deltie.net archive

Extracted abstract

Many municipalities in the United States have experienced conflict over immigration, but the intensity of these conflicts varies. It is important to understand why one place experiences more conflict than another, because highly intense conflicts can have big impacts on local responses to immigration. This paper contributes to the immigration literature by constructing a measure of conflict intensity to assess what factors affect the intensity of local conflicts over immigration. A dataset on local conflict over day laborers in California in the period 1985-2020 is employed to assess how conflicts vary by time and space. This paper argues that demographics and politics do not play a major role when it comes to conflict intensity, while contact mediated by time and space is a decisive factor. Initial contact can boost conflict intensity as it may magnify differences and breed feelings of fear. However, endured encounters provide the opportunity to develop connections and allow communities to get familiar with immigrant day laborers.

Immigration Relief Act implemented by the city encountered protest from various advocacy organizations (Steil and Ridgley 2012) . Although each of the mentioned cases involves intense local conflicts over immigration, this is not to say that all communities that deal with immigration experience highly intense conflicts.

Conflicts are dynamic and their intensity varies (Ramsbotham et al. 2011; Schattschneider 1957 ). The intensity of a conflict, for instance, is affected by the number of actors involved and whether the involved actors engage in hostile behavior. When the number of actors that is involved in a conflict increases, conflict intensity may grow as new participants add new interests and demands to the existing conflict. In addition, the display of hostile behavior may strain mutual relationships thereby heightening conflict intensity (Ramsbotham et al. 2011) . Conflict intensity is important because highly intense conflicts oftentimes result in delicate political affairs that precipitate policy intervention, while less intense conflicts are more likely to recede into the background and have fewer political and policy implications (Schattschneider 1957) . Thus, the intensity of conflict matters and should be considered when studying conflict.

Scholars exploring local responses to immigration have found that the settlement of immigrants is characterized by conflict, but they have not examined the intensity of these conflicts in a generalizable way. Some large-scale quantitative studies have identified factors that help to explain the direction of local responses to immigration. These factors include growth of the Latino population, the proportion of homeowners, voter preference, and saliency of immigration in public debate (Hopkins 2010; Ramakrishnan and Wong 2010; Visser and Simpson 2019; Walker and Leitner 2011) . Others have conducted case studies of local conflict over immigration and have found contrasting results when it comes to the impact of contact on conflict. Some have argued that conflict may dwindle due to contact between established residents and immigrants, while others have found that close proximity to immigrants may spur conflict (Amin 2002; Hondagneu-Sotelo and Pastor 2021; Valentine 2008; Valentine and McDonald 2004; Wilson 2017) . While the literature has provided us with important insights into local conflict over immigration, it largely ignores the impact of variation in conflict intensity. It is important to examine conflict intensity for two reasons. First, conflict intensity varies and a highly intense conflict may have a much bigger impact on local policy responses than a less intense conflict. To understand local responses to immigration it is imperative to assess why one place experiences more conflict than another. Second, it is essential to acknowledge that conflict is not static, but dynamic. It is not simply a matter of conflict versus no conflict, as conflicts can vary in their intensity. A conflict may start small, intensify, and cede over time. Thus, the goal of this paper is twofold: first, to trace conflict trajectories to assess differences in conflict intensity, and second, to identify the factors shaping these conflict trajectories to understand what is driving variation in conflict intensity across space and time.

The immigration literature lacks a metric to assess conflict intensity, and this paper addresses that gap by turning to the literature on conservation conflict to develop a measure of conflict intensity. Cusack et al. (2021) have explored conflict over the conservation of biodiversity; this model forms the foundation for studying the intensity of local conflict over immigration. The model of Cusack et al. (2021) is applicable to the case of immigration, because conflicts over both immigration and biodiversity conservation revolve around actors competing for contrasting goals and in both cases actors may engage in conflict behavior to reach their goals. By constructing a quantifiable measure of conflict intensity it becomes possible to assess the intensity of conflict over immigration across space and time, and to understand what factors are responsible for conflict.

This paper focuses on immigrant day laborers as a case of local conflict over immigration. Day laborers use informal and formal hiring sites in public spaces to solicit for jobs, and these sites can ignite conflict when other community members experience nuisance from these sites or believe that sites are an inappropriate use of public space (Crotty 2020; Crotty and Bosco 2008; Sandercock 2003; Valentine 2008 ). The presumed immigration status of day laborers oftentimes plays an important role in local conflicts (Crotty 2017; Esbenshade 2000; Valenzuela 1999; Varsanyi 2008) , making local conflict over day laborers a suitable case in which to study the intensity of local conflict over immigration. A dataset on day laborers in California in the period 1985-2020 is employed to explore the intensity of conflicts. The dataset includes data on day laborer hiring sites, conflict over day laborers, and demographic and political characteristics. Ordinal multilevel regression models are fitted to the data to study factors affecting the intensity of conflicts.

Local Immigration Conflict • 91

LOCAL CONFLICT OVER IMMIGR ATION

The impacts of immigration on communities -such as costs of integrating newcomers and the legitimacy of unauthorized immigrants (Ellis 2006; Rodriguez 2008 ) -can ignite conflict. A conflict is made up of three elements: incompatibility, actors, and action (Bartos and Wehr 2002; Wallensteen 2019) . First, incompatibility refers to disagreement, in the sense that involved actors strive for goals that cannot be met at the same time or by the same resources-the goals are incompatible (Bartos and Wehr 2002; Ramsbotham et al. 2011; Wallensteen 2019) . Second, actors refers to the individuals or groups that strive for incompatible goals. Finally, action refers to the behavior displayed by actors to achieve their incompatible goals and to behavior that is deployed to express hostility (Bartos and Wehr 2002; Ramsbotham et al. 2011; Wallensteen 2019) . Thus, local conflict can be defined as a situation in which at least two actors compete with each other to achieve incompatible goals and/or express hostility by means of conflict behavior (Bartos and Wehr 2002; Wallensteen 2019) .

The growth of the day laborer population since the late 1980s has sparked conflict in several places throughout the United States. A significant share of this population consists of Latino immigrants (Crotty 2014 (Crotty , 2015;; Esbenshade 2000; Valenzuela 1999; Valenzuela, Kawachi, and Marr 2002) , which has led to immigration playing an important role in many local conflicts. Day laborers use hiring sites, a location where day laborers congregate and solicit for jobs (Crotty 2015 (Crotty , 2018)) . Hiring sites are often found in the vicinity of businesses that provide goods to potential employers, such as homeimprovement stores, lumber yards and paint stores. In addition, hiring sites are also found near busy streets and other locations that are convenient for both potential employers and day laborers (Crotty 2014; Valenzuela 1999; Valenzuela and Melendez 2003) . Day laborers who solicit for jobs in public spaces are highly visible to the local community, and this can be a source of friction, as community members, day laborers, and others raise questions about who should be allowed to use public space and how this public space should be used (Crotty 2020; Crotty and Bosco 2008; Sandercock 2003; Valentine 2008) .

Factors Impacting Local Conflict

Scholars studying local impacts of immigration have identified multiple factors that affect how community members respond to the impacts of immigration. One of these factors is contact between different groups. Differences can be based on race, age, income, and religion, among others (Valentine 2008; Valentine and McDonald 2004) . Proponents of the contact hypothesis have argued that unfamiliarity between groups can breed hostility as people feel uncertain and threatened by the unknown other. For instance, fears about the impacts of immigration on the local and national identity can be enhanced among community members when they have little experience with immigrants (Gilbert 2009; Walker 2014) . Hostility can then be reduced by contact between different groups, because contact raises familiarity, which makes people less fearful and uncertain (Hondagneu-Sotelo and Pastor 2021; Valentine 2008; Valentine and McDonald 2004) . However, the contact hypothesis has been criticized because contact in itself would not bridge hostility between groups. On the contrary, close proximity and contact between different groups can fuel frustration and resentment, as it exacerbates differences (Amin 2002; Valentine 2008; Valentine and McDonald 2004; Wilson 2017) . In this sense, contact between different groups in public spaces -such as between day laborers soliciting for work and other community members -can heighten hostility because contact is superficial and emphasizes how "we" behave versus how "they" behave (Valentine and McDonald 2004) .

The criticism on the contact hypothesis aligns with the findings of researchers exploring ethnic competition. It has been argued that when boundaries between ethnically or racially different groups start to crumble, ethnic conflict can be spurred (Olzak 1992) . This, for instance, is the case when growth of a minority population is accompanied by economic decline, leading to increased competition over a shrinking pool of resources between the growing population of newcomers and the established population (Koopmans and Olzak 2004; Olzak and Shanahan 1996) . The same can be said when a minority population experiences prosperity, which results in increased competition with the members of the established population that find themselves just above the minority population (Olzak 1992; Olzak and Shanahan 1996) . Some scholars have argued that contact between different groups should be considered in relation to time and proximity of conflict. Time matters when it comes to contact, because continued contact over a longer period of time can alleviate hostility as casual contact may grow into a connection (Hondagneu-Sotelo and Pastor 2021; Wilson 2017). For instance, in the case of immigrant day laborers, Crotty (2018) has found that enduring the existence of a hiring site results in a decline of conflict as community members get used to its presence. In addition to the passage of time, proximity of conflict matters as well. Group dynamics can spur hostility, because persons who express hostility and prejudice can bolster the behavior and attitudes of like-minded others (Valentine and McDonald 2004) . This suggests that hostility and conflict can spread to areas outside the original location of antagonism. Hostility can be reinforced when different groups come into contact while conflict exists nearby, such as in adjacent neighborhoods or cities. Thus, contact matters when it comes to local conflict over immigration, but contact should be considered in relation to time and proximity of conflict.

Other factors that have found to impact local conflict over immigration are demographics, politics, characteristics of place, and issue saliency. Demographics, such as a rapid growth of the Latino population, a higher population of home-owners, and a lower level of educational attainment have been identified as spurring hostility towards immigrants (Visser and Simpson 2019; Walker and Leitner 2011; Wong 2012) . Political preferences -such as predominantly voting for Republican candidates -has been connected to more hostile responses to immigrants as well (Huang and Liu 2018; Ramakrishnan and Wong 2010; Walker and Leitner 2011) . Other factors impacting local conflict over immigration are place-based characteristics, and, in the case of day laborers, the location of hiring sites plays an important role. Businesses in the vicinity of day laborer hiring sites complain about the loss of customers, customers report feeling unsafe, and residents argue that day laborer activities decrease their property values and quality of life (Castaneda and Beck 2017; Crotty 2015 Crotty , 2017;; Esbenshade 2000) . This suggests that a hiring site in a mostly remote area may generate little conflict, as the activities of day laborers do not interfere with the activities of other community members. Lastly, saliency of immigration may provide local actors with frames to define immigrants as a threat, thereby spurring hostility towards immigrants (Hopkins 2010) .

Thus, the literature has identified multiple important variables that impact local conflict over immigration and it has suggested that factors can vary across space and time. However, the literature does not provide us with a measure to assess conflict across space and time in a generalizable manner. It is important to study conflict over time and space, because conflicts are dynamic and their intensity may vary (Ramsbotham et al. 2011) . The variation in intensity may impact the political significance of a conflict and may also affect the likelihood of local authorities to intervene (Schattschneider 1957) . The intensity of conflict matters, but currently no standard metric is available in the immigration literature to assess conflict intensity. This paper addresses that gap by turning to the literature on biodiversity conservation, to develop a measure of conflict intensity.

Conflict Intensity

Scholars studying the escalation and de-escalation of conflict over conservation of biodiversity have constructed a measure to assess conflict intensity that consists of six stages: 1) coexistence or collaboration, 2) latent disagreement, 3) expressed disagreement, 4) unilateral action, 5) multilateral action, and 6) violence (Cusack et al. 2021 ). This model (Cusack et al. 2021 ) forms a solid groundwork to measure the intensity of local conflict over immigration, because both conflict over biodiversity conservation and immigration include multiple stakeholders who may compete for incompatible goals and may engage in conflict behavior to achieve them.

The measure of conflict intensity deployed in this paper consists of five stages instead of six, because the available data sources do not allow for exploring "latent disagreement" -a stage in which actors express disagreement within their private spheres, but not in public discourse. The model focuses on conflict over day laborers -a case of local conflict over immigration -and consists of five stages (see Table 1 ). The first stage -coexistence and collaboration -is characterized by an absence of competition between involved actors. At this stage, no conflict is observed in the public realm. The second stage -disagreement -features the occurrence of conflict in public discourse. At this stage conflict does not surpass the expression of disagreement in public discourse. The third stage -unilateral action -is characterized by the involvement of a single actor (group) in non-violent conflict behavior. This

Local Immigration Conflict • 93 type of behavior includes demonstrations, lobbying, and implementing policies, among others. When multiple actor (groups) get involved in non-violent conflict behavior the fourth stage -multilateral action -is reached. Finally, the fifth stage -violence -features violent conflict behavior.

In sum, this paper develops a standard metric to assess the intensity of local conflicts over immigration, to assess how local conflict varies across space and time. The immigration literature has identified several important variables that impact local conflict over immigration -such as contact, demographics, politics, place-based characteristics, and saliency -but lacks a measure to explore why conflict intensity varies between places and over time. Scholars studying conflict have found that the intensity of conflict matters, as conflicts are dynamic and their intensity varies. The literature on conflict analysis provides tools to develop a measure of conflict intensity for local conflict over immigration. This metric is employed to assess why one place experiences more intense conflicts over immigration than another.

METHODS

The data used for the analyses in this paper has been derived from a largescale dataset on day laborers in California between 1985 and 2020. This dataset has been built based on a list of all places in California that were included in the decennial Census of 2010. The list includes 1,523 places in California. For each of the places included in this list, extensive online searches have been performed to gather information on the existence of day laborer hiring sites and on the existence of conflict over day laborers.

Four online sources have been deployed to obtain data on the existence of conflict over day laborers. The sources include newspaper articles, city council archives, studies, and Google searches. The search keys "day laborer(s)" and "day worker(s)" have been combined with the names of each of the 1,523 places identified above to search the archives of Newsbank for relevant newspaper articles. Then, for all places the availability of online city council archives has been checked. When online archives were available, the archives have been searched for documents in which day laborers were discussed. In addition, Google searches have been performed using the same key word combinations that were used for the newspaper searches. These searches produced newspaper articles not included in the databases of Newsbank, scholarly articles, and reports by advocacy organizations.

When articles or reports referred to inactive websites, these links have been searched in the archives of the Wayback Machine. For instance, the Wayback Machine has been deployed to access old and inactive versions of city council websites and websites from activist groups. All relevant results have been coded for the existence of conflict over day laborers (no conflict = 0, conflict = 1). As mentioned previously, conflict is defined as a condition in which at least two actors compete with each other to attain incompatible goals and/or express animosity through conflict behavior (Bartos and Wehr 2002; Wallensteen 2019) . For each of the instances in which conflict over day laborers was identified to be present, the intensity of conflict was coded according to the day labor conflict intensity measure (see Table 1 ). This means that instances where no signs of conflict are present, are coded as "coexistence or collaboration. " When disagreement between day laborers and another actor(s) is apparent in public discourse -such as in newspaper articles or in city council meetings -conflict intensity is coded as "disagreement. " As soon as actions that target day laborers surpass disagreement in public discourse and involve one actor or actor group, conflict intensity is coded as "unilateral action. " When more than one actor or actor group engages in actions that target day laborers and these actions surpass public discourse, conflict intensity is coded as "multilateral action. " Finally, when actions targeting day laborers include violence, conflict intensity is coded as "violence. "

The set of sources mentioned in the previous paragraph has been supplemented with Google Streetview searches to identify day laborer hiring sites (see Table 2 ). The use of Google Streetview as a data source in the field of urban studies has become more common in recent years (for instance, see Hwang and Sampson 2014; Wang and Vermeulen 2020) . It is likely that sources such as newspaper articles and studies on day laborers include more hiring sites that have experienced conflict than sites that have not experienced conflict. Google Streetview, on the other hand, provides the opportunity to explore areas that have the potential of being used by day laborers as a hiring site. Such areas include home improvement stores, lumber yards, rental facilities and paint stores. These are all locations that have a connection to the informal day labor market, since day laborers are often employed in construction and landscaping (Crotty 2014; Valenzuela 1999; Valenzuela and Melendez 2003) . Thus, Google Streetview allows to sample sites based on the potential of being a space in which day laborers solicit for jobs, thereby reducing the bias of solely selecting sites that have experienced conflict. This is important because this study aims to explore conflict intensity, and to explore variation in conflict intensity, the dataset should not be sampled based on the dependent variable of conflict (McAdam and Boudet 2012) .

Google Streetview has been deployed to explore areas in all 1,523 Californian places that have the potential to be used by day laborers as hiring sites. These areas include Home Depot, Lowe's, paint stores, lumber yards, and rental facilities (such as U-Haul). Each area has been examined to determine if day laborers were visible in these areas. Day laborers are defined as men who stand around in small or larger groups, while wearing working clothes and/or backpacks. It has been taken into consideration that people may gather for other activities (such as to wait for public transport), and those instances have not been included into the dataset. In all instances where day laborers were identified in an area that has the potential to be used as hiring site, the site was added to the dataset. The number of images available through Google Streetview varies by location. All images that were available on Google Streetview for potential hiring sites have been checked for the presence of day laborers. This has made it possible to track day laborer hiring sites over time. As the availability of images on Google Streetview is limited by time -started from 2008 onwards -and place, other sources have been deployed as well to monitor hiring sites over time. These sources include newspaper articles, reports, and scholarly articles.

All identified hiring sites have been added to a dataset including for each site the location, the type of site, the years in which a site was active, and a short description of the location of the hiring site. Whether a hiring site was actively used by day laborers to find work in a certain year is determined by a combination of newspaper articles, studies, and Google Streetview. When day laborers were visibly present at a hiring site in a certain year and/or newspaper articles or studies described activity at the site in that specific year, the site was coded as active. Although the approach chosen for this study to gather data online cannot match the ethnographic depth of fieldwork, it does allow for capturing the breadth and variety of day laborer hiring sites in California.

This paper focuses on the intensity of conflict over day laborers, and zooms in on active day laborer hiring sites in California. This means that the dataset employed for the statistical analyses includes a row of data for each year a hiring site was determined to be actively used by day laborers. A total of 432 hiring sites in California have been identified that are located in 179 places and 26 counties. Counties and places can accommodate multiple day laborer hiring sites. The number of years sites have been active varies, and this has resulted in a total of 3,675 observations. The unit of analysis is year(s), which means that if a hiring site experienced multiple conflicts in the same year, the highest level of conflict intensity that can be observed in a certain year is included in the dataset.

Multilevel Generalized Linear Model

The data employed for the statistical analyses in this paper include repeated measures and clusters, which consist of repeated measurements over time and clusters of hiring sites that are located in places and counties. To account for the clusters and repeated measures, a multilevel generalized linear model has been fitted to the data (Finch, Bolin, and Kelley 2019) . The model includes an ordinal dependent variable -intensity of day laborer conflict -that consists of five levels (1 = coexistence or collaboration; 2 = disagreement; 3 = unilateral action; 4 = multilateral action; 5 = violence). To explore whether the dependent variable is really ordinal -i.e., that the levels of conflict intensity are orderedmultilevel multinomial models have been fitted to the data as well. The results of the multilevel multinomial analyses are quite similar to the ordinal analyses, which suggests that the dependent variable can be treated as ordinal. This means that ordinal models with the proportional odds assumption have been conducted to analyze the data. The output of the multilevel multinomial analyses can be found in the appendix. 1 The independent variables employed in this study can be separated into three categories. The first category refers to demographic characteristics and political preferences of the area in which a hiring site is located. Demographic variables have been obtained from the decennial Census and the American Community Survey and refer to the Census Tract level. Many hiring sites are located at the intersection of roads, which means that depending on the side of the intersection that is selected a different Census Tract is found. To obtain demographic data that reflect the area in which a hiring site is located, a buffer of 250 meters is created around each hiring site. For each hiring site it has been identified which Census Tract(s) intersect the buffer of 250 meters. Subsequently, all demographic variables represent the mean of all Census Tracts that are found within a 250-meter radius of a hiring site. A buffer of 250 meters has been chosen because day laborer hiring sites are a very local phenomenon.

Responses to immigration do not have to be uniform across a city, as one neighborhood may be more hostile or friendly than another (Furuseth, Smith, and McDaniel 2015) . Nuances between different neighborhoods would be overlooked when demographic data on the city level is used. For instance, a city might house multiple day laborer hiring sites that are located in neighborhoods with different characteristics. To assess if variation in neighborhood characteristics -such as household income or level of unemployment -impact conflict intensity, it is necessary to employ data measured at the neighborhood level.

The demographic variables consist of the population that is Latino, the growth of the Latino population, the percent of owner-occupied housing, median household income, and the percent of the population that is unemployed. The median household income is measured as an ordinal variable that consists of three categories. The U.S. median household income has been calculated for each year that data have been obtained from the decennial Censuses and the American Community Survey. Middleincome is calculated as two third to double the U.S. median household income. All incomes that fall below the middle-income threshold are defined as low-income, while all incomes that are above the upper threshold of middle-incomes are defined as high-income. The political preferences of community members (percentage voting Republican at the County level) is included as well. Unfortunately, these data were not available at the tract level, and are therefore measured at the County level.

The second category of independent variables includes characteristics of day laborer hiring sites. These variables include the type of hiring site -unmanaged (0) versus managed (1), the type of zoning of the parcel on which a hiring site is located -residential (1), commercial (2), industrial (3), or other (4), and the distance in meters to the nearest day laborer hiring site. In addition, it has been determined if businesses (no = 0 versus yes = 1) and homes (no = 0 versus yes = 1) are located within a buffer of 250 meters around a hiring site. The third category refers to variables that are measured beyond the local scale. These include the presence of day laborer hiring sites and conflict within a radius of ten kilometers (no site and no conflict = 1, site and no conflict = 2, or site and conflict = 3), and saliency of immigration in California. The presence of other hiring sites and conflict is measured within a 10-kilometer radius of a specific hiring site. The 10-kilometer radius allows assessing whether other sites exist within close proximity and if these sites experience conflict and how these factors impact conflict intensity. It is possible that sites and or conflict in another part of a city or a neighboring city may spread to other places, and this variable allows exploring whether such spill-over effects exist. Saliency of immigration is measured according to the approach used by Hopkins (2010) . This means that the appearance of the topics "immigration" and "immigrants" in the Los Angeles Times and all California newspapers included in the Newsbank archives have been counted for each year in the period 1985 to 2020. The saliency measure is then calculated by dividing the number of stories published in a certain newspaper in each year by the average number of stories that the newspaper published on these topics between 1985 and 2020. The outcomes of each newspaper are then summed up for each year to get saliency of immigration by year.

FINDINGS

In this section the results of the ordinal multilevel regression models are discussed. The models examine what factors impact the intensity of conflicts over day laborers. When the day laborer hiring sites are categorized according to the highest level of conflict intensity that they have experienced, we see that 180 hiring sites reached conflict intensity level one (coexistence or collaboration); 49 hiring sites entered level two of conflict intensity (disagreement); 174 hiring sites reached level three of conflict intensity (unilateral action); 21 hiring sites entered level four of conflict intensity (multilateral action); and 8 hiring sites reached level five of conflict intensity (violence).

An empty model and three models to which independent variables were added have been fitted to the data (see Table 3 ). The empty model does not include any independent variables. Model 1 contains demographic and political factors to assess to what extent demographic and political characteristics affect conflict intensity. Model 2 adds characteristics of the hiring site and features of the neighborhood in which a site is located to explore to what degree spatial characteristics of hiring sites impact conflict intensity. Lastly, Model 3 adds a variable on the existence of sites and conflict within a 10-kilometer radius and saliency of immigration in California to explore how the visibility of day ** 0.41 0.87 * 0.39 2.54 *** 0.48 public disagreement| unilateral action 2.30 *** 0.09 1.76 *** 0.41 1.41 *** 0.40 3.19 *** 0.48 unilateral action| multilateral action 4.91 *** 0.18 4.46 *** 0.44 4.13 *** 0.43 6.23 *** 0.52 multilateral action| violence 6.45 *** 0.36 6.00 *** 0.54 5.68 *** 0.53 7.79 *** 0.60 Random Effects σ 2 3.29 3.29 3.29 3.29 Between group variance (τ 00 ): τ 00 Sites: in Places: in Counties 0.13 0.01 0.00 0.17 τ 00 Places: in Counties 0.25 0.32 0.27 0.30 τ 00 Counties 0.00 0.02 0.00 0.00 Intraclass Correlation Coefficient (ICC): ICC Sites: in Places: in Counties 0.04 0.00 0.00 0.05 ICC Places: in Counties 0.07 0.10 0.08 0.09 ICC Counties 0.00 0.01 0.00 0.00 Observations 3675 3675 3675 3675 Marginal R 2 / Conditional R 2 0.000 / NA 0.199 / 0.275 0.252 / NA 0.383 / NA * p<0.05 ** p<0.01 *** p<0.001 Table 3. Continued Local Immigration Conflict • 99

laborers and conflict within close proximity and saliency of immigration in California affect conflict intensity.

The empty model does not include any factors that can help to explain the intensity of conflict over day laborers. However, the intraclass correlation coefficients (ICC) can provide insight into the amount of variance that is explained by the cluster structures (Finch et al. 2019 ). The ICC is.04 at the sites nested within places and counties level, suggesting that 4 percent of the total variance is accounted for by hiring sites. At the places nested in the counties level, the ICC is.07, indicating that 7 percent of the total variance can be explained by places. At the county level the ICC is.00, suggesting that no variance is accounted for by counties. Thus, the ICC scores show that intensity of conflict scores correlate among hiring sites and among places, but not among counties. As clustering at the hiring site and place level impacts the intensity of conflict, it is useful to conduct multilevel modeling to account for clustering.

The first model contains demographic and political variables, to assess to what extent local demographics and political preferences matter when it comes to the intensity of conflict over day laborers. The model suggests that differences in race and class do not significantly affect the intensity of conflict over day laborers. The percent of the population that is Latino, the growth of the Latino population and the median household income are not significantly impacting the intensity of conflict in any of the three models; nor does the share of homeowners in an area surrounding a hiring site significantly impact conflict intensity. The same can be said about unemployment and political preferences. The percentage of the population that is unemployed and the share of community members that vote for a Republican governor does not significantly affect conflict intensity in any of the three models. These findings suggest that demographic and political characteristics are not of overriding importance when it comes to the intensity of conflict over day laborers.

As was mentioned previously, multilevel multinomial analyses have been performed as well (see the appendix). It could be the case that demographic and political factors affect specific levels of conflict intensity. The multilevel multinomial models show that some demographic and political factors are significantly impacting conflict intensity when comparing specific levels of conflict intensity. However, the effect sizes of these significant variables are marginal. The effect sizes range from 0 to 1.5 percent for the predictor percent Latino population, from 0 to 1.8 percent for having a lower median household income versus a middle median household income, from 0 to 0.7 percent for the percent unemployed, and from 0 to 0.9 percent for the percentage voting Republican. Thus, the multilevel multinomial models confirm the findings of the ordinal models that demographic and political characteristics do not have a decisive impact on conflict intensity.

The second model adds characteristics of hiring sites and characteristics of the neighborhood in which hiring sites are located as predictors. The first predictor is the type of hiring site, indicating whether a hiring site is unmanaged or managed. The model indicates that the type of hiring site does not have a significant effect on the level of conflict intensity. The zoning of the parcel on which a hiring site is located does not significantly affect the level of conflict intensity. This finding indicates that a hiring site's location, whether it is located on a residential, industrial, or other type of parcel compared to a commercial parcel, does not impact the intensity of conflict over day laborers. Model two suggests that the distance from one day laborer site to another significantly reduces the intensity of conflict. This finding indicates that when the distance between a hiring site and the nearest hiring site is larger, the more likely it is that the intensity of conflict over day laborers is reduced. This finding is in line with critics of the contact theory who have argued that close proximity can heighten conflict (Amin 2002; Valentine 2008; Valentine and McDonald 2004) . However, the effect of distance to the nearest hiring site on conflict intensity is modest. For instance, the predicted probabilities indicate that when the nearest hiring site is located at 24 kilometers instead of 1 kilometer, the likelihood of finding conflict intensity level three (unilateral action) decreases from 13 percent to 10 percent. In addition, the effect does not seem to be very robust, as the distance to the nearest hiring site is not found to be significant in model three.

Model two shows that conflict intensity is significantly reduced when the number of years a site exists increases. For instance, predicted probabilities show that when a site exists 1.4 years the chances of finding conflict intensity level three are 15 percent. When a hiring site exists for 15.7 years, the likelihood of finding conflict intensity level three has been reduced to 8 percent. The number of years a site exists significantly impacts the intensity of conflict in model three as well. These findings suggest that time matters when it comes to contact, as endured contact can lower conflict intensity.

Two characteristics of the neighborhood that have been included into model two are the existence of businesses and/or homes within a 250-meter radius of a hiring site. The existence of businesses within close proximity of a hiring site does not significantly impact the level of conflict intensity over day laborers in the ordinal models. However, in some of the multinomial models (see appendix) the existence of businesses within 250 meters of a hiring site does significantly impact the level of conflict intensity. For instance, when there are no businesses within 250 meters of a hiring site, predicted probabilities indicate that the likelihood of finding unilateral action (level three) instead of coexistence (level one) increases from 4 to 11 percent compared to when there are businesses present within 250 meters of a hiring site. The effect of businesses is limited to moving from coexistence to unilateral action and from public disagreement to unilateral action. This finding suggests that conflict intensity is more likely to reach level three of conflict intensity (unilateral action) when there are no businesses located within close proximity of a hiring site. The multilevel generalized linear models were not able to capture this variation, as these analyses do not allow examining specific levels of conflict intensity.

Model two indicates that the absence of homes within a 250-meter radius of a hiring site significantly reduces the intensity of conflict. For instance, the predicted probabilities show that the likelihood of finding level one of conflict intensity (coexistence or collaboration) reduces from 83 percent when there are homes present to 77 percent when there are no homes present. On the other hand, predicted probabilities indicate that the chances of finding conflict intensity level three (unilateral action) increases from 10 percent when there are no homes present to 14 percent when there are homes present within the vicinity of a day laborer hiring site. In addition, model three shows a significant positive relationship between the presence of homes and the intensity of conflict as well. However, it must be noted that none of the multinomial models showed a significant impact of the presence of homes within 250 meters of a hiring site. Thus, there seems to be some effect from the existence of homes around hiring sites on the intensity of conflict, but the impact is absent in the multinomial analyses.

Model three adds two more variables to explore how the existence of other sites, conflict and saliency affect conflict intensity. The variable on sites and conflict within ten kilometers explores if other day laborer hiring sites can be found within a 10-kilometer radius and if conflict occurred over these sites. The variable refers to sites and or conflict existing in the same year, to explore if proximity of sites and/or conflict affects conflict intensity. The results of model three indicate that having no site and no conflict within a 10-kilometer radius of a hiring site increases conflict intensity compared to having one or more sites that do not experience conflict within 10 kilometers. The predicted probabilities illustrate that the likelihood of a higher level of conflict intensity increase substantially when other hiring sites that experience conflict are found in close proximity (see, Figure 1 ).

For instance, predicted probabilities show that experiencing a conflict intensity level of three (unilateral action) increases from 4 percent when there are sites and no conflict within 10 kilometers to 10 percent when there are no sites and no conflict within 10 kilometers. This finding is in line with the contact theory, which suggests that contact between newcomers and longtime residents can reduce conflict. On the other hand, the existence of hiring sites that experience conflict within 10-kilometer radius compared to having sites without conflict, do significantly increase the level of conflict intensity. For instance, predicted probabilities indicate that the likelihood of finding conflict intensity level three (unilateral action) when there are sites without conflict is 4 percent, and this increases to 28 percent when there are hiring sites present within a 10-kilometer radius that experience conflict over day laborers. Thus, conflict over day laborer hiring sites may spread to adjacent hiring sites. This finding indicates that proximity of conflict may spur the intensity of a conflict, which supports the argument that contact itself is not enough to reduce conflict as conflict interacts with other factors such as time and proximity of conflict (Amin 2002; Hondagneu-Sotelo and Pastor 2021; Valentine 2008; Valentine and McDonald 2004) . Lastly, the impact of saliency of immigration in California on conflict intensity is examined. Saliency of immigration in California does significantly increase the intensity of conflict over day laborers. For example, predicted probabilities indicate that when the saliency of immigration score is 2.7, the likelihood of finding conflict intensity level three (unilateral action) is 12 percent. This percentage increases to 16 percent when the saliency of immigration score is 7.9.

Local Immigration Conflict • 101 Downloaded from https://academic.oup.com/socpro/article/73/1/90/7801216 by Universiteit van Amsterdam user on 08 April 2026

DISCUSSION

Scholars studying local conflict over immigration have identified multiple factors that impact local conflict. For instance, Crotty and Bosco (2008) have argued that differences in race and class are likely to increase conflict. Based on this argument it can be expected that when a hiring site is located in an affluent area that is predominantly populated by non-Latino residents, finding conflict there is more likely. This is expected because the population in these areas differs to a great extent from the day laborer population that to a large extent consists of Latino immigrants who have little money to spend (Valenzuela 1999; Valenzuela et al. 2002) . In addition, researchers have argued that a higher percentage of home-owners increases the likelihood of hostile responses towards immigrants (McCabe 2016; Mitchel 1997; Stuart, Armenta, and Osborne 2015; Walker and Leitner 2011) , and that a preference for Republican candidates increases hostility towards immigrants as well (Huang and Liu 2018; Ramkrishnan and Wong 2010; Walker and Leitner 2011 ). An economic downturn, on the other hand, would foster more friendly responses to day laborers (Crotty 2014 ). However, the findings of this study suggest that demographics and politics do not have a decisive impact on the intensity of conflict over day laborers. Differences in race and class, the share of home-owners, the median household income, the rate of unemployment, and political preferences of community members do not significantly impact conflict intensity.

In addition, some characteristics of hiring sites, such as the type of hiring site and the zoning of the parcel on which a hiring site is located, also fail to explain variation in conflict intensity. The finding that the type of hiring site (unmanaged versus managed) has no significant impact on conflict intensity is in line with the literature, in the sense that the literature is inconclusive. Some scholars argue that managed hiring sites may alleviate concerns of unmanaged hiring sites, while others argue that managed hiring sites contribute to the level of conflict intensity as opponents frame these sites as unlawful support to unauthorized immigrants (Castaneda and Beck 2017; Crotty 2015 , 2017; Crotty and Bosco 2008; Esbenshade 2000; Rodriguez 2008 ). 2 10 • M. de Wilde

An important factor impacting conflict intensity turns out to be contact mediated through time and space. The analyses indicate that matters such as the distance between hiring sites, the number of years sites have been active, the existence of nearby conflict, and saliency of immigration significantly affect the intensity of conflict over day laborers. When hiring sites are located in closer proximity to each other, the likelihood of conflict is higher. This finding is in line with critics of the contact hypothesis who have argued that contact or familiarity with immigrants in itself is not sufficient to reduce hostility and conflict (Amin 2002; Valentine 2008; Valentine and McDonald 2004) . The mere presence of more than one day laborer hiring site in close proximity does not lower conflict intensity. On the contrary, encountering multiple sites within a relatively small distance may even aggravate community members. However, the findings also suggest that contact over time may alleviate tensions and reduce conflict intensity. When a hiring site exists for a longer period of time, it may integrate into the community (Crotty 2018) , and endured contact between day laborers and community members provides the opportunity to develop connections (Hondagneu-Sotelo and Pastor 2021). Lastly, the analyses indicate that conflict intensity is influenced by the type of contact in close proximity. If nearby places experience conflict over day laborers, the likelihood of finding a higher level of conflict intensity increases. This finding is in line with the argument of Valentine and McDonald (2004) that when hostility is expressed by other persons in close proximity of someone, this may fuel that person's hostility. Nearby conflicts over day laborers may spur the level of conflict intensity as actors are provoked by the hostile behavior in nearby places. In this sense, saliency of immigration is found to be fueling conflict intensity as well, as it may provide actors with frames to express their disagreement in public about day laborer activities (Hopkins 2010) .

Taken together it can be argued that demographics and politics do not have a major impact on the intensity of conflict over day laborers. This is an important finding for immigration scholars to take into consideration, because demographic and political factors play a prominent role in the immigration literature. This is not to say that demographics and politics do not matter, but that researchers have to take into account that other factors have a more decisive impact on the intensity of local conflict. This study shows that the intensity of conflict over day laborers is to a large extent driven by contact mediated through time and space. When day laborers encounter other community members in public spaces -where day laborers solicit for work while others use the same public spaces for different purposes -proximity of difference and unfamiliarity can breed hostility and heighten conflict intensity. However, contact between day laborers and community members can also alleviate tensions over time and space. Endured contact enables the formation of connections and the existence of other hiring sites that do not experience conflict nearby may make a hiring site seem less threatening.

This paper has aimed to show that local conflicts are dynamic and that it is important to take the intensity of conflicts into account. The paper contributes to the immigration literature along four dimensions. First, the paper emphasizes that more attention should be paid to intensity of local conflicts instead of treating conflict as a factor that is either present or absent. Many municipalities in the United States have experienced conflict over immigrants in general or day laborers in particular, but the intensity of these conflicts may vary widely. The measure of conflict intensity constructed in this paper can be employed to compare conflict across localities and explore why one place experiences more conflict than another. In addition, the measure allows an examination into how different levels of conflict intensity may have different impacts on local policies. For instance, it can be expected that conflicts that never go beyond public disagreement may have less impact on local policies than conflicts that reach the stage of multilateral action. Further research is necessary to explore how variation in conflict intensity affects local policy responses.

Second, this paper adds to the immigration literature by deploying a quantitative measure of conflict intensity which makes it possible to study intensity of conflicts over immigrants in a generalizable manner. The analyses in this paper have provided insight into how different factors affect the intensity of conflict over day laborers. The findings suggest that demographic and political factors do not have a substantial impact on the intensity of conflict. This is not to say that demographics and politics do not matter, but that other factors may play a more important role in explaining the intensity of local conflict over day laborers.

Third, the paper contributes to immigration literature by demonstrating that the interaction of contact with other factors should be incorporated into the contact hypothesis. The results suggest that Local Immigration Conflict • 103 contact -considering time and space -significantly impacts conflict intensity. On one hand, contact between day laborers and other community members may exacerbate differences and heighten conflict intensity. On the other hand, enduring contact and the existence of other hiring sites nearby that do not experience conflict may help to alleviate tensions and reduce conflict intensity. These findings support the argument of Hondagneu-Sotelo and Pastor (2021) that a revised version of the contact hypothesis that considers how contact interacts with other factors, can be very valuable. So, a key takeaway from this study for immigration scholars is that when it comes to conflict over immigrants it is not just contact that matters, but I is the interplay between contact and other factors that plays an important role.

The fourth contribution to immigration literature is providing a metric to assess the intensity of local conflicts, to incite scholars studying local conflicts to explore conflict intensity. The metric employed here allows scholars to explore the intensity of local conflicts over immigrants, but can also be deployed to study conflicts over other groups. Local conflicts are an inevitable part of everyday life as communities struggle with ways to accommodate immigrants, the homeless, and other marginalized groups. To get a better understanding of how local conflicts evolve and the role of different factors in these conflicts it is important to pay attention to the dynamic character of conflict. Conflicts vary in intensity and this variation in intensity may have important consequences when it comes to local responses and policies.

References

1. Ethnicity and the Multicultural City: Living with Diversity Ash Amin 10.1068/a3537 Environment and Planning A: Economy and Space Environ Plan A 0308-518X 1472-3409 34 6 2002 SAGE Publications
2. Using Conflict Theory Otomar J Bartos Paul Wehr 10.1017/cbo9780511613692 Using Conflict Theory New York Cambridge University Press 2002
3. Right to the Suburb? Rethinking Lefebvre and Immigrant Activism Genevieve Carpio Clara Irazabal Laura Pulido 10.1111/j.1467-9906.2010.00535.x Journal of Urban Affairs 33 2 2011
4. Stigmatizing Immigrant Day Labor: Boundary-making and the Built Environment in Long Island, New York Ernesto Castaneda Kevin R Beck 10.4324/9781315100371-4 in Immigration and Categorical Inequality, Migration to the City and the Birth of Race and Ethnicity E Castaneda New York Routledge 2017
5. The Social Geography of Day Labor: Informal Responses to the Economic Downturn Sean M Crotty 10.1353/pcg.2014.0011 Yearbook of the Association of Pacific Coast Geographers 76 1 2014
6. Locating day-labor employment: toward a geographic understanding of day-labor hiring site locations in the San Diego metropolitan area Sean M Crotty 10.1080/02723638.2015.1049481 Urban Geography Urban Geography 0272-3638 1938-2847 36 7 2015 Informa UK Limited
7. Can the Informal Economy Be “Managed”?: Comparing Approaches and Effectiveness of Day‐Labor Management Policies in the San Diego Metropolitan Area Sean M Crotty 10.1111/grow.12180 Growth and Change Growth and Change 0017-4815 1468-2257 48 4 2017 Wiley
8. Strategic Visibility and the Production of Day-Labor Spaces: A Case Study from the San Diego Metropolitan Area Sean M Crotty 10.1177/1078087416659767 Urban Affairs Review 54 3 2018
9. Origins of Informality: Examining the Historical and Spatial Roots of Informal Day-Labor Hiring Sites Sean M Crotty 10.1007/978-3-030-32120-8_8 Temporary Appropriation in Cities, Human Spatialisation in Public Spaces and Community Resilience Alessandro Melis Jose Antonio Lara-Hernandez James Thompson Cham Springer Nature Switzerland 2020
10. Racial geographies and the challenges of day labor formalization: a case study from San Diego County Sean M Crotty Fernando J Bosco 10.1080/08873630802433830 Journal of Cultural Geography Journal of Cultural Geography 0887-3631 1940-6320 25 3 2008 Informa UK Limited
11. Measuring the intensity of conflicts in conservation Jeremy J Cusack Tom Bradfer‐lawrence Zachary Baynham‐herd Sofia Castelló Y Tickell Isla Duporge Håvard Hegre Lara Moreno Zárate Vincent Naude Sahil Nijhawan John Wilson Dario Gerardo Zambrano Cortes Nils Bunnefeld 10.1111/conl.12783 Conservation Letters CONSERVATION LETTERS 1755-263X 14 3 2021 Wiley
12. UNSETTLING IMMIGRANT GEOGRAPHIES: US IMMIGRATION AND THE POLITICS OF SCALE Mark Ellis 10.1111/j.1467-9663.2006.00495.x Tijdschrift voor Economische en Sociale Geografie Tijd voor Econ & Soc Geog 0040-747X 1467-9663 97 1 2006 Wiley
13. The “Crisis” over Day Labor Jill Esbenshade 10.1111/j.1743-4580.2000.00027.x-i2 WorkingUSA Journal of Labor and Society 1089-7011 1743-4580 3 6 2000 Wiley
14. Multilevel Modeling Using R W Holmes Finch Jocelyn E Bolin Ken Kelley 10.1201/9781351062268 Multilevel Modeling Using R New York Chapman and Hall/CRC 2019
15. The Logic of Institutional Interdependency: The Case of Day Laborer Policy in Suburbia Lorrie A Frasure Michael Jones-Correa 10.1177/1078087409355640 Urban Affairs Review Urban Affairs Review 1078-0874 1552-8332 45 4 2010 SAGE Publications
16. Belonging in charlotte: multiscalar differences in local immigration politics and policies Owen Furuseth Heather Smith Paul Mcdaniel 10.1111/j.1931-0846.2014.12048.x Geographical Review Geographical Review 0016-7428 1931-0846 105 1 2015 Informa UK Limited
17. Immigration as Local Politics: Re-bordering Immigration and Multiculturalism through Deterrence and Incapacitation Liette Gilbert 10.1111/j.1468-2427.2009.00838.x International Journal of Urban and Regional Research 33 1 2009
18. South Central Dreams. Finding Home and Building Community Pierrette Hondagneu-Sotelo Manuel Pastor 10.18574/nyu/9781479804023.001.0001 18574/nyu/9781479804023.001.0001 2021 New York University Press South L.A. New York
19. Politicized Places: Explaining Where and When Immigrants Provoke Local Opposition Daniel J Hopkins 10.1017/S0003055409990360 American Political Science Review 104 1 2010
20. Welcoming Cities: Immigration Policy at the Local Government Level Xi Huang Cathy Yang Liu 10.1177/1078087416678999 Urban Affairs Review 54 1 2018
21. Divergent Pathways of Gentrification Jackelyn Hwang Robert J Sampson 10.1177/0003122414535774 American Sociological Review Am Sociol Rev 0003-1224 1939-8271 79 4 2014 SAGE Publications
22. Discursive Opportunities and the Evolution of Right‐Wing Violence in Germany Ruud Koopmans Susan Olzak 10.1086/386271 American Journal of Sociology American Journal of Sociology 0002-9602 1537-5390 110 1 2004 University of Chicago Press
23. Putting Social Movements in Their Place Doug Mcadam Hilary Schaffer Boudet 10.1017/CBO9781139105811 Explaining Opposition to Energy Projects in the United States, 2000-2005 New York Cambridge University Press 2012
24. Building Community? Homeownership and the Politics of Exclusion Brian J Mccabe 10.1093/acprof:oso/9780190270452.001.0001 No Place Like Home: Wealth, Community, and the Politics of Homeownership Brian J Mccabe New York Oxford University Press 2016
25. The Annihilation of Space by Law: The Roots and Implications of Anti-Homeless Laws in the United States Don Mitchell 10.1111/1467-8330.00048 Antipode 29 3 1997
26. The Dynamics of Ethnic Competition and Conflict Susan Olzak 1992 Stanford University Press Stanford, CA
27. Deprivation and Race Riots: An Extension of Spilerman's Analysis Susan Olzak Suzanne Shanahan 10.2307/2580387 Social Forces 74 3 1996
28. Partisanship, Not Spanish: Explaining Municipal Ordinances Affecting Undocumented Immigrants S Ramakrishnan Tom Tak Karthick Wong 10.1017/S002238161200076X Taking Local Control: Immigration Policy Activism in U.S. Cities and States Monica Varsanyi Stanford, CA Stanford University Press 2010
29. Oliver Ramsbotham Tom Woodhouse Hugh Miall Contemporary Conflict Resolution. The Prevention, Management and Transformation of Deadly Conflicts Cambridge, UK Polity Press 2011
30. The Significance of the Local in Immigration Regulation Cristina M Rodriguez Michigan Law Review 106 4 2008
31. Cosmopolis II. Mongrel Cities of the 21st Century Leonie Sandercock 10.5040/9781472545527 2003 London
32. Intensity, Visibility, Direction and Scope Elmer E Schattschneider 10.2307/1952444 The American Political Science Review 51 4 1957
33. Small-town Defenders': The Production of Citizenship and Belonging in Hazleton, Pennsylvania Justin Steil Jennifer Peter Ridgley 10.1068/d0109 Environment and Planning D: Society and Space 30 6 2012
34. Legal Control of Marginal Groups Forrest Stuart Amada Armenta Melissa Osborne 10.1146/annurev-lawsocsci-120814-121433 Annual Review of Law and Social Science 11 2015
35. Living with difference: reflections on geographies of encounter Gill Valentine 10.1177/0309133308089372 Progress in Human Geography Progress in Human Geography 0309-1325 1477-0288 32 3 2008 SAGE Publications
36. Gill Valentine Ian Mcdonald Understanding Prejudice. Attitudes towards Minorities London Stonewall 2004
37. Day Laborers in Southern California. Preliminary Findings from the Day Labor Survey. Center for the Study of Urban Poverty Abel Valenzuela 1999 Los Angeles
38. Seeking Work Daily: Supply, Demand, and Spatial Dimensions of Day Labor in Two Global Cities Valenzuela Janette A Abel Matthew D Kawachi Marr 10.1177/002071520204300205 International Journal of Comparative Sociology 43 2 2002
39. Day Labor in New York: Findings from the NYDL Survey Abel Valenzuela Edwin Melendez Community Development Research Center 2003 Center for the Study of Urban Poverty New York and Los Angeles
40. Immigration Policing through the Backdoor: City Ordinances, the 'Right to the City,' and the Exclusion of Undocumented Day Laborers Monica W Varsanyi 10.2747/0272-3638.29.1.29 Urban Geography 29 1 2008
41. Determinants of County Migrant Regularization Policymaking in the United States: Understanding Temporal and Spatial Realities M Visser Sheryl-Ann Anne Simpson 10.1177/0308518X18797134 Economy and Space 51 1 2019
42. Immigration, Local Policy, and National Identity in the Suburban United States Kyle E Walker 10.1080/02723638.2014.890423 Urban Geography 35 4 2014
43. The Variegated Landscape of Local Immigration Policies in the United States Kyle E Walker Helga Leitner 10.2747/0272-3638.32.2.156 Urban Geography 32 2 2011
44. Understanding Conflict Resolution Peter Wallensteen 2019 SAGE Publications London
45. Life between Buildings from a Street View Image: What Do Big Data Analytics Reveal about Neighbourhood Organizational Vitality? Mingshu Wang Floris Vermeulen 10.1177/0042098020957198 Urban Studies 58 15 2020
46. On Geography and Encounter: Bodies, Borders, and Difference Helen F Wilson 10.1177/0309132516645958 Progress in Human Geography 41 4 2017
47. 287(g) and the Politics of Interior Immigration Control in the United States: Explaining Local Cooperation with Federal Immigration Authorities Tom K Wong 10.1080/1369183x.2012.667983 Journal of Ethnic and Migration Studies Journal of Ethnic and Migration Studies 1369-183X 1469-9451 38 5 2012 Informa UK Limited