Are refugees really welcome? Understanding Northern Ireland attitudes towards Syrian refugees

granular research will be needed to highlight these nuances.


Introduction
In June of 2018, the Belfast Telegraph reported that the people of Northern Ireland (NI) had provided a 'welcoming' community to resettled Syrian refugees (McHugh 2018). Another news article stated that the Department of Communities in NI had 'overwhelming support from communities throughout NI to assist refugee families' (Edwards 2018:1). NI has played an important role in addressing the Syrian humanitarian crisis starting in 2011. Since 2015, NI has taken in 1,258 Syrian refugees since December 2018 (Edwards 2018). This effort was due primarily to the United Kingdom's (UK) enactment of the Syrian Vulnerable Person Resettlement Programme (SVPRP), which was established to resettle at least 20,000 Syrians from overburdened Middle Eastern refugee camps (Home Office 2017). As of 2017, the UK SVPRP program had only settled 7,307 of its promised 20,000 resettlements. Thus, NI has taken in around 17 per cent of all the refugees resettled in the UK, but NI only represents 2.8 per cent of the total UK population (NISRA 2018; Home Office 2017). As of 2019, NI had the highest number of Syrian refugees resettled in its communities than any other UK country with 1,637 individuals (Meredith 2019).
To verify NI's news claims being a welcoming society, Michael and Devine (2018) examined recent data from NI. They found that at least 52 per cent of respondents supported allowing asylum-seekers. Although only providing a bivariate analysis, they suggested that much of this hospitality came from NI citizen sentiments that the country has a duty to protect refugees. Also, the most substantial support from respondents was people who lived in religiously-mixed and urban areas; however, NI remains a divided society where most neighbours live in primarily Protestant or Catholic areas (Shuttleworth and Lloyd 2013) and have a substantial rural population (36 per cent) (NISRA 2020). There also seemed to be general support for refugee resettlement across most NI age groups, political affiliations, and national identities (i.e. British, Irish, or neither).
Although this study and news coverage suggested some positive attitudes towards Syrian refugees, researchers have also suggested increasing anti-immigrant attitudes over the last two decades (see Doebler et al. 2018). Notably, researchers have suggested that NI has become the 'Race Hate Capital of Europe' (Gilligan and Lloyd 2006). For example, NI citizens' negative views of Muslim and Eastern European groups have increased from the early 2000s to now (see Doebler et al. 2018;Hayes and Dowds 2006;Michael 2017). Amnesty International (2018) pointed out that more than half of the people surveyed in NI would not willingly accept a Muslim as a close friend or a relative by marriage. They also asserted that around 25 per cent stated that they would not willingly work beside a Muslim. Other researchers also found that other groups were racist or prejudiced attitudes, including Eastern Europeans, Roma, and Asian subgroups (see Knox 2011). Studies also documented little to no support of Syrian refugees in NI within smaller isolated subgroups, including Democratic Unionist Party (the largest Unionist-Protestant leaning political party) supporters and Protestant sixteenyear-olds (Michael and Devine 2018). Although these studies have focused on Muslims and not necessarily Syrian refugees, like other researchers (see Abba 2019), this study recognizes that these two groups are not the same. However, European populist national framing at the beginning of this crisis has conflated the two identities as similar due to terrorism and the possible threat these outgroups may present towards European religious and cultural beliefs.
Recent NI news reports have also suggested increasing numbers of racist attacks and hate crimes over the last decade due to these identities' conflation. A Belfast Telegraph (2018) article reported that of the 335 violent incidents reported to NI's Public Prosecution Service, 38.5 per cent were racially motivated, whereas 24.8 per cent were sectarian. Moreover, there has also been some violence targeting Syrians. For example, a 15-year-old Syrian refugee was attacked and threatened at school by NI native-born British and Irish students by holding him down by his throat and pouring water into his mouth (Cockburn 2018). Participation and Practice of Rights (2018) also released a report on six Syrian families' testimonies of racially-motivated abuse and discrimination. These acts included not having to access adequate housing and racist attacks directed towards their children to the point of being afraid to send their kids out to play or to school.
Researchers have often pointed out that the NI context matters in explaining the varied attitudes and reports. Cultural identity in NI is incredibly complex, with intersecting politics, nationalism, and religion. These are inherently difficult to parse out as both distinct markers with strong overlapping and reinforcing components (Mitchell 2013). These identities are shaped by historical events that have led to a deeply divided society that broadly fall into the two communities of Unionist-British-Protestant and Republican-Irish-Catholic. Particularly, NI has been labelled as a country filled with hate due to the decades of sectarian violence and institutional discrimination during the Troubles. For much of the country's history, Catholics and Protestants found themselves in conflict and almost fully segregated across many institutions, including neighbourhoods, schools, workplaces, the government, and even within graveyards (McKittrick and McVea 2002). This sectarian divide was further complicated and escalated because the religious identities were intertwined with political and nationalist divisions related to whether NI should be united with the Republic of Ireland as one Irish nation (i.e. Nationalists and Republicans) or should stay under British rule and be a part of the UK (i.e. Loyalists and Unionists). Notably, the debate on where the border was drawn for NI was heavily based on concentrating the Protestant colonizers remaining under the British rule, which fed into ongoing debates in NI on who is the 'majority,' 'minority,' or other groups (Anderson and Shuttleworth 1998). Boal and Livingstone (1984) identified these complex identity formations as representative of 'ethnonationalism,' which has uniquely shaped views of insiders and outsiders within a country consistently dealing with conflict, violence, and segregation among its native-born British and Irish citizens. Although the Good Friday (Belfast) Peace Agreement in 1998 pushed for an end to the violence and for equality in sharing resources, civil rights, and integration among its divided citizens, NI has continued to be segregated across neighbourhoods, schools, and Northern Ireland attitudes towards Syrian refugees 3 workplaces due to the lasting sectarian divisions (see McKittrick and McVea 2002;Jarman et al. 2009). Importantly, these identities transcend religious affiliation and ideology. Although there are religious influences such as churches fostering communities, these identities are culturally embedded and reinforced in NI society's fabric to the point of being ethnic community groups (Mitchell 2013).
Because of this complicated history, researchers have struggled to tease out the social conditions and problems foreign-born minorities have faced. Moreover, Knox (2011) argued that racism has always existed in NI and sectarian issues were often racially motivated because of perceptions of ethnic differences between Catholic-Irish-Republicans and Protestants-British-Unionists. Particularly, racist or discriminatory actions against foreign-born minorities were often lost because of the Troubles and have only become a focus in the last couple of decades (see McVeigh 2015). It should also be noted that negative attitudes, prejudice, and discrimination towards other ethnic groups (i.e. Muslims, Syrians, and Eastern Europeans) is a recent phenomenon simply because the past conflicts kept the number of foreign-born arrivals low in NI for decades. As of 2001, the foreignborn population born outside of the UK and the Republic of Ireland was around 1 per cent. By 2011, it grew to around 4.5 per cent (The Migratory Observatory 2014). Thus, until the early 2000s, NI has been racially homogeneous, existing as a primarily 'white' European country. Overall, researchers have become more interested in understanding how NI citizens and their complicated history and identities might predict their views of foreign-born populations.
Current research on attitudes towards other ethnic minority groups and particularly Syrian refugees resettling in Europe has only scratched the surface, only providing descriptive explanations (see Michael and Devine 2018; Steele and Abdelaaty 2019). Our research seeks to fill this empirical gap. First, we attempted to uncover which theory-informed factors have shaped NI's attitudes towards Syrian refugees. Specifically, we examined variables that were matched to four broad theoretical explanations of local/national/native-born and citizen attitudes towards foreign-born and immigrant populations, including economic selfinterests, social exposure (i.e. contact hypothesis), cultural marginality, and group threat or conflict based on ethnonationalist politics. Second, we investigated the explanatory power of the various NI identities that may shape NI respondent attitudes and actions. We also used the Northern Ireland Life and Times Survey (NILT) 2015 dataset that specifically asked respondents about their views of Syrian resettlement. Since 1998, the NILT has collected data on social attitudes of people living in Northern Ireland, providing a wide range of information related to social policy issues (see NILT 2021). We argued that by using these theories and the unique contextual history of NI for the basis for our analysis, we could better understand the nuances behind NI people's attitudes towards Syrian resettlement. More importantly, we can contextually discern whether racism, sectarianism, or both have influenced attitudes towards Syrian refugees and the possible acceptance of these individuals who will influence NI life and politics as they continue to work and live in NI.

Theoretical and Empirical Background
In this paper, we are investigating attitudes towards Syrian refugee resettlement in NI using secondary data analysis. Notably, we cannot isolate some terminologies with crucial nuances. For instance, we cannot distinguish between views of asylum seekers from refugees, even though these groups have different opportunities and experiences when migrating. We consider views on Muslims and racial minorities but acknowledge many Syrians are not Muslim, and the non-white racial categorization is debatable on which racial group Syrian refugees would fall into in the context of NI. Although the data prevents pinpointing what characteristics respondents associate with Syrians resettling, their responses to specific group categories (i.e. Muslim, Syrian, Refugee, or Asylum Seeker), we rely on the known NI context to determine the likely interpretations would consist of refugees that are primarily Muslim and non-white.
With this limitation noted, this study attempts to provide a precise theoreticallydriven examination of what factors have impacted people's attitudes towards refugees has been scant. In most studies, researchers have relied primarily on theoretical explanations focusing on immigration and general theories on group threat and ethnic competition. Steele and Abedelaaty (2019) argued that as the discussion on refugee attitudes has been relatively recent for scholars in the western hemisphere, it stands to reason that theories examining immigration attitudes have served as a sufficient starting point. They also pointed out that most of these theories were written to predict ingroup (i.e. native-born citizens) views of outgroups (i.e. foreign-born non-citizens). Therefore, we focus on four theoretical perspectives regularly used in explaining attitudes towards immigration: economic self-interests, social exposure, cultural marginality, and group threat/competition. We also made sure to note how NI's ethnonationalist context and conflict might shape attitudes towards Syrian refugees.

Economic Self-Interest
Social scientists have often examined whether attitudes towards immigrants or foreign-born groups were based on economic self-interest. Downs (1957) first suggested that individual political attitudes were shaped by a person's assessment of whether a policy or political action economically benefited or harmed them. More recently, Fetzer (2000:7) argued that this perspective could be applied to attitudes on any politically-charged discussion like immigration. He described this as an argument that 'contends that people's [immigration] attitudes primarily reflect their narrow, material self-interest.' He also argued this allowed researchers to use a range of demographic variables to predict immigrant (and refugee) attitudes, including income, employment status, occupation, and education levels. In other words, individual attitudes towards Syrian refugee settlement may depend on a person's economic status and whether they see resettlement as a benefit or cost to their financial wellbeing or socioeconomic status.

Northern Ireland attitudes towards Syrian refugees 5
This theory would predict that those with lower incomes, education, and being unemployed would lead to higher anti-immigrant attitudes, but findings have been mixed. For example, Hayes and Dowds (2006) found that lower economic status measured by unemployment, low household income, and subjective views of their NI's economy's overall health had little to no effect on anti-immigrant attitudes. More recently, Naumann et al. (2018), examining 15 European nations, found that individual economic concerns did not increase anti-immigrant attitudes. These results were mirrored in several other European studies (i.e. Ceobanu and Escandell 2010;Chandler and Tsai 2001;Citrin et al. 1997;Fetzer 2000;Hayes and Dowds 2006;Kehrberg 2007).
Although a person's socioeconomic status does not necessarily influence their attitudes, researchers found that when individuals were asked about their economic well-being or even the country's economic stability may be threatened by immigrants; then economic self-interest became more critical. For example, a prominent finding among European studies shows that native-born and citizen workers with matching skills of incoming foreign-born immigrant workers tended to be more anti-immigrant based on the probable direct competition (i.e. Borjas 2016;McLaren and Johnson 2007;Naumann et al. 2018). Also, researchers found that anti-immigrant attitudes increased when individuals considered whether immigrants could access social benefits coveted by citizens (i.e. health insurance, economic welfare programs) (i.e. Espenshade and Hempstead 1996;Degen et al. 2019;Mayda 2006;van Oorschot 2006;Senik et al. 2009). Thus, attitudes towards refugees could be swayed by an individual's economic situation; mainly, if they were in direct competition for resources. It should be noted that within the NI, Syrian refugees have access to social benefits and housing as well as are allowed to work (Department of Communities NI 2020). NI also has competition on seeking benefits; for example, in 2014-2015, almost 40,000 social housing applications were allocated only about a quarter (NISRA 2019). Thus, NI's situational factors may cause different results than what has been found in other locales with different welfare systems.

Social Exposure (Contact Hypothesis)
Social scientists have also examined whether there are theories to predict positive attitudes towards foreign-born populations. One prominent theory has been the contact hypothesis (i.e. social exposure). First discussed by Allport (1958), the theory posed that when ingroup individuals had close and sustained contact with outgroup individuals, that contact would foster positive attitudes towards the outgroup.
These close connections to outgroup members included co-workers, schoolmates, or close friendships between ingroup and outgroup members. For example, Hayes and Dowds (2006) found that NI citizens had more positive attitudes towards immigrants when their friendships included foreign-born immigrants. Another more recent NI study by Doebler et al. (2018) found that regular contact with Eastern Europeans and other minorities linked to more tolerance of these groups, especially if NI citizens were already involved in religiously mixed schools or other social settings. Kehrberg's (2007) multinational European study of immigrant attitudes confirmed that quality exposure was vital to more positive attitudes. However, Croucher et al. (2014) found that increased contact escaladed feelings of threat and prejudice for Spanish citizens towards Muslims. The researchers also suggested that although this seems to contradict other studies on contact, it is clear that they did only measured frequency, not the quality of the contact. Therefore, quality contacts such as close friendships and not merely exposure to outgroups may be vital in predicting positive attitudes towards Syrian resettlement in NI.

The Case for NI: Cultural Marginality and Ethnic Competition
NI represents a unique context for examining attitudes towards refugees. Notably, the centuries-old divisions between ethnonationalist identities have fuelled decades of contention and inequality. As noted by several scholars (e.g. Brewer 1992;McKittrick and McVea 2002), these divisions have caused civil unrest, and many of these groups have faced significant hardships and violence, leaving deep rifts between native-born British and Irish ingroup or local communities. Even after the signing of the 1998 Good Friday (Belfast) Agreement, many of these communities remain highly segregated and still report marginalized feelings (see Doebler et al. 2018). Thus, some scholars argue that these divisions and feelings in NI make predicting attitudes towards immigrants or refugees distinctive (see Knox 2011;McVeigh 2015). They also postulated that the sectarian and nationalist divides might encourage sympathy (or rejection) to other groups perceived as facing similar inequality and marginality levels.
One theoretical argument that helps explain attitudes towards immigrants and refugees arriving in NI's unique context is cultural marginality. This theory suggests that native-born citizens will have more positive views of foreign-born immigrants if their communities have experienced marginalization or systemic oppression in their communities (Fetzer 2000;Hayes and Dowds 2006). Looking in France, Germany, and the US, Fetzer (2000) found that antiimmigrant attitudes decreased if an individual belonged to a religious or racial/ ethnic minority group or was foreign-born immigrants. Rustenbach's (2010) study using European Social Survey data controlled for other variables that represented human capital, political affiliation, and economic competition and found that cultural marginality did not affect immigrant attitudes.
Contact theory applied to NI has some mixed results. For example, Hayes and Dowds (2006) found that Catholics were more supportive of immigrants than Protestants, and identifying as 'British' or 'Irish' had no significant effects. However, in a more recent study, Doebler et al. (2018) argued that it was not necessarily if an NI citizen identified as a Catholic or Protestant, Irish or British; however, negative attitudes towards immigrants and ethnic minorities was due to NI citizens' ethno-identity that includes both religious and nationalist interactions. They also concluded that although NI citizens do have some negative Northern Ireland attitudes towards Syrian refugees 7 attitudes to foreign-born minorities, it still is not at the levels recorded in the rest of the UK.
Other researchers have argued that some of this negativity towards other ethnic minorities, particularly Muslims, is more a consequence of the growing fears of terrorism, the rise in immigration to the UK, and the subsequent populist movements (i.e. Brexit) (Abbas 2019;Croucher et al. 2014;Esses et al. 2008Esses et al. , 2017Lewin-Epstein and Levanon 2005). Also, these scholars have suggested that it may be possible that NI citizens may have shifted focus away from each other and have now turned to disliking foreign-born and ethnic minorities as scapegoats for possible shifts away from NI white ethnic traditions (see Doebler et al. 2018;Knox 2011;McVeigh 2015). Thus, we contend that ethnic competition may explain negative attitudes.
Several studies have also argued and found something different in shaping attitudes beyond shared experiences of cultural marginality. The opposite could occur if in-groups do not feel any kinship with outside groups but rather threat and competition may shape attitudes towards foreign-born or outgroup populations. Bloch (2004) found that when native-born UK citizens perceived severe threats to their access to labour markets in Great Britain, they were more antiimmigrant, supporting a policy that restricted immigrant labour market participation. Emmenegger and Klemmensen (2013) and Meuleman et al. (2009) found that native-born citizens were mostly against any redistribution of social benefits and jobs to incoming immigrants within European countries. Although there was some variation based on an individual's humanitarian beliefs, most respondents with fears of losing social benefits had more anti-immigrant sentiments. Schneider (2007) made an important distinction in how ethnic competition creates anti-out group attitudes; precisely, perception and fear of the unknown, not an actual reality of outgroup members taking ingroup resources, was key to understanding the power of ethnic competition shaping attitudes. This study confirmed Blumer's (1958) proposal that prejudice increased when the ingroup thought they should have privileged access to resources over the outgroup.
Social scientists have also assessed whether perceived threats to 'cultural interests' impacted individuals' views of immigration. According to Higham (2000) and Jaret (1999), cultural interests include cultural elements that defined being a citizen of a particular country. For instance, an increased and overwhelming influx of immigrants who do not speak the native language, eat traditional/local foods or understand cultural practices in the workplace make ingroup or citizens feel that new groups may destroy or change citizens' 'normal' customs (see Lippard 2011).
Recently, researchers have documented this idea of ethnic competition when studying Muslim immigration to Europe. Savelkoul et al. (2011) found that native-born citizens perceived Muslims would challenge cultural norms and values in the Netherlands due to their religious practices. Croucher et al. (2014) found that Spanish citizens felt their religious beliefs and way of life would be under attack with Muslim immigrants' arrival. A study of anti-Muslim prejudice across 30 different countries had similar findings (Strabac and Listhaug 2008). Notably, anti-Muslim prejudice increased significantly for many Europeans who thought Muslims would harm their cultural traditions or believed they would not assimilate into European cultures(see Croucher et al. 2014).
Although we understand that various negative attitudes towards Muslim immigrants are not synonymous with views towards Syrian refugees in NI, it is clear that media and the public have often conflated the two identities. For example, Tyler's (2018) research on reactions towards Syrian refugees in Europe found that reactions and attitudes were mostly racist, and the public often suggested that Syrian refugees were not only all Muslims but were terrorists working for ISIS (ISIL). Sadeghi (2019) also found that Germans interviewed about the recent refugee crisis regularly viewed Iranian immigrants and naturalized citizens as radical Muslims and a threat to a German way of life.
In NI, ethnic threat and competition may be growing. Principally, the growth of ethnic competition may be due to NI's continuing divisions and recent demographic shifts. In 2014 42 per cent of the population was Protestant, whereas 41 per cent was Catholic (NIRSA 2014b). Thus, Catholics may soon become the majority population for the first time. Additionally, Unionists (representing mostly Protestant and British-identifying NI citizens) lost their majority in the government in 2017. Overall, these significant population and political shifts may push some NI Protestants to worry about losing the majority, and thus they may be more antirefugee if these groups align with Catholic interests. Although Catholics may feel a kinship to other marginalized groups, Protestants may express more angst towards a growing minority population that threatens their social position in the country. Doebler et al.'s (2018) study of negativity towards immigrants and ethnic minorities in NI from 2004 to 2015 found that sectarianism impacted negative attitudes of newly-arriving minorities. Although they could not conclude that it was due to the notion that minorities posed a threat to NI citizen resources or cultural interests, they did suggest further research would need to determine how NI's continued religious and ethnic segregation among its white citizens may spill over into the social, economic, and political treatment of new groups. Also, Vieten and Murphy (2019) suggested in their study of asylum seeker and refugee treatment that Belfast might be more racist and discriminatory because of the long history of ethnonationalist struggle over obtaining equal rights, treatments, and resources in a divided society. Thus, any strain on already-bitter fights for resources might be further magnified by the arrival of other groups wanting equal access. Thus, adding other outside groups like Syrian refugees may make NI citizens feel more threatened that they may lose access to resources (and rights) despite positive progress since the peace agreement in 1998.

Data and Sample
The study utilizes data from the 2015 Northern Life and Times Survey (NILT). The NILT is an ongoing attitudinal cross-sectional panel survey sampling the NI population. The survey measures attitudes on a range of issues, including a module on minority issues. This survey has a representative sample drawn from Northern Ireland attitudes towards Syrian refugees 9 postcodes with 1202 men and women aged 18 years old and older. The 2015 dataset was chosen because it specifically included a question on attitudes towards Syrian refugees and a range of variables that capture the multiple theories of interest. After excluding missing responses, the analysed sample was 1,181. The majority of nonresponse (don't know, refused, cannot choose, other, or missing) was included in the analysis as missing flags as these could be non-random responses, but further analysis showed no pattern of nonresponses across variables. Imputation was not done as most nonresponses were not missing (such as don't know). Only 21 cases were dropped due to having a deficient number of nonresponses (14 or less) deemed too small to warrant a missing flag.

Measures
The dependent variable derives from the question: 'Thinking about the current situation in Syria, how much do you agree or disagree that people from Syria should be allowed to come to Northern Ireland.' Although not explicitly using the term refugee, the wording captures the concept without drawing on jargon that respondents may not understand. The original responses were a 5-point Likert scale, including a 'don't know' response. The five Likert scale categories were collapsed into three categories of 'agree' (strongly agree and agree), 'neutral' (neither agree nor disagree and don't know), 'disagree' (disagree and strongly disagree) to allow for multinomial regression analysis and interpretation. Preliminary analysis using diagnostic tests on ordinal regression with a 5-point Likert scale suggested poor model fit. It did not provide a meaningfully different story than the final analysis using the three categories that no longer violated model assumptions. It also allowed a direct comparison to those that agree and disagree.
Additionally, there are ongoing debates on if 'don't know' responses can be treated as a neutral response, especially if respondents distinctly choose midpoint responses differently than 'don't know' (see Sturgis et al. 2014). In light of no clear consensus in the literature, we used robustness checks to determine any differences between models excluding 'don't know' from the analysis compared to combining the group with 'neither agree or disagree' into a neutral category, which produced no notable differences. The decision was to present the latter model due to the 'don't know' responses capturing a modest minority of the sample (7.2 per cent).
Four measures commonly used to represent economic self-interests were household income, educational attainment, employment activity, mobility, and homeownership. Household income originally had 38 ordinal income responses recoded into three categories for parsimony. Low household income includes household earning £0 to 14,559, middle household income captures £14,460 to 36,399, and high household income is earning £36,400 and higher (based on average household income quintiles for 2014/2015 for the bottom, 2nd, and 3rd, and 4th and top, respectively, (see ONS 2018). A missing flag was created for those who reported they did not know or refused to respond (not uncommon for income measures). Highest attained educational level was recoded into four categories. Degree level or higher; higher education, or equivalent was combined into 'Higher degree'; A-level and GCSE or equivalents were combined into 'Secondary degree'; No qualifications consisted of those with less than a secondary degree; Other qualifications or level unknown were combined into an 'Other category.' Employment was measured as either has employment or not actively employed. A missing flag variable identified the small number of nonresponse. A measure was included to capture geographic mobility, indicating whether the respondent had lived outside of NI (yes ¼ 1; no ¼ 0).
Social exposure was examined in two dimensions to capture frequency and quality of contact: direct interaction and stated preferences that can influence outgroup contact (Pettigrew 1998). Direct social exposure consisted of Muslim contact frequency, avoided ethnic contact in shop/work, used different shop/service, and had no minority friends. Muslim contact asks about the frequency of direct contact. Willingness to accept refugees in Europe is lowered by anti-Muslimism sentiments (Bansak et al. 2016). Syria is a majority Muslim country, estimated at over 90 per cent in 2010 (Pew Research Center 2015); therefore, it remains a reasonable measure for identifying an outgroup social exposure to Muslims that would likely impact views on Syrian refugees. Due to the small sample size, daily, once or twice a week, and once or twice a month were recoded into 'at least monthly response,' as well as maintaining original categories of 'very rarely,' 'not at all,' and 'don't know.' A skip pattern excluded five Muslim respondents, which were recoded into the highest frequency of contact category.
To capture ethnic contact, the following question: 'Have you ever deliberately avoided contact with someone from a minority ethnic or different country to your own in a shop or at work?' was recoded into 'Yes' (combining 'Yes, because I was worried there might be language difficulties' and 'Yes, because of other reasons'); 'No'; and 'Cannot choose.' Used different shop/service derives from the question 'Ever chose to use a different shop or service because it was used more by people of another religious background.' Responses were recoded into 'Yes' ('Yes definitely' and 'Yes, probably') and 'No' ('Probably not' and 'Definitely not') for parsimony and preliminary evidence showing no variation on the outcome by the level of support. Simultaneously, a small amount of 'don't know' responses (n ¼ 8) was excluded from the analysis. This measure is likely drawing on sectarian attitudes on Catholic-Protestant contact, which provides information on the degree of ingroup self-segregation that can provide insight into how sectarianism and broader outgroup exposure may relate to views on Syrians. Having minority friends was recorded from a series of questions asking if the respondent was friends with any Black, Chinese, South Asian, Irish Traveller, Portuguese, Polish, Bulgarian, Romanian, Other Eastern European, Filipino, or Other individuals. Respondents who reported none of these were coded as 'No minority friends' than 'Yes minority friends' reporting any of the above groups as friends. Although Syrians were not explicitly mentioned, there is evidence that outgroup contact can lower prejudices towards other out-groups not involved in that contact, called the secondary transfer effect (Pettigrew 2009).
Stated preferences of social exposure included a created index variable that counted the number of situations a respondent would not be comfortable with a Muslim person ranging from 0 (comfortable in all situations) to 6 (not comfortable in all situations) with a Cronbach's alpha score of .906. As a NI tourist, NI resident (living and working), resident in the local area, work colleague, close friend, and married to a family member. Five Muslims were again excluded from the question and recoded as 0. Respondents were also asked, 'In relation to colour and ethnicity, I prefer to stick to people of my own kind.' The five-category Likert scale from strongly agree to strongly disagree was collapsed for parsimony into agree, neither agree nor disagree (including 12 don't know responses), and disagree. Ones missing were excluded from the analysis.
To measure cultural marginality and ethnic competition, respecting the culture, local political power, and ethnopolitical identity was included in the models. Feeling if culture was respected by society originally had a five-point Likert scale from strongly agree to strongly disagree and a 'don't know' response, collapsing into 'agree,' 'neither agree nor disagree' (includes 22 'don't know' responses) and 'disagree' with four missing responses dropped. The second variable measured if respondents felt they had any influence when it comes to any of the local decisions made that was collapsed into three categories due to cell size and preliminary analysis showing no difference in the outcome between combined groups: 'Yes' ('Yes, definitely' and 'Yes, probably'); 'No' ('No, Probably not' and 'Definitely not'); and 'Don't know. ' For anonymity purposes, the publically available NILT dataset grouped religious responses into three categories: Catholic, Protestant, and no religion. Other religions were recorded within missing responses, which for this study were recoded into a fourth other/missing religion category. Religion has a robust ethnopolitical identity salience in NI that captures distinct ethnic communities rather than strictly religious affiliation (Mitchell 2013). For robustness checks, we also include supplementary analysis that uses alternate ethnopolitical identity measures: political (nationalist; unionist; neither; other or don't know) and nationality (more Irish, equally Irish and British, other or missing). These identities all have substantial overlaps that cause multicollinearity issues within the same model; the additional findings are similar to the main findings and can be seen in Table 3. Lastly, gender and age were included as control variables as prior research has established these can relate to outgroup attitudes (Dempster and Hargrave 2017).

Analysis
The research uses multinomial logistic regression to examine the relative risk ratio of the independent measures on the likelihood of disapproving of Syrian refugees versus approving Syrian refugees and neutral towards Syrian refugees versus approving of Syrian refugees. All data are weighted. Table 1 presents the percentage of distributions and means of respondent characteristics. Household income was mostly evenly spread with around 20 per cent reporting in each across the three income categories, with 40 per cent nonresponse, which, although high, is not uncommon in survey responses to income. Almost half the respondents had completed the secondary degree although almost a third had a higher degree, and 19 per cent had no qualifications. Slightly more than half of the respondents were used, and 30 per cent had lived outside of NI. There was a low frequency of social interaction with Muslims, with only 19 per cent reporting daily or monthly contact than 41 per cent reporting no contact. A minority of respondents would purposely avoid contact with others in public spaces; 15 per cent stated that having avoided someone in a shop or work due to being an ethnic minority, and 11 per cent reported they would definitely or probably use a different shop or service if people from another religious background frequented it. When it comes to the quality of contact, slightly over half of the respondents reported not having minority friends. Also, the mean number of incidences that respondents would not accept Muslims was 2.85, close to the index midpoint (0-5). Over half of the respondents disagreed that they would prefer to be with their kind compared to about a quarter for both agree and neutral.

Descriptive Statistics
When considering cultural marginality measures, over two-thirds agreed that their culture was respected compared to 14 per cent that disagreed. Slightly over a quarter of respondents felt empowered to make local decisions compared to 69 per cent, not feeling they had power. In terms of religious identification and how identity politics matter in this context, Catholics represent 39 per cent of the sample compared to 41 per cent identifying as Protestant and 16 per cent reporting no religion. This distribution closely resembles distributions reported in the 2011 census (NISRA 2014a).

Multivariate Analysis
Table 2 presents multinomial logistic regressions of relative risks for disapproving or feeling neutral towards accepting Syrian refugees versus approving towards accepting Syrian refugees. The first column shows the risks of disapproving versus approving Syrian refugees. Among the economic self-interest measures, being in the lowest household income bracket (£0 to 14,559) compared to the middle household income bracket (14,399) has almost twice the likelihood of disproving versus approving Syrian refugees. Education attainment, employment, and living outside NI do not reach statistically significant differences in the full model. This finding partially supports that low income may spark negative attitudes towards outgroups (i.e. Syrian refugees) who may need the same resources, such as jobs and low-income housing.
Among the direct social exposure (capturing specific behaviours over preferences and the quality of interactions) measures, having no minority friends had a higher likelihood of disapproving over approving Syrian refugees. The other direct social exposure measures (Muslim contact, avoided contact with an ethnic minority in shop/work, and used different shop/service to avoid another religious group) did not reach statistically significant differences. More evidence was found among preference for social exposure; the more incidences that a respondent 14 Cameron D. Lippard and Cate McNamee would not accept, Muslims are associated with a higher likelihood of disapproving than approving Syrian refugees (relative risk, 1.7). Those who reported preferring to hang out with their own kind were almost three times more likely to disapprove versus approve of Syrian refugees. Like other researchers, low frequency and quality of contacts (i.e. having minority friends) increased anti-Syrian refugee settlement attitudes (Croucher et al. 2014).
Regarding the cultural marginality and ethnic threat measures, neither feelings towards culture being respected or feeling if you had power over local decisions reached statistically significant differences. However, when considering religious affiliation as a possible indicator of marginality and ethnic threat, Protestants had 3.5 times higher risk of disapproving than Catholics.
The second column presents the relative risk ratios of feeling neutral compared to the approval of Syrian refugees. Those with no qualifications had a higher risk of neutral versus approval than those with a secondary degree. None of the other economic self-interest variables reach statistically significant differences. Among the social exposure direct measures, those who have avoided contact with an ethnic minority in a shop or at work had higher risks of feeling neutral than those who had not avoided contact. Also, having no minority friends associated with a higher likelihood of feeling neutral over approval. Among the social exposure preference measures, the number of situations where Muslims are not accepted is positively associated with feeling neutral compared to approving Syrian refugees (relative risk, 1.2) whereas prefer own kind remains insignificant. For the cultural marginality measures, those that felt neutral about the statement of their culture being respected were more likely to feel neutral over agree with Syrian refugees compared to those that agreed their culture was respected. The local decision was not significant in the model. Lastly, compared to Catholics, Protestants are almost three times as likely to feel neutral, whereas those that are other or missing were three times as likely to feel neutral than approve.
To further tease out the impacts of identity politics as a proxy for ethnic threat/ competition, Table 3 presents a supplemental multinomial logistic regression analysis of how NI alternative ethnicity indicators influence attitudes towards Syrian refugees. As noted in Table 3, 'Alternate 1' identity examining the differences between a 'nationalist' versus 'unionist' identity suggests a significant difference in that those NI respondents who identified as Unionists (i.e. those who believe NI should be a part of the UK) disapproved of Syrian refugee resettlement in . Also, the 'Alternate 2' identity found that those who identified as 'More British' disapproved more than those who identified as 'More Irish.' Combined with the results in Table 2, these results suggest that sectarian and political identities are crucial in explaining attitudes towards Syrian refugees. Moreover, being Protestant, Unionist, and 'More British' are significant indicators in explaining anti-refugee attitudes.
Overall, this study adds to the emerging literature focusing on refugee settlements and social integration into various European communities. The findings for examining disagreeing versus agreeing are broadly similar to neutral versus agreeing, suggesting similar processes are at play. Thus, those disagreeing and those neutral share some of the same characteristics and attitudes that distinguish them from those agreeing. These results provide partial support for economic selfinterests, social exposure (contact), and cultural marginality and ethnic competition theories.

Discussion and Conclusions
This study investigated attitudes towards Syrian refugees in NI through the lens of immigrant/foreign-born outgroup theories: economic self-interest, social exposure, cultural marginality, and ethnic threat/competition. The findings suggest that all these theories may explain views towards Syrians in NI, which requires a nuanced understanding of the context of NI people's attitudes towards this group. Northern Ireland attitudes towards Syrian refugees 17 In particular, having lower socioeconomic status, having no minority friends, having a dis-preference towards Muslims, holding prejudice, Protestant, Unionist, and More British identities, and older ages are associated with less welcoming attitudes towards Syrian refugees. Like other research, these theories and associated variables help explain attitudes towards Syrian refugee resettlement in NI and create a complicated explanation of native-born British and Irish citizen views of foreign-born immigrants. This research agrees with other researchers that NI provides a unique context due to a post-conflict society's cultural and structural implications. How does a society that continues to be homogeneously white and divided based on religious and nationalist positions respond to the recent arrival of a foreign non-white outgroup? Thus, we examined if racism, sectarianism, or both (represented by measures of self-reported prejudice and various identities) have influenced NI attitudes towards Syrian refugee resettlement.
Our findings point to both. For instance, measures that captured racist attitudes such as social situations where the respondent would not accept Muslims were related to their negative views of Syrians, conflating these two groups as at least similar, matching other European and UK findings. Additionally, results suggest that sectarianism continues to matter in NI. Protestants hold less favourable views towards Syrians than Catholics. Those who self-identify as Unionists or More British tend to disapprove of Syrian refugee resettlement compared to Nationalist and More Irish identities. These findings point to the majority group (Protestant, Unionist, and More British) being more resistant towards an outside group as they might be perceived as a threat to maintaining a dominant sectarian identity. This finding supports Blumer's (1958) claims that prejudice will increase if a dominant group feels threatened by ethnic minority outgroups. Alternatively, as argued by other scholars (see Higham 2000;Jaret 1999;Lippard 2011), racist nativism increases among native-born and citizen populations that feel or perceive a severe threat to their well-established and privileged group positions. Another explanation could be related to pre-existing community divides on the Israeli-Palestinian conflict, which in NI has been viewed as a parallel struggle but was not measured in the 2015 survey. The division in support reflects the majority and minority groups (Protestants supporting Israel and Catholics supporting Palestine), which can be seen played out in murals and flags throughout NI that reinforce group solidarity (Arar 2017). Therefore, these group differences in support for the majority Muslim Syrian population might mirror pre-existing sides taken towards the majority Muslim -Palestine conflict, but more granular measurement is needed.
Considering the findings from this study, cultural marginality and ethnic competition emerged as a promising avenue of study in understanding how ethnoidentities shape attitudes towards Syrian refugees and other foreign-born groups. However, to establish a more precise explanation of how NI attitudes evolve with growth in diversity, further research to capture the nuances of how the complexity of NI identities shapes these attitudes would be fruitful. Notably, researchers should interview NI citizens to determine whether they feel any threat to their livelihoods or culture with the arrival of groups different from the centuries-old Northern Irish demography. Additionally, the data used in the study was collected in 2015 at the start of the European refugee 'crisis'; studies have found that European attitudes shifted from welcoming to feeling threatened (e.g. Dempsey and McDowell 2019). In the NI context, Brexit, which is tied to views on migrants, has intensified ethno-identity debates and identification due to the potential impacts on causing a hard border or Irish unification (Hayward and Rosher 2020). It will also be critical to consider the consequences of the COVIDpandemic on views towards migrants going forward. Therefore, the overall and group difference in the views towards 'others' may have intensified in this rapidly changing climate. We also assert that as direct measures on the quality of intragroup relationships were outside the scope of the study dataset (i.e. no clear measures that suggest the quality of social contact), then more research has to collect in-depth information from both groups and particularly, the newcomers and their treatment in NI for decades to come. These future studies will better inform and develop social policies that help integrate new populations arriving in NI by knowing the social, economic, and political barriers they may face.