 2020, 11, 604 5 of 30 Ocampo et al. 2018). Similarly, sociologists and political scientists studying Muslim American grassroots politics and organizations show how a growing sense of group deprivation under both the George W. Bush and Donald Trump administrations led to an increased participation by Muslim Americans in activism aimed at defending Muslim American civil rights and challenging particular policies such as Trump's "Muslim ban" (Love 2017;Oskooii 2016;Tekelioglu 2019;Yukich 2018).
As with previous research on Blacks, Asians, and Latinx Americans, the existing studies on Muslim Americans and politics both contain suggestions of a link between perceived group deprivation and increased intergroup solidarity and leave this question largely unexplored. One strain of this literature highlights consistent, though anecdotal, evidence that Muslim American activists and organizations reached out to others to form political coalitions and express intergroup support during the post-9/11 and Trump years (Tekelioglu 2019). In these works, we find accounts of Muslims forging partnerships with other racial, ethic, and religious minorities (Vali 2012), Black Lives Matter activists (Gibbons et al. 2015), and civil liberties advocates and organizations (Craun 2014). Another strain notes attitudinal shifts among Muslim Americans, especially when it comes to their increasing acceptance of homosexuality (Pew Research Center 2017b;Selod 2018). But, as with research focused on Black, Asian, and Latinx American politics, none of these studies has taken on the issue of intergroup solidarity as its outcome of interest. It is this outcome-and its relationship with perceived in-group deprivation-that we directly assess in this paper, using the case of Muslim Americans.

Data
We use data from the Pew Research Center's repeated cross-sectional surveys of American Muslims, conducted in 2007Muslims, conducted in , 2011Muslims, conducted in and 2017 These data are the only known national, probability samples of this population and therefore provide our best chance at a representative sample. Center researchers stratified US counties based on the estimated density of Muslims within them and then conducted random digit dials of cell phones and landlines in all but the least dense stratum. Screening interviews were conducted until a large enough sample (over 1000 respondents) was gathered for each year (1050( Muslim adults in 2007( , 1033( in 2011( , and 1001. 3 All samples were weighted to adjust for the differing probabilities of selection given the sampling frames and non-coverage of geographic areas with a low density of Muslims. Interviews were conducted by phone in English, Arabic, Urdu and Farsi, gathering information on demographic characteristics, household composition, political opinions, and religiosity (Pew Research Center 2007, 2011, 2017b. Parts of our analysis utilize all three years of the survey (N = 3084) and others focuses on 2017 (N = 1001), when new survey questions were introduced that are particularly relevant to this project.

Analytic Strategy
Do Muslims who feel marginalized as Muslims show greater solidarity with other identity groups? To answer this question, we use two proxies for sense of deprivation: respondents' experiences of being discriminated against as a Muslim and respondents' perception of discrimination against Muslims as a group. We test the relationship between these marginalization variables and a series of questions on attitudes towards other minority groups using logistic regressions. Our outcome variables encompass respondents' views on the status of Black Americans, lesbian and gay (LG) 4 2 Data from all three survey years are publicly available on the Pew Research Center's website: www.pewresearch.org. Analysis was conducted using Stata 16, and all files associated with the project are available upon request. 3 For more details on the sampling process for each survey, see the Pew Research Center's website and reports (Pew Research Center 2007, 2011, 2017b. 4 As will be noted below, the survey questions inquire after homosexuality and the status of lesbian and gay individuals. To hue close to the data, we use the term LG instead of the broader term LGBTQIA+ when referring to our measures and results.
Religions 2020, 11, 604 6 of 30 individuals, immigrants, Hispanics 5 and Jews. We present the full tables below and then calculate and graph predicted probabilities of each outcome at different levels of our two independent variables.
In separate analyses, whose results are included in Tables A1-A3 of Appendix A, we run bivariate regressions on each pairing of independent and dependent variable. The differences between the multivariate and bivariate models were largely trivial, so we focus our discussion on the results of the multivariate regressions. Where there was significant, observable overlap between one of the groups and the Muslim population (i.e., between Blackness and being Muslim), we conducted separate analyses for each group to ensure our results still held. We will discuss the results of these sensitivity tests in more detail below.
Since all of our control variables are categorical, we create a missing category for each control to minimize loss of observations. We then use listwise deletion to remove observations that are missing on either the main independent variable or dependent variable in each model. As a result, the number of observations varies from model to model. 6 This does not threaten our results, because we do not compare coefficients across models. Next, we discuss how each variable was constructed and discuss its distribution across the sample.

Independent Variables
Our two independent variables are experiences of discrimination as a Muslim and perception of discrimination as a Muslim. These two measures tackle the question of whether Muslims feel a sense of deprivation-individually or collectively.

Experiences of Discrimination as Muslims
In 2007 and 2011, respondents were told "I am going to read a list of things that some Muslims in the US have experienced. As I read each one, please tell me whether or not it has happened to you in the past twelve months. First, in the past twelve months, [randomized items] because you are a Muslim or not". The randomized items were "have people acted as if they are suspicious of you", "has someone expressed support of you", "have you been called offensive names", "have you been singled out by airport security, have you been singled out by other law enforcement officers", 7 or "have you been physically threatened or attacked". In 2017, the same prompt and response items were listed, but interviewers added "have you seen anti-Muslim graffiti in your local community, or not".
To create a measure of discriminatory experiences, we tallied the number of negative items respondents said had happened to them (whether the respondent received support was omitted). To keep the scale consistent across all three years of the survey, we omitted the 2017 graffiti item. The resultant scale ranged from 0 to 5. Since categories 4 and 5 had very few respondents, the variable was top coded such that 3, 4 and 5 were combined into category "three or more". 8 Three percent of observations on this variable were missing. As Table 1A shows, among those observed, the mean (across 5 The survey questions involved used the term Hispanic, but we acknowledge that the word and its derivatives have a problematic and contested history (Alcoff 2005). To maintain accuracy, we use the term when directly discussing the data but use Latinx in our broader discussions (Salinas and Lozano 2019;Morales 2018). 6 To ensure that our results are not sensitive to listwise deletion of missing observations, we impute missing data using chained multivariate regressions (20 iterations) and re-run our analysis using the imputed observations. The resulting estimates (Tables A4-A6 in Appendix A) are nearly identical to those presented in the main text below. We use the non-imputed data in the main tables, because deriving predicted probabilities from imputed data is a cumbersome process that does not lend itself easily to replication by future scholars. For most models, the non-imputed data also give a slightly more conservative estimate of the association between our main variables. 7 Targeting by airport security and targeting by law enforcement were separate items, but they were always mentioned consecutively. 8 This coding choice did not substantially change our results, tables available upon request.
Religions 2020, 11, 604 7 of 30 all three years) was 0.72. 9 The mean is relatively low because approximately 58.5% of respondents reported no experiences of discrimination. Approximately 20.5% reported having one experience, another 12% reported two experiences and 9% reported three or more. We were surprised by this distribution given the prevalence of marginalization in qualitative accounts of Muslim life in the United States (Cainkar 2009;Love 2017;Peek 2011;Selod 2018). As we will discuss below, this may reflect the difficulty of measuring marginalization in a survey setting.

Perceived Discrimination against Muslims
In 2017, respondents were asked "Just your impression, in the United States today, is there a lot of discrimination against [randomized items], or not? How about [next randomized item]?" The randomized items included Muslims, Jews, Blacks, gays and lesbians, and Hispanics. We created a binary variable for whether or not Muslims answered yes to this question, indicating a belief that Muslims do face a lot of discrimination in the US today. Approximately 2% of observations were missing, and as Table 1A shows the proportion of Muslims who answered yes to this question was 78%. Surveys of the general public on the same year show that 69% think Muslim Americans face discrimination (Pew Research Center 2017b). The discrepancy between the proportion of respondents who report experiencing specific Islamophobic incidents and the proportion who believe Muslims as a whole experience discrimination suggests that the marginalization faced by Muslims could be so normal as to be rendered invisible in daily life. One example of a quotidian form of marginalization that would not have been captured in the discrimination scale is the intensive surveillance by law enforcement of Muslim communities, what some activists have called "the feeling of being watched" (Bechrouri 2018;Boundaoui 2018;Selod 2018;Shams 2018 We have 2983 observations (out of 3084) for this variable which is reflected in the "Sample" column of Table 1A.

Dependent Variables
We gathered all variables in the surveys that indicated Muslims' views on the discrimination and deprivation faced by other groups. Of course, several of these groups have overlap with the Muslim population. By "other" here we do not mean that these groups represent mutually exclusive identities, but rather that they are groups defined by a different boundary than Muslimness-a boundary within which the respondent may or may not fall. Indeed, the interplay between Islam and Black Americans has been tremendously important for both the religious landscape of Black America and for the trajectory of American Islam (Abdul Khabeer 2016; Chan-Malik 2018; Jackson 2005Jackson , 2014Karim 2009).

Equality for Black People
The first set of outcome variables pertains to the respondent's views on Black people's status in America. In 2017, respondents were asked which statement comes closer to their own views: "our country has made the changes needed to give blacks equal rights with whites" or "our country needs to continue making changes to give blacks equal rights with whites". The order in which the statements were read by the interviewer was randomized. We created a binary variable, which we refer to as "change needed for equality for Black people", where those who chose the first statement receive a one and those who chose the second receive a zero. Approximately 6% of observations were missing. Seventy-seven percent of those who answered the question in the 2017 sample believed that change was still needed to achieve equality for Black people. This proportion among Muslims is larger than that of the general public, of whom 61% believed in the need for change (Pew Research Center 2017a).
As mentioned above, respondents in 2017 were also asked whether they believed Black people were subjects of a lot of discrimination in the US today. This is another binary variable we use as an outcome. Three percent of observations were missing, and of those who were observed, approximately three-quarters (76%) answered yes to this question. Again this is considerably higher than the proportion of the general public who answered yes to the same question (59%) (Pew Research Center 2017b).

Equality for LG People
In all three years of the survey, respondents were asked which statement came closest to their views. "Homosexuality should be accepted by society" or "homosexuality should be discouraged by society". 10 Those who chose "should be accepted by society" were assigned a one in our binary variable, whereas those who said discouraged were assigned a zero. One limitation of using this measure is that approximately 13% of respondents (across the three years) refused to answer the question. Of those observed, the proportion saying homosexuality should be accepted across the three years is 46%. As with the general population, however, responses to this variable changed considerably over the decade. In 2007, only 32% say homosexuality should be accepted by society (compared to 41% of the general public in 2006); in 2011, this proportion rises to 41% (45% in the general public); and by 2017, it is well over a majority of respondents at 61% (lower but comparable to the proportion of 70% among the general public) (Pew Research Center 2017a). 11 We also use the item asking whether the respondent believes gays and lesbians are victims of a lot of discrimination in the US today-a question asked only in 2017. Approximately nine percent of observations were missing, but of those observed, 65% of respondents answered yes. This proportion is higher than that of the general public in the same year, which is at 56% (Pew Research Center 2017b).

Other Solidarity Questions-Immigrants, Jews and Hispanics
Across all three years of the survey, respondents were also asked to choose between the following statements "immigrants today strengthen the US because of their hard work and talents" or "immigrants today are a burden on the US because they take our jobs, housing and healthcare". The binary variable, "approval of immigrants", assigns a one to those who say immigrants strengthen US society and a zero to those who say immigrants are a burden. Seven percent of possible observations are missing on this question. Given the high proportion of the sample that is foreign born or born to foreign born parents (see Table 1B), it is unsurprising that 87% of our observed sample approves of immigrants. In the general public, only 50% approved of immigrants across comparable survey years (Pew Research Center 2017a). 10 In 2007, the wording was slightly different, asking respondents to choose between "Homosexuality is a way of life that should be accepted by society" and "Homosexuality is a way of life that should be discouraged by society". Experiments conducted by the Pew Research Center suggested that removing the phrase "is a way of life" made no difference in the overall distribution of responses. 11 In sensitivity checks, we explored the possibility of an interaction effect between experiences of discrimination and year.
We did not find that the association between perceiving discrimination against one's self and accepting homosexuality is different across different years of the survey. Tables are available upon request. We could not conduct the same sensitivity analysis with our independent variable, perception of discrimination against Muslims, because that question was only asked in 2017.
Similar to the questions regarding Muslims, Blacks, and gays and lesbians, respondents were also asked whether in the US today, Hispanics and Jews faced a lot of discrimination. Seven percent of respondents did not answer the question regarding Hispanics, but of those who did, 67% answered yes. The proportion among the general public in the same year was 56% (Pew Research Center 2017b). Approximately ten percent of respondents did not answer the question regarding Jews, but of those who did, 42% answered yes. This proportion is close to that of the general public, which is 38% (ibid.). Table 1B shows means and standard errors on all controls from all three years in the first column and then separately from 2017 in the second column. Since all the controls are categorical, each is treated as a series of indicator variables. We control for age and gender as well as income. We chose income, rather than education, as a proxy for class, because returns on education depend on whether degrees were acquired domestically or abroad (Chiswick and Miller 2008;Friedberg 2000)-information the survey did not collect. 12 We also control for ethnicity, race and migration through a composite variable. The Muslim American community is extremely ethnically and racially diverse. However, standard race measures can be misleading for this population (Mohamed 2020). The largest ethnic groups among American Muslims-Arabs, Central Asians and South Asians-are racially ambiguous for complicated, layered reasons (Gualtieri 2009;Maghbouleh 2017;Morning 2001). Unfortunately, ethnicity was not directly observed in the Pew Surveys. Further complicating the situation, approximately 20% of American Muslims are Black, but we cannot be sure just from the race measure whether these respondents are converts, immigrants or descendants of either-groups that have had different trajectories.

Control Variables
To create a variable that captures the cleavages of race, ethnicity and migration in this complex population, we rely on a combination of respondent's birthplace (or parent's birthplace if the respondent was born in the US) and the respondent's categorization of her or his race. The resultant categories are: 1st or 2nd generation 13 MENA region, 1st or 2nd generation South Asian, 1st or 2nd generation Non-Black Other, 1st or 2nd generation Black, 3rd generation (or more) Black, 3rd generation (or more) White, and 3rd generation or more Other. If both the respondent and her or his parents were born in the United States, the interviewers did not ask about their ancestry. We consider these respondents third generation or more, but we cannot be sure whether they are descendants of immigrants or descendants of converts. We further divide this category by their racial identification (Black, white or other). Since 1st and 2nd generation immigrants are combined in this variable, we also include a separate generation control.
In the United States, religiosity tends to influence social and political attitudes (Schnabel 2020) and the same has been found of American Muslims (Barreto and Dana 2008;Chouhoud et al. 2019;Ocampo et al. 2018). We include both frequency of prayer and mosque attendance as controls for religiosity. The former captures personal religiosity while the latter captures the special role of mosques (Dana et al. 2011). 14 Lastly, one could argue that respondents who are more liberal would see themselves as more marginalized as well as pick up on the marginalization of other groups. To make sure we were not simply picking up a liberalism effect, we included general political ideology as a control in our models. 12 In a sensitivity analysis, we replace income with education. The coefficients on our independent variable of interest are nearly identical when we do. Results are available upon request. 13 We use the standard sociological definition wherein those born outside the United States are considered first generation and those born in the United States whose parents are foreign born are second generation. 14 Some scholars have noted that community embeddedness can influence Muslims' political leanings (Ocampo et al. 2018).
In an earlier iteration of these analyses, we controlled for the proportion of the respondents' close friends who were Muslim. Including this variable did not substantially change coefficients on our independent variables of interest, though it occasionally had its own associations with the outcome variables net of other controls. While further exploring these relationships is beyond the scope of this paper, it could be a topic for future research.
Respondents were asked "In general, would you describe your political views as very conservative, conservative, moderate, liberal or very liberal". The category moderate has a plurality of the sample at 44%, with another 28% identifying as liberal or very liberal. Where models included data from multiple survey years, we controlled for survey year as well. In earlier iterations of this analysis, we also included controls for sect and conversion status, but neither showed consistent, interesting relationships across models nor did their inclusion change the significance or magnitude of coefficients on our primary independent variables.

Results
We group our outcome variables topically, showing first attitudes towards Black people's position in America, then attitudes towards LG people, and finally attitudes towards Hispanics, Jews, and immigrants. Table 2 shows results from multivariate regressions predicting-in models 1 and 2-belief that more change is needed to achieve equality for Black people and-in models 3 and 4-perceiving Black people as victims of discrimination. Models 1 and 3 use discrimination experiences scale as the primary independent variable whereas models 2 and 4 switch in perceived discrimination against Muslims as the primary independent variable. Both independent variables have positive, significant associations with each outcome.

Equality for Black People
To visualize these associations, consider Figure 1, which graphs the predicted probability of believing more change is needed to ensure equal treatment of Black people from models 1 and 2 (panels 1 and 2 respectively). On the left, panel 1 illustrates the increase in support for change as levels of experienced discrimination rise. Muslims who report three or more experiences of discrimination are approximately a fifth more likely than those who report no experiences of discrimination to support change (86% predicted probability compared to 72%). The influence of perception of discrimination against Muslims is even more stark. Believing that Muslims are discriminated against is associated with a 39% increase in the probability of believing that change is needed for to ensure equality for Black people. Figure 2 graphs results from models 3 and 4, where the outcome variable is whether the respondent believes that in the US today, Black people face "a lot of discrimination". Again, both experiences of discrimination as a Muslim and perception of discrimination against Muslims as a group have a positive, significant association with also recognizing discrimination against Black people. Those who report no experiences of being discriminated against as Muslims are a third less likely to say that Black people face discrimination than those who report three or more experiences of discrimination. Recognizing discrimination against Muslims, both individually or as a group, is closely linked to recognizing discrimination against Black people. Those who perceive Muslims as a group as facing a lot of discrimination are nearly twice as likely to perceive Black people as also facing a lot of discrimination-the predicted probability jumps from 30% to 89%. + This category had too few observations and was therefore omitted from the analysis. ++ This category had too few observations and was therefore omitted from the analysis in models 1 and 2. * p < 0.05, ** p < 0.01, and *** p < 0.001.
Again, Black and Muslim are not mutually exclusive categories. Approximately one-fifth of the 2017 sample used in these models identifies as Black (see Table 1B). Both immigrant and non-immigrant Black people were significantly more likely than the reference group (non-Black immigrants from the MENA region) to believe in the need for change and to recognize discrimination against Black people.
We wondered whether this relationship differs by whether the respondent her or himself is Black. In results not shown, we run the same regressions after excluding any respondents who identified as Black, the results were substantially the same. Those who experienced discrimination as Muslims or who perceived Muslims to be discriminated against were more likely to believe in the need for change and to see discrimination against Black people.
Religions 2020, 11, x FOR PEER REVIEW 14 of 33 Figure 1. Predicted probability of believing change is still needed to achieve equality for Black people, computed from models 1 (left) and 2 (right) of Table 2, which utilize the 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology. Figure 2 graphs results from models 3 and 4, where the outcome variable is whether the respondent believes that in the US today, Black people face "a lot of discrimination". Again, both experiences of discrimination as a Muslim and perception of discrimination against Muslims as a group have a positive, significant association with also recognizing discrimination against Black people. Those who report no experiences of being discriminated against as Muslims are a third less likely to say that Black people face discrimination than those who report three or more experiences of discrimination. Recognizing discrimination against Muslims, both individually or as a group, is closely linked to recognizing discrimination against Black people. Those who perceive Muslims as a group as facing a lot of discrimination are nearly twice as likely to perceive Black people as also facing a lot of discrimination-the predicted probability jumps from 30% to 89%. Figure 1. Predicted probability of believing change is still needed to achieve equality for Black people, computed from models 1 (left) and 2 (right) of Table 2, which utilize the 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.  Table 2, which utilize the 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.
Again, Black and Muslim are not mutually exclusive categories. Approximately one-fifth of the 2017 sample used in these models identifies as Black (see Table 1B). Both immigrant and nonimmigrant Black people were significantly more likely than the reference group (non-Black immigrants from the MENA region) to believe in the need for change and to recognize discrimination against Black people. We wondered whether this relationship differs by whether the respondent her or himself is Black. In results not shown, we run the same regressions after excluding any respondents who identified as Black, the results were substantially the same. Those who experienced discrimination as Muslims or who perceived Muslims to be discriminated against were more likely to believe in the need for change and to see discrimination against Black people.  Table 2, which utilize the 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology. Table 3 shows results from the questions of accepting homosexuality in society and whether gays and lesbians face discrimination. Similar to the set up in Table 2, we separate out models for each independent variable of interest. Experiences of discrimination as a Muslim serves as the main independent variable in models 1 and 3, and it is swapped for perceived discrimination against Muslims in models 2 and 4. + This category had too few observations and was therefore omitted from the analysis in models 2, 3, and 4. ++ This category had too few observations and was therefore omitted from the analysis in model 2. * p < 0.05, ** p < 0.01, and *** p < 0.001.

Equality for LG People
Experiences of discrimination have a positive relationship with both accepting homosexuality in society (model 1) and seeing LG people as victims of discrimination (model 3), though the magnitudes of the associations are relatively small. On the left side panel of Figure 3, we can see that the predicted probability of accepting homosexuality rises from 43% to 49% as experiences of discrimination rise from zero to three or more-approximately a 14% increase. The left side of Figure 4 shows that those who have experienced the highest levels of discrimination as Muslims are approximately a fifth more likely than those who have had no such experiences to recognize LG people as victims of discrimination.
Religions 2020, 11, x FOR PEER REVIEW 17 of 33 Experiences of discrimination have a positive relationship with both accepting homosexuality in society (model 1) and seeing LG people as victims of discrimination (model 3), though the magnitudes of the associations are relatively small. On the left side panel of Figure 3, we can see that the predicted probability of accepting homosexuality rises from 43% to 49% as experiences of discrimination rise from zero to three or more-approximately a 14% increase. The left side of Figure  4 shows that those who have experienced the highest levels of discrimination as Muslims are approximately a fifth more likely than those who have had no such experiences to recognize LG people as victims of discrimination.  Table 3, which utilize all three survey years (2007, 2011, and 2017) and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.
When the predictor is whether one perceives Muslims as discriminated against, the positive correlation is stronger for both outcomes. The difference in predicted probability of accepting homosexuality between those who do and those who do not perceive Muslims as victims of discrimination is approximately 21% (11 percentage points). The magnitude of the association between perceived marginalization of Muslims and perceived marginalization of LG people is much stronger-those see Muslims as victims of discrimination are twice as likely as those who do not to also see LG people as victims of discrimination. 15 Coefficients on the controls in the models show that, in general, when it comes to both acceptance of homosexuality and perceived discrimination against LG people, younger Muslims are more tolerant than their elders at almost every age group. Women are more tolerant than men. Thirdgeneration people were generally more tolerant than immigrants, among whom there were not consistent generational differences. Those in the highest income range are more accepting than those in the lowest. The most religious Muslims and the least religious Muslims differ significantly from each in their level of acceptance of homosexuality, but the negative association between religiosity and tolerance appears stronger in the frequency of prayer variable than in mosque attendance. This 15 Here too, we are considering a category that is not mutually exclusive with Muslim. Unfortunately, the survey questionnaire does not enquire about sexual orientation, so we are unable to distinguish LG Muslims from non-LG ones. That is a limitation of this study.  Table 3, which utilize all three survey years (2007, 2011, and 2017) and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.
Religions 2020, 11, x FOR PEER REVIEW 18 of 33 may suggest that religious Muslims' personal convictions are a stronger deterrent to accepting LG people than embeddedness in Muslim congregations, a hypothesis that future study could potentially test with more extensive data. 16 Evidence for an association between religiosity and perceived discrimination against LG people appears less clear. One interpretation of these divergent results could be that even while maintaining discomfort with homosexuality, Muslims may continue to see LG people as subjects of discrimination and therefore as an oppressed group. LGBTQIA+ Muslims have criticized mainstream Muslim community leaders for showing solidarity with Queer people outside the Muslim community but not within it (Graham 2016).  Table 3, which utilize the 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.
More politically liberal respondents were more accepting of homosexuality but not more likely to see discrimination against LG people. Finally, positive coefficients on survey year in model 1 reflect the rapid increase in acceptance for homosexuality among Muslims over the decade between the first survey (2007) and the last (2017).  Table 3, which utilize the 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.

Other Solidarity Questions-Immigrants, Jews and Hispanics
When the predictor is whether one perceives Muslims as discriminated against, the positive correlation is stronger for both outcomes. The difference in predicted probability of accepting homosexuality between those who do and those who do not perceive Muslims as victims of discrimination is approximately 21% (11 percentage points). The magnitude of the association between perceived marginalization of Muslims and perceived marginalization of LG people is much stronger-those see Muslims as victims of discrimination are twice as likely as those who do not to also see LG people as victims of discrimination. 15 Coefficients on the controls in the models show that, in general, when it comes to both acceptance of homosexuality and perceived discrimination against LG people, younger Muslims are more tolerant than their elders at almost every age group. Women are more tolerant than men. Third-generation people were generally more tolerant than immigrants, among whom there were not consistent generational differences. Those in the highest income range are more accepting than those in the lowest. The most religious Muslims and the least religious Muslims differ significantly from each in their level of acceptance of homosexuality, but the negative association between religiosity and tolerance appears stronger in the frequency of prayer variable than in mosque attendance. This may suggest that religious Muslims' personal convictions are a stronger deterrent to accepting LG people than embeddedness in Muslim congregations, a hypothesis that future study could potentially test with more extensive data. 16 Evidence for an association between religiosity and perceived discrimination against LG people appears less clear. One interpretation of these divergent results could be that even while maintaining discomfort with homosexuality, Muslims may continue to see LG people as subjects of discrimination and therefore as an oppressed group. LGBTQIA+ Muslims have criticized mainstream Muslim community leaders for showing solidarity with Queer people outside the Muslim community but not within it (Graham 2016).
More politically liberal respondents were more accepting of homosexuality but not more likely to see discrimination against LG people. Finally, positive coefficients on survey year in model 1 reflect the rapid increase in acceptance for homosexuality among Muslims over the decade between the first survey (2007) and the last (2017).

Other Solidarity Questions-Immigrants, Jews and Hispanics
In the final table, we consider Muslims' views on immigration as well as whether they identify Jewish and Hispanic people as victims of discrimination in the United States-our final measures of intergroup solidarity.
The first two columns of Table 4 and the corresponding Figure 5 show results predicting approval of immigrants. As the left side panel shows, there is no significant increase in immigrant approval as Muslims experience more discrimination. Only perception of discrimination has the positive association we saw in other models, the magnitude is small and the coefficient is only significant at the p < 0.05 level. But as Figure 5 (and Table 1A) also shows, of all the outcome variables, approval of immigrants was the highest (87%). The lack of variation on this outcome could explain our weak findings here. Where we do have a significant association, in model 2 (right side of Figure 5), those who perceived of Muslims as facing discrimination had a 93% predicted probability of approving of immigrants, five percentage points higher than those who did not. The latter's approval of immigrants was still at a whopping 88%. Recall that among the general public, the average approval of immigrants across three corresponding survey years was only 50%. * p < 0.05, ** p < 0.01, and *** p < 0.001. + This category had too few observations and was therefore omitted from the analysis in models 2, 5 and 6. ++ This category had too few observations and was therefore omitted from the analysis in model 2.
a 62% probability of seeing Hispanics as victims of discrimination, a probability that rises to 82% among those who have had three or more experiences of discrimination as Muslims. Once again, perceived discrimination against Muslims is a stark dividing variable. Those who see Muslims as victims of discrimination are over three times as likely as those who do not to also identify Hispanics as victims of discrimination. 17 Figure 5. Predicted probability of approving of immigrants, computed from models 1 (left) and 2 (right) of Table 4, which utilize all three survey years (2007, 2011, and 2017) and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.
While overall, Muslims are less likely to see Jews as victims of discrimination in the United States, the extent to which they do increases both with experiences of and perception of discrimination against themselves as Muslims. As Figure 7 shows, the predicted probability of seeing Jews as victims of discrimination is 45% higher for those who have experienced discrimination as Muslims (54%) than those who have not (37%). Further, yet again, the great divider is whether Muslims perceive themselves as marginalized. Those who do not only have an 8% probability of seeing Jews as facing discrimination, while those who do are at 51%-a difference of over five-fold! Figure 5. Predicted probability of approving of immigrants, computed from models 1 (left) and 2 (right) of Table 4, which utilize all three survey years (2007, 2011, and 2017) and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology. Models 3-6 use perceptions of discrimination against Jews and Hispanics as the outcome variables. Here again, both experiences of discrimination as Muslims and perceptions of discrimination against Muslims as a whole are associated with higher perceptions of solidarity against these groups. As Figure 6 illustrates, those who report no experiences of discrimination have a 62% probability of seeing Hispanics as victims of discrimination, a probability that rises to 82% among those who have had three or more experiences of discrimination as Muslims. Once again, perceived discrimination against Muslims is a stark dividing variable. Those who see Muslims as victims of discrimination are over three times as likely as those who do not to also identify Hispanics as victims of discrimination. 17 While overall, Muslims are less likely to see Jews as victims of discrimination in the United States, the extent to which they do increases both with experiences of and perception of discrimination against themselves as Muslims. As Figure 7 shows, the predicted probability of seeing Jews as victims of discrimination is 45% higher for those who have experienced discrimination as Muslims (54%) than those who have not (37%). Further, yet again, the great divider is whether Muslims perceive themselves as marginalized. Those who do not only have an 8% probability of seeing Jews as facing discrimination, while those who do are at 51%-a difference of over five-fold! Overall, our results suggest that Muslims who experience discrimination as Muslims or who perceive Muslims as facing discrimination are more solidaristic towards other identity groups. Intergroup recognition of solidarity was the strongest and most consistent result across our models, suggesting that the ability to recognize one's self as marginalized translates to also "seeing" other forms of marginalization.
Religions 2020, 11, x FOR PEER REVIEW 22 of 33 Figure 6. Predicted probability of perceiving discrimination against Hispanics, computed from models 3 (left) and 4 (right) of Table 4, which utilize 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.

Figure 7.
Predicted probability of perceiving discrimination against Jews, computed from models 5 (left) and 6 (right) of Table 4, which utilize 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.  Table 4, which utilize 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.
Religions 2020, 11, x FOR PEER REVIEW 22 of 33 Figure 6. Predicted probability of perceiving discrimination against Hispanics, computed from models 3 (left) and 4 (right) of Table 4, which utilize 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.  Table 4, which utilize 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.  Table 4, which utilize 2017 data and include controls for age, gender, ethnicity, generation, income, mosque attendance, frequency of prayer, and political ideology.

Discussion
Overall, our results suggest that Muslim Americans who experience individual discrimination based on their Muslim identity or who perceive Muslims as facing discrimination as a group show more attitudinal and political support towards other minority groups, including African Americans, immigrants, gays and lesbians, Jews, and Latinx people. This finding-that experiences and perceptions of in-group deprivation also result in feelings of intergroup solidarity-has also been suggested, but not fully emphasized, in previous research on Black, Asian, and Latinx Americans (e.g., Cain et al. 1991;Espiritu 1992;Tate 1993;Dawson 1994;Sanchez 2008). While the great majority of research investigating the consequences of experiences and perceptions of in-group marginalization have focused on in-group solidarity as the outcome of interest, our findings suggest that intergroup solidarity is a phenomenon that deserves greater attention in studies of Muslim Americans and US minority groups in general.
While both of our measures of perceived deprivation-individual experiences of discrimination and the belief that Muslims as a group faced discrimination-predicted greater intergroup solidarity among our Muslim respondents, it was a sense of group discrimination that was a stronger and more consistent predictor of intergroup solidarity. In other words, seeing oneself as a part of an aggrieved group is more likely to lead Muslim Americans to support and empathize with members of other marginalized groups than are individual experiences of anti-Muslim discrimination. Notably, this finding is consistent with the few studies of intergroup solidarity among African Americans, Asian Americans, and Latinx Americans, in which a shared sense of group deprivation-called, alternatively, "group consciousness", "minority group status", "power discontent", or "perceived disadvantage"-has been a consistent predictor of expressions of concern for, support of, or commonality with other socially marginalized minority groups (Cain et al. 1991;Gurin et al. 1980;McClain et al. 2009;Uhlaner 1991).
Why would perceived group deprivation be a stronger predictor of intergroup solidarity than individual experiences of discrimination? From our perspective, it seems likely that group deprivation has a stronger association with feelings of intergroup solidarity than do individual experience of discrimination because the first two forms of cognition that involve seeing the social world primarily through the lens of "groups" (Brubaker 2004), and aggrieved groups in particular. If a person has learned to see oneself as a member of a deprived group, it seems logical that they could then extend this cultural cognitive template, or "schema", outward towards other groups who may seem similar in their social condition or situations (DiMaggio 1997;Sewell 1992;Swidler 1986). While individual experiences of discrimination also have some relationship with feelings of intergroup solidarity, we feel that the sense of being part of a group that faces discrimination may be a more powerful driver of cross-group empathy because both of these modes of thought are grounded in "groupness" (Brubaker 2004).
Our proposed explanation for the strong link between group deprivation and intergroup solidarity may seem more convincing and less abstract when it is connected to the particular mechanisms, or social processes, though which members of racialized minority groups might move from feelings of in-group deprivation to intergroup solidarity. Remaining with Muslim Americans as our empirical example, we can speculate that one tangible means through which Muslims could both gain a sense of themselves as members of a socially deprived group and think in terms of intergroup solidarity is through communal organizational and political action. Indeed, previous studies of minority group coalition formation in US politics emphasizes practical, communal political action as a crucible in which participants may develop a strong in-group consciousness along with a sense of common cause with other minority groups (Espiritu 1992;Garcia Bedolla 2005;Padilla 1985). Likewise, case studies and qualitative research reveal clear efforts by Muslim organizations and leaders to both build a sense of in-group collective identity as well as build coalitions with and support the social concerns of an array of other social groups-Jewish Americans on anti-Semitism and harassment, African Americans on police brutality and racism, Latinx Americans on ending DACA and supporting DREAMERs, and (at times) sexual minorities on harassment and discrimination (Calfano and Lajevardi 2019;Tekelioglu 2019;Yukich 2018). This evidence suggests the possibility of institutional and organizational factors conditioning the relationship between the perception of in-group deprivation among American Muslims and their support and empathy for other minority groups. While this is a relationship we cannot measure using existent survey data (which do not contain sufficient inquiries of respondents' political organizing or organizational memberships), the potential importance of organizers and activists in providing the social and organizational contexts within which a sense of group deprivation and intergroup solidarity can be cultivated, among Muslims as well as other marginalized US minorities.
Another way to think about the connection between the sense of in-group deprivation among Muslims and their increased sense of intergroup solidarity is to consider the significance of the common political opposition faced by American Muslims and the other groups for which they express intergroup solidarity, namely, right-wing political forces of white Supremacy and Christian nationalism, who also support President Donald Trump (Braunstein 2019;Dahab and Omori 2019;Edgell 2016;Lajevardi and Oskooii 2018;Perry and Whitehead 2015). While the previous mechanism suggests that Muslims and other minority groups might develop a sense of in-group deprivation and then intergroup solidarity through political action and community organizing, this mechanism centers on the recognition by those who feel they are members of socially deprived groups that they share the same political opposition as other minority groups in their midst. Just as American Muslims heard and saw Trump make defamatory statements about Muslims, appoint openly Islamophobic officials, and try to enact a restriction on Muslims entering the country, so did they see and hear these other racialized and marginalized groups-Latinx and African immigrants, Black Americans, Jewish Americans, and gays and lesbians-face similar social and political attacks from Trump and his key constituency (Bail 2014;Braunstein 2019;Dahab and Omori 2019;Edgell 2016;Selod 2015;Whitehead and Perry 2020). The cross-group solidaristic political attitudes expressed by American Muslims-and others-who feel group deprivation could be relational in nature, and might be shaped in part in response to a political climate in which they are one of many marginalized social groups under attack from the same right-wing political coalition. This may be why those Muslims who feel the strongest sense of group deprivation also feel the strongest support for these other marginalized groups-these "enemies of their enemy" are considered political friends.

Conclusions
The fact that the great majority of US Muslims feel a strong sense of group deprivation, and that this sense predicts empathy and concern for other politically marginalized groups, may mean that Muslims are taking their place as a politicized racial and ethnic group, as much as a religious one, and one whose concerns extend out to other social groups similarly marginalized in contemporary American politics-Blacks, LGBTQIA+ individuals, Hispanics, Jews, and immigrants. These results not only suggest the potential for cross-group political coalitions and allegiances, but also demonstrate that this sense of allegiance with other marginalized groups-this solidaristic attitude-is powerfully shaped for American Muslims by their own perceptions of in-group deprivation. In this way, our paper contributes to and extends the social scientific literature on the racialization and political behavior of American Muslims in contemporary American society. While many have looked at the processes of racialization and discrimination and their effect on Muslims in terms of their own self-perceptions, our project demonstrates how these processes impact the political attitudes and expressions of Muslim Americans, not simply as recipients of discrimination but as agentive political thinkers and actors in their own right.
Finally, our results offer a testable hypothesis with important implications for organizers and activists. Future research could test whether perceived marginalization leads to intergroup solidarity among other minority groups. If it does, this affirms the importance of in-group consciousness raising as a crucial component of intergroup coalition building.     Multiple imputations conducted using chained regressions over 20 iterations in Stata 16. Corresponds to Table 2 in the main text. * p < 0.05, ** p < 0.01, and *** p < 0.001.  Multiple imputations conducted using chained regressions over 20 iterations in Stata 16. Corresponds to Table 3 in the main text. * p < 0.05, ** p < 0.01, and *** p < 0.001. Multiple imputations conducted using chained regressions over 20 iterations in Stata 16. Corresponds to Table 4 in the main text. * p < 0.05, ** p < 0.01, and *** p < 0.001.