ATTITUDES TOWARDS IMMIGRATION: ECONOMIC VERSUS CULTURAL DETERMINANTS. EVIDENCE FROM THE 2011 TRANSATLANTIC TRENDS IMMIGRATION DATA

ATTITUDES TOWARDS IMMIGRATION: ECONOMIC VERSUS CULTURAL DETERMINANTS. EVIDENCE FROM THE 2011 TRANSATLANTIC TRENDS IMMIGRATION DATA A Thesis submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Master of Public Policy in Public Policy By Elif Can, M.A. Washington, DC March 25, 2013

Copyright 2013 by Elif Can All Rights Reserved ii

ATTITUDES TOWARDS IMMIGRATION: ECONOMIC VERSUS CULTURAL DETERMINANTS. EVIDENCE FROM THE 2011 TRANSATLANTIC TRENDS IMMIGRATION DATA Elif Can, M.A. Thesis Advisor: Yuriy Pylypchuk, Ph. D. ABSTRACT Immigration is an issue that is always on the political agenda. Today we see that the United States and European countries are reassessing immigration policies based on changed realities and needs of their countries. However, policy makers are not always in-sync with public opinion. Surveys show that the public generally is less concerned about the economic impact of immigration but care more about the cultural impact of immigration. The predominant narrative in political discussion and electoral campaigns, however, seems to be increasingly focused on the effects immigrants might have on the labor market and the welfare system. This paper analyzes individual attitudes towards immigration based on the 2011 Transatlantic Trends Immigration data for the United States, Germany, the United Kingdom, France, Spain and Italy. The results indicate that economic factors are not the most important predictors of public attitudes towards immigration. iii

I would not be here without the love and support of my inspiring parents who migrated from Turkey to Belgium had 4 daughters one of whom migrated to the United States wrote this thesis and is loved by a Mexican-American. My love to all. iv

TABLE OF CONTENTS 1. Introduction...1 2. Literature review...2 3. Conceptual framework and hypotheses......6 4. Data description and summary statistics...8 5. Analysis plan...16 6. Results...17 7. Limitations of the study...29 8. Policy implications and conclusion...30 Bibliography...32 v

1. INTRODUCTION The economic crisis hit hard across the European Union and the United States. At the same time, in recent years there has been a growth in anti-immigrant sentiment across the globe (Wilson and Hainsworth, 2012). Greece is such an example where the neo-nazi party, Golden Dawn entered parliament for the first time with 21 seats in May 2012 (Papapostolou, 2012). The financial crisis seems to have increased the focus on immigration. Mid 2012, the economic crisis led to a border crisis where France and Germany started to rethink the principle of freedom of movement within the European Union by asking to temporarily secure the internal borders. Currently, there is also increased funding for the European Union External Borders Fund to toughen border security management ( 1 ). During the 2012 American presidential elections, immigration reform was a crucial issue addressed by both sides that ultimately might have determined the results. Current debates and initiatives on immigration reform in the European Union and the United States show that immigration and the economy are one of the more important issues today. Since immigration and its link to the economy is a hot political topic in Europe and the United States, it is interesting to see if the public has the same beliefs. Several surveys show that the public is concerned about immigration and the economy. However, the connection between these two issues is not as clear. Although immigration proves to benefit a country and immigrants are needed for the good of the economy, there is a great focus on the negative aspects of immigration. One of those negative aspects is the belief that 1 http://ec.europa.eu/dgs/home-affairs/financing/fundings/migration-asylum-borders/external-bordersfund/index_en.htm 1

immigrants take jobs from natives (Espenshade and Hempstead, 1996; Novak, 2010). The Take Our Jobs campaign by the American United Farm Workers accurately deals with the stereotype of undocumented workers taking American jobs. The campaign asks all Americans to work in the agricultural field to see firsthand the reality of the industry. As one can imagine, the campaign was successful in showing that jobs are not taken away from natives but rather filled in by much needed workers. Economic motivations seem to drive the political discourse regarding immigration. With this paper, I will look into the public beliefs regarding immigration. If the political discourse is in-sync with the public attitude, economic factors should also drive public beliefs. The goal of this paper is to see which factors determine attitudes towards immigration. 2. LITERATURE REVIEW There is extensive literature regarding the factors which influence international public opinion on immigration and immigration policy. Individual-level and country-level explanations are vast. Country-level explanations use aggregated data and usually place emphasis on data collected for a country or region e.g. GDP, unemployment rate or the size of the immigrant population. Individual explanations of attitudes towards international migration generally focus on three categories: demographics, economic factors and cultural factors that are related to perceptions on immigrants. Although literature agrees that attitudes are likely to be influenced by a combination of these individual-level factors, there is some disagreement on which factors play the most crucial role. Several studies have emphasized the role played by non-economic factors, 2

and in particular cultural factors (Citrin et al., 1997; Burns and Gimpel, 2000; Hainmueller and Hiscox, 2007). Other studies have found that economic factors can influence public attitudes on immigration (Kessler, 2001; Scheve and Slaughter, 2001; Mayda, 2006, Facchini and Mayda, 2012). I argue that economic factors are not the strongest predictors of public attitudes towards immigration. Economic controls refer to an assessment of the economic costs and benefits of immigration as a result of fears about labor market competition. Studies show that individuals with less economic security are more likely to have negative attitudes towards immigrants (Dustmann and Preston, 2006; O Rourke and Sinnott, 2006, Mayda, 2006). According to trade theory, highly-skilled individuals have a more positive view of immigrants because immigrants, who are generally low-skilled, do not compete directly with them in the labor-market (Wilson, 2001; Mayda, 2006; O'Connell, 2011). In tough economic times, people seem to be even more skeptic about immigrants and worries about the national economy indicate that people have a more negative outlook towards immigrants (Espenshade and Hempstead, 1996; Citrin et al., 1997; Burns and Gimpel, 2000; Fennelly and Federico, 2008). This theory does not seem unreasonable since resources are scarce in a society, so that people might feel economically threatened by immigrants that have the same skill set as them. However, the assessment of the economic costs and benefits of immigration encompasses more than just labor-market competition and also pertains to general welfare effects, taxes and public burden (Dustmann and Preston, 2006; Facchini and Mayda 2012). 3

The other perspective in literature is the emphasis on cultural controls and demographics. Various studies show that education, age, gender, residency, ideology and ethnocentric attitudes all have an independent effect on attitudes towards immigration. Cultural controls or determinants tap into beliefs about a national identity, cultural identity, habits and language. Perceived threats to the country's culture have a statistically significant and negative affect on attitudes towards migration (Chandler and Tsai, 2001). Research has shown that economic concerns are not the principal determinants of public attitudes on international immigration (Ford, 2011; Hainmueller and Hiscox, 2010; Sides and Citrin 2007). In accordance with the hypotheses of this paper, the labor market status of individuals do not show a statistically significant relationship with attitudes towards immigrants (Paas and Halapuu, 2012; Kehrberg, 2007; Fetzer, 2000). Culture is manifested when people agree on the importance of certain societal symbols although they can have different meanings to different people (Schildkraut, 2003). One symbol of culture is language. Schildkraut states that support for English as a symbol of the American national identity sometimes reveals anti-immigrant sentiments since not all immigrants speak English when they first enter the country. Several studies found that feelings of threat to one s cultural identity and general hostility toward ethnic others are all indicative of negative feelings towards immigration (Citrin et al., 1997; Sides and Citrin, 2007). O Rourke and Sinnott show that nationalist sentiments have a negative and highly significant effect on positive feelings towards immigration which indicates that those more likely to view cultural diversity as a benefit rather than a cost are more positive towards immigration and immigrants. 4

The most important demographic factor seems to be education and in several studies education is shown to be consistently significant (Citrin et al., 1997; Espenshade & Hempstead, 1996). Education however is an indicator of cultural capital and economic capital (Manevska, and Achterberg, 2011). As a result, the explanation as to why the more educated are more tolerant is twofold (Sides and Citrin, 2008). On the one hand, there is agreement that highly educated people are just more tolerant because of increased cultural capital which leads to an educated understanding and acceptance of cultural differences (Manevska, and Achterberg, 2011; Hainmueller and Hiscox, 2007). On the other hand, higher tolerance might follow from the fact that, in accordance with the trade theory, the highly educated do not compete for the same jobs as the mostly low-skilled immigrants (Mayda, 2006). Furthermore, when it comes to ideology it shouldn t be a surprise that more conservative individuals oppose permissive immigration policies and have a more negative attitude towards immigration (Mayda, 2006; Burns and Gimpel, 2000; Chandler and Tsai, 2001; Citrin et al., 1997; Espenshade and Hempstead, 1996; Scheve and Slaughter, 2001; Wilson, 2001). Studies show that age and gender have an effect on attitudes towards immigration but the effect is minor and sometimes it is insignificant (Scheve and Slaughter, 2001). Generally age and gender tend to be negatively correlated with support for immigration. Women tend to have more negative attitudes than men (Francois and Magni-Berton, 2013; Mayda, 2006; O Rourke & Sinnott, 2006). Age has been shown to have mostly a negative effect on tolerance towards immigration with the older generation being less positive about immigration (Ford, 2012; Mayda, 2006; Hainmueller and Hiscox, 2007; Dustmann and Preston, 2006; Citrin et al, 1997). 5

Lastly, individuals living in urban areas are more tolerant towards immigrants than those living in rural areas (Paas and Halapuu, 2012). Migrant heritage also plays a significant role in perceptions towards immigrants (Mayda, 2006; Scheve and Slaughter, 2001). Studies show that foreign-born individuals (Paas and Halapuu, 2012; Fennelly and Federico, 2008; Burns and Gimpel, 2000) and people with foreignborn parents (O Rourke and Sinnott, 2006) have more positive beliefs about immigration. My thesis is meant to complement the debate on which of these determinants plays the most important role in explaining attitudes towards immigration. 3. CONCEPTUAL FRAMEWORK AND HYPOTHESES In this paper I will analyze which determinants play a statistically significant role in shaping people's attitudes towards immigration. The focus of this paper will be individual level determinants. As literature suggests, there are a variety of individual level factors that can explain people's attitudes towards immigration and these can generally be divided into three categories: demographics, economic factors and cultural factors. Demographics and personal factors are generally considered to be education, age, gender, residency, ideology, migrant heritage, employment and income. Among the economic factors are fears and feelings about labor-market competition, satisfaction with the economy and personal finances, worries about the economy or unemployment, the perceived fiscal burden on welfare services but also the demographic factors of employment and income. Finally, cultural considerations are those that address people's perceptions regarding the cultural make-up of their country; the perceived 6

national identity, its habits and values and the respondent's preference for cultural homogeneity. It also encompasses any perceptions people might have about immigrants and the preference for cultural unity. For my analysis I will be using all of these categories of explanatory variables - demographic, economic and cultural - to try to explain what determines the current attitudes towards immigration. When mentioning attitudes toward immigrants this usually refers to the general perceived consequences of immigration and more specifically it refers to the desired level of immigrants or the preference for either restrictive or permissive immigration policies. However, for this paper I will focus on two distinct perceived consequences of immigration as dependent variables since I am interested to see if there are any differences in the magnitude and significance of the explanatory variables based on the perceived consequences my hypothesis is that there is no difference. The attitudes I will analyze are the perceived cultural and economic consequences of immigration. My hypotheses are that regardless of the perceived consequences of immigration - economic or cultural -, cultural and demographic explanatory variables will have a greater explanatory power than economic variables. In other words, economic factors will have no statistically significant effect on the perceived economic or cultural consequences of immigration after controlling for all other effects. Instead I expect that once controlled for education, ideology, immigrant heritage, age, gender, residency and employment, the cultural determinants will have an independent effect on the perception of the consequences of immigration 7

4. DATA DESCRIPTION AND SUMMARY STATISTICS Public attitudes on immigration are measured in various national and international surveys. Usually the questions regarding attitudes on immigration are part of more general social surveys that try to explain the interaction between values, beliefs and attitudes of diverse populations. The data that will be used for this paper is unique as it is one of the few surveys that solely focuses on public views on international immigration and integration and does not ask about other more general beliefs. For this reason, it is able to capture beliefs that other surveys cannot. The 2011 Transatlantic Trends Immigration Survey Data measures individual level public opinion on these issues on both sides of the Atlantic. The six countries that were included in the fourth year of the Transatlantic Trends Immigration data were the United Kingdom, France, Germany, Italy, Spain and the United States. The Transatlantic Trends Immigration is a project of the German Marshall Fund of the United States and its partners. A random sample of about 1000 adults per country were surveyed between August 25 and September 18, 2011 by means of computer-assisted telephone interviews as shown in table 1. Table 1: Sample overview Country N USA 1001 UK 1000 France 1000 Germany 1000 Italy 1000 Spain 1000 TOTAL 6001 8

4.1. DEPENDENT VARIABLES The key dependent variables are the questions Q18.1 and Q31.2 that ask the respondents about their agreement with two perceived consequences of immigration for their country: economic consequences and cultural consequences (Table 2). Table 2: Dependent variables Q18.1. Do you agree with the statement that immigrants take jobs away from native born? 1. Strongly agree 2. Somewhat agree 3. Somewhat disagree 4. Strongly disagree Q16. Some people think that immigration enriches (NATIONALITY) culture with new customs and ideas. Others think that these new customs and ideas negatively affect (NATIONALITY) culture. Which comes closer to your point of view? 1. Immigration enriches (NATIONALITY) culture 2. Immigration negatively affects (NATIONALITY) culture The first question is used as a measure of perceived economic consequences of immigration while the second question is used to measure perceived cultural consequences of immigration. These variables are recoded into the following dichotomous variables: Immigrants Take Jobs (Q18.1) and Cultural Effect Immigration (Q16) and will serve as key variables for the analysis of the attitudes towards immigration. When recoding, 0 was used to indicate a negative attitude towards immigration and denotes agreement with the statement that immigrants take jobs from natives and denotes the belief that immigration negatively affects culture. A 1 indicates a positive attitude towards migration. 9

4.2. INDEPENDENT VARIABLES The independent variables that I am most interested in are the economic security variables which are operationalized as the questions on the satisfaction with personal finances and the satisfaction with the economy. In order to see if my hypothesized effect of economic determinants holds, I include various other explanatory variables in the model which can be categorized in three groups: demographic, economic and cultural controls. DEMOGRAPHIC CONTROLS The first set of control variables are the demographic and personal variables that are provided by the survey data: age, gender, education, employment, ideology, residency, citizenship, foreign born, and foreign born parents. Age is a categorical variable that has 6 categories starting with the 18-24 category and then increasing in 10 year increments. Male is a dichotomous variable where 1 is male and 0 is female. The 7 categories of the variable Education were recoded into 3 categories: primary education, secondary education and post-secondary education. Employment is coded as a pair of dichotomous variables Employed and Unemployed (1= yes, 0 = other). Ideology was measured on a seven-point scale ranging from very left to very right and is recoded into: left (=1), center (=2), right (=3). The area people live in is no longer the categorical variable with 5 values but is instead recoded into the Urban dichotomous variable where 1 indicates a big city (includes the suburbs or outskirts of a big city) and 0 is a small city or a country village (includes a farm or home in the countryside). The next three variables pertain to the immigrant heritage of the respondent: Citizen, Foreign 10

Born, and Foreign Born Parents where 1 is coded as affirmative to the statement and 0 is negative. Table 3 provides an overview by country of the means of the demographic and personal control variables I will be using in the model. Table 3: Summary statistics of demographic controls by country - Means Variables Male Age Education Ideology Employed Unemployed Urban Citizen FB* FBP** Country USA.486 4.075 2.553 2.184.541.185.467.977.077.175 UK.408 4.173 2.480 2.053.540.135.357.980.086.150 France.436 4.157 2.277 1.854.473.102.283.978.062.170 Germany.482 3.902 2.183 1.785.626.204.388.953.103.194 Italy.455 3.964 2.019 1.848.488.166.348.986.044.041 Spain.478 3.692 2.147 1.942.529.225.370.967.062.063 *Foreign Born; *Foreign Born Parents ECONOMIC CONTROLS The variables in the Transatlantic Trends Immigration data that can best capture feelings of economic insecurity are those that measure the respondents' assessment of the financial situation of their own household and the nation's economy over the past year. The responses to the first two economic variables Satisfaction Personal Finances and Satisfaction Economy were recoded into a three-point scale where higher scores indicate that the economy or their personal financial situation is believed to have gotten worse over the past 12 months. Furthermore, ranking the economy or unemployment as the most important issue facing their nation today is also indicative of feelings of economic insecurity. Unemployment Most Important and Economy Most Important are two dichotomous variables where 1 indicates that the issue was ranked first among the 8 possible issues (of the others only immigration will be used as a control variable) and 0 indicates they it was not ranked as the most important issue. 11

CULTURAL CONTROLS AND IMMIGRANT PERCEPTION The last set of control variables are the cultural determinants that address the beliefs people might have regarding the cultural make-up of their country and its cultural norms, its cultural identity and the respondent's preference for cultural homogeneity. It reflects worries about perceived traits of immigrants and views on multiculturalism. The Transatlantic Trends Immigration survey has a myriad of variables that can be used to serve as controls to test my hypothesis. My selection attempts to address the potential for endogeneity bias and covers a range of cultural identity topics. The first variable is Immigrant Education Job where respondents were asked to indicate their policy preference on the type of immigrants entering their country where government should either give preference to immigrants who have a high level of education but no job offer, or to immigrants who have a job offer but a lower level of education. For this dichotomous variable 1 represents the preference for highly educated immigrants with no job offer while 0 reflects the opposite policy preference. The dichotomous variable Immigrant as Neighbor indicates whether respondents are comfortable with an immigrant as their neighbor (comfortable=1, uncomfortable=0). In the survey, respondents were asked which attribute was the most important precondition to become a citizen. Two of those answers were transformed into separate dichotomous variables Citizenship: Lived in country and Citizenship: Same Cultural values. The first variable refers to the precondition of having lived in the country for most of one's life while the second one refers to the precondition of sharing the nation s cultural values. For each of these variables regarding preconditions to citizenship, 0 indicates agreement with the importance of the precondition and 1 indicates disagreement. The ranking of 12

immigration as the most important issue facing the nation today is coded into the dichotomous variable Immigration Most Important where 1 indicates that the issue was ranked first among the 8 possibly issues and 0 indicates they it was not ranked as the most important issue. The last three cultural determinants are probably the most important as I predict that these variables will have a statistically significant relationship with the dependent variable Immigrants Take Jobs. Respondents were asked if they thought most of the immigrants in their respective countries were legal or illegal. This question was recoded as the dichotomous variable Most Immigrant Legal where 1 reflects agreement with the statement that most immigrants are legal and 0 indicates that most immigrants are illegal. The variable Immigrants should act like native was recoded from a four-point scale of agreement into a dichotomous variable where 0 indicates agreement and 1 indicates disagreement with the statement that people who come to a country should try to act like people from that country. Lastly, the variable Cultural Effect Immigration was described above and will be used as an independent variable as well as a dependent variable to demonstrate that regardless of the fact that the dependent variable pertains to economic consequences (Immigrants Take Jobs) or cultural consequences (Cultural Effect Immigration), economic determinants play a less important role when trying to explain public attitudes towards immigration and immigrants. Table 4 gives an overview of the most important variables and the dependent variables with their means and t-statistic testing whether there is a statistical significant difference between the mean score of that country for that particular variable and the mean score of the other countries for that variable. 13

Table 4: Average attitude regarding immigrants Variables Immigrants Take jobs Cultural effect Immigration Should act like native Most Immigrants Legal Country Mean T stat Mean T stat Mean T stat Mean T stat USA.431 15.173 *** UK.415 16.211 *** France.783-10.464 *** Germany.813-12.671 *** Italy.740-7.328 ***.627 1.195.454-13.312 ***.479 11.589 ***.682-2.607 **.732-6.186 ***.693-3.323 ***.390-8.251 ***.127 11.869 ***.305-1.749 * Spain.645-0.4812.652-0.484.146 10.251 *** Notes: Significant at the ***1, **5, *10 percent level..431 9.044 ***.630-4.012 ***.709-9.254 ***.876-22.218 ***.266 1.231.277 19.703 ***.455 7.271 *** For Spain the mean scores for the variables Immigrants Take Jobs and Cultural Effect Immigration are not statistically significantly different than the mean scores for the other countries. For the variable Immigrants should act like native there is no statistically significant difference between the mean score for Italy and the other countries. The mean score for Cultural Effect Immigration for the United States is not statistically significantly different than the other countries. An overview of all the variables of the Transatlantic Trends Immigration survey that will be used in the analysis model and their descriptive statistics are presented in table 5. 14

Table 5: Variable description and summary statistics for pooled sample Variables Description Coding Immigrants Take Jobs Do you agree with the statement that 0: Agree immigrants take jobs away from 1: Disagree native born? Demographic and economic controls Ideology In politics, people sometimes talk of 1: Left left and right. Where would you place 2: Center yourself? 3: Right Citizen Are you a citizen of country? 1: Yes 0: No Foreign Born Where you born in country or in 1: Foreign Born another country? Foreign Born Parents Which of these correspond to where your parents were born? 0: Born in country 1: One or both parents foreign born 0: Both parents born in country Male 1: Male 0: Female Age Age respondent (categorical) 1: 18-24 2: 25-34 3: 35-44 4: 45-54 5: 55-64 6: 65+ Education Employed Unemployed Urban Satisfied Personal Finances Satisfied Economy Unemployment Most Important Economy Most Important Time of completing full-time education. 1: Primary 2: Secondary 3: Post-secondary Which of the following best describe 1: Employed your current employment status? 0: Other Which of the following best describe 1: Unemployed your current employment status? 0: Other Which best describes the area where 1: Big city you live? 0: Small city or a country village How does the financial situation of 1: Better your household compare with what it 2: Neutral was 12 months ago? 3: Worse Would you say that over the past 1: Better year the nation's economy has 2: Neutral gotten better, stayed about the same 3: Worse or gotten worse? Unemployment most important issue 1: yes facing country today 0: no Economy most important issue 1: yes facing country today 0: no Number of Std. Min Max Mean Observations Dev. 5880 0 1.638.481 5248 1 3 1.945.824 5966 0 1.973.161 5959 0 1.073.259 5949 0 1.132.338 6001 0 1.457.498 5926 1 6 3.993 1.570 5776 1 3 2.276.658 6001 0 1.533.499 6001 0 1.169.375 5948 0 1.369.483 5907 1 3 2.438.651 5914 1 3 2.609.660 6001 0 1.312.463 6001 0 1.325.469 15

Cultural controls and migrant perceptions Immigrant Education Should government give preference versus Job to immigrants with a high level of education but no job offer, or immigrants who have a job offer but a lower level of education Immigrant as Neighbor Comfortable with immigrant as neighbor Most Immigrants Legal Are most immigrants legal? Citizenship: Lived in country Citizenship: Same cultural values Cultural Effect Immigration Immigrant should act like native Having lived in country for most of one s life most important value for citizenship Sharing cultural values most 1: High education/ no job offer 0: Low education/ job offer 1: Comfortable 0: Uncomfortable 1: Yes 0: No 0: Agree 1: Disagree 0: Agree important value for citizenship 1: Disagree Some people think that immigration 1: Immigration enriches culture with new customs enriches country's and ideas. Others think that these culture new customs and ideas negatively 0: Immigration affect culture. Which comes closer to negatively affects your point of view? country's culture People who come to [COUNTRY] 0: Agree should try to act like people from 1: Disagree (COUNTRY) 5071 0 1.296.456 5665 0 1.867.340 5284 0 1.568.495 5886 0 1.107.309 5886 0 1.155.363 5519 0 1.645.479 5851 0 1.282.449 Immigration Most Important Immigration most important issue facing country today 1: Yes 0: No 6001 0 1.070.256 5. ANALYSIS PLAN To test the hypothesis that economic factors do not play a statistically significant role in perceptions on the consequences of immigration, this paper will run ordinary least squares (OLS) regressions that controls for demographic and cultural factors on each of the perceived consequences. The first regression model (Model 1) includes the demographic and economic variables. The second regression model is the full model (Model 2) where I add the cultural variables to Model 1. I decided not to include country dummy variables in the models but instead 16

decided to run regressions separately for each country allowing the regression slope to vary for the six countries. Model 1 Immigrants Take Jobs = β1 Demographic controls + β2 Economic controls + ε Cultural Effect Immigration = β1 Demographic controls + β2 Economic controls + ε Model 2 Immigrants Take Jobs = β1 Demographic controls + β2 Economic controls + β3 Cultural controls + ε Cultural Effect Immigration = β1 Demographic controls + β2 Economic controls + β3 Cultural controls + ε 6. RESULTS As hypothesized, there is strong evidence that demographic and cultural variables rather than economic variables - are statistically significant and correlated with both the economic and cultural consequence variables. Table 6 reports the regression estimates for the pooled sample for the most important variables. The United States is the baseline country that was left out of the regression. The results of the regressions using model 1 and model 2 for the Immigrants Take jobs variable are presented in tables 7a and 7b. Tables 8a and 8b present the result for the regression on the Cultural effect Immigration variable. As both dependent variables are dichotomous variables, the predicted values will represent a percentage point increase or a 17

decrease in the predicted probability of disagreeing with the statement that immigrants take job (Disagree=1) and the probability of believing that immigration enriches a country's culture (Enriches culture=1). Table 6: Attitudes towards immigrants - estimation results for pooled sample (USA is baseline) Variables Immigrants Take jobs Country Coeff. Std. Dev. Cultural effect Immigration Coeff. Std. Dev. Should act like native Coeff. Std. Dev. Most Immigrants Legal Coeff. UK.044*.026 -.098***.026 -.099***.026.259***.026 France.382***.027.076***.027 -.419***.027.311***.028 Germany.332***.027.058***.027 -.252***.027.411***.027 Italy.385***.027.146***.027 -.212***.027 -.105***.029 Spain.296***.027.079***.027 -.361***.027.060*** 0.03 Notes: Significant at the ***1, **5, *10 percent level. Std. Dev. 18

Table 7a: Attitudes towards immigrants -estimation results for each country Dependent variable Immigrants Take jobs Country USA UK France Model (1) (2) (1) (2) (1) (2) Independent variables Ideology -.073*** -.013 -.102*** -.067*** -.096*** -.042** (.021) (.025) (.023) (.025) (.017) (.021) Citizen -.081 (.137) -.093 (.148).001 (.135).136 (.138).103 (.109).053 (.127) Foreign Born.202** (.086).148 (.096).039 (.084) -.043 (.096).044 (.074) -.035 (.088) Foreign Born Parents.005 (.055).033 (.062).147** (.063).229*** (.0737).006 (.044) -.029 (.051) Male.062* (.034).064 (.040) -.047 (.036) -.028 (.041) -.064** (.028) -.041 (.032) Age.001 (.001).0002 (.001).002 (.001).005*** (.002).003* (.002).003* (.002) Education.088** (.035).033 (.042).069** (.034) -.031 (.039).118*** (.025).092*** (.030) Employed -.009 (.049) -.080 (.053).008 (.049).040 (.054).039 (.047).041 (.055) Unemployed.069 (.055).061 (.063) -.066 (.059) -.028 (.068) -.023 (.061).015 (.070) Urban.036 (.035) -.003 (.041).075** (.037).044 (.042).013 (.031).007 (.035) Satisfied Personal Finances.038** (.017).027 (.020).023 (.018).019 (.020) -.008 (.017) -.016 (.020) Satisfied Economy -.057** (.027) -.011 (.031) -.047 (.031) -.042 (.035) -.005 (.029) -.007 (.033) Unemployment Most Important.018 -.077.106**.089.017 -.012 (.046) (.058) (.046) (.056) (.032) Economy Most Important.005 -.075.086**.001.056 (.041) (.052) (.040) (.049) (.036) Immigrant Education - Job.012 -.084** (.041) (.041) Immigrant as Neighbor.129***.117** (.066) (.057) Most Immigrants Legal.136***.005 (.073) (.044) Citizenship: Lived in country.023 -.156*** (.065) (.052) Citizenship: Same cultural values.067 -.028 (.054) (.050) Cultural effect immigration.171***.227*** (.047) (.046) Immigrant should act like native.081***.075*** (.020) (.021) Immigration Most Important -.215** -.081 (.098) (.063) Constant.361* -.032.350 -.019.483** (.204) (.241) (.221) (.249) (.18*) Observations 828 542 772 547 780 546 Adjusted R².049.170.063.209.077.135 Notes: Significant at the ***1, **5, *10 percent level. Standard deviations are in parentheses. 19 (.038).006 (.042) -.069 (.043).197*** (.062).084** (.039) -.005 (.063).031 (.041).090** (.040).032 (.022) -.093 (.079).171 (.229)

Table 7b: Attitudes towards immigrants -estimation results for each country Dependent variable Immigrants Take jobs Country Germany Italy Spain Model (1) (2) (1) (2) (1) (2) Independent variables Ideology -.049*** (.017) -.005 (.018) -.126*** (.019) -.079*** (.022) -.106*** (.018) -.028 (.021) Citizen -.025 (.069) -.020 (.071) -.023 (.197) -.053 (.223) -.040 (.118).007 (.131) Foreign Born -.054 (.055) -.078 (.058.119 (.102).088 (.115).104 (.098) -.051 (.114) Foreign Born Parents.119*** (.042).119*** (.044).000 (.115).035 (.122).115 (.097).155 (.116) Male.057** (.026).097*** (.028) -.033 (.032) -.029 (.036).015 (.032).024 (.036) Age -.001 (.0008).001 (.001).002** (.001).002* (.001).003*** (.001).002* (.001) Education.108*** (.020).077*** (.021).098*** (.024).062** (.029).139*** (.023).067** (.027) Employed -.045 (.030) -.034 (.032) -.016 (.038) -.021 (.044).053 (.043).070 (.051) Unemployed -.007 (.033).015 (.036) -.093* (.052) -.118** (.060).016 (.050).091 (.059) Urban.008 (.025).008 (.027) -.053* (.032) -.023 (.037).050 (.032).020 (.037) Satisfied Personal Finances.024 (.015).014 (.017) -.007 (.020).016 (.023).060*** (.018).023 (.021) Satisfied Economy -.045*** (.016) -.011 (.017) -.098** (.039) -.080* (.043) -.027 (.037) -.068 (.041) Unemployment Most Important -.063* (.034) -.051 (.037) -.073* (.038) -.005 (.052).109** (.054) -.051 (.074) Economy Most Important -.008 (.030) -.013 (.032).035 (.039).081 (.052).147*** (.054).009 (.075) Immigrant Education - Job -.003 (.027) -.046 (.046) -.104** (.045) Immigrant as Neighbor.213*** (.047).226*** (.052).324*** (.062) Most Immigrants Legal.134*** (.042).036 (.040).133*** (.038) Citizenship: Lived in country.027 (.066) -.013 (.066) -.101** (.049) Citizenship: Same cultural values.024 (.044).009 (.054) -.093*.048 Cultural effect immigration.135*** (.034).157*** (.044).145*** (.043) Immigrant should act like native.048*** (.018).071*** (.020).051** (.023) Immigration Most Important -.103 (.073).127* (.075) -.237 (.152) Constant.735*** (.117).102 (.145) 1.037*** (.248).459 (.293).206 (.198).054 (.240) Observations 900 693 698 471 844 567 Adjusted R² 0.08.189.166.246 0.113.239 Notes: Significant at the ***1, **5, *10 percent level. Standard deviations are in parentheses. 20

DEMOGRAPHIC CONTROLS As expected, ideology, age, gender, education, immigrant heritage all have an independent effect on the attitudes towards immigrants taking jobs from natives. When controlled for cultural variables Ideology and Education have the strongest statistically significant relationship with the variable Immigrants Take Jobs. Holding other variables constant, an increase in the level of Education is associated with an increase in the probability of disagreeing with the statement that immigrants take jobs from natives in the limited and the full model. Only for the United States and the United Kingdom, the level of education has no longer a significant relationship with the probability of disagreeing with the statement after controlling for cultural variables. In the full model the biggest increase in the probability of disagreeing with the statement that immigrants take jobs based on a higher educational level can be found in France where there is an increase in probability of 9 percentage points. Ideology has a negative and significant relationship with the attitudes on immigrants taking jobs from natives in all countries in Model 1. In the full model, this significance disappears for Germany, Spain and the United States and the negative impact on the predicted probability of disagreeing with the statement that immigrants take jobs is the highest in Italy with a decrease of 8 percentage points the more right people are. Age has a positive and statistically significant effect on the probability of disagreeing with the said statement. However, due to the very small effect (half a percentage point as the biggest effect) the age variable does not add much to the understanding of attitudes towards immigration which is in line with other studies. 21

On the other hand, the effects of migrant heritage are somewhat interesting. Whether or not somebody is a Citizen does not seem to have any significant impact on their attitudes toward immigrants taking jobs from natives for any country in any of the models. Being Foreign Born, surprisingly, does not have an impact either except for the United States in the limited model where foreign born individuals are 20 percentage points more likely than natives to disagree with the statement that immigrants take jobs. Having Foreign Born Parents has the expected positive effect on the probability of disagreeing and only has a statistically significant effect in Germany and the United Kingdom. The significant effect of Foreign Born Parents in these two European countries shouldn't surprise, as these countries had the highest number of immigrants of the European countries in 2011 according to 2011 Eurostat data ( 2 ). According to the same data, the United Kingdom had the largest number of immigrants with a total of 566,044, Germany following with 489,422, Spain with 457,649 and Italy lastly with 385,793. The same data shows that these four countries together accounted for about 60 % of all immigrants to EU-27 Member States. For Germany and the United Kingdom the positive effect holds in the full model where individuals with foreign born parents are about 20 percentage points more likely than individuals without foreign born parents to disagree with the statement that immigrants take jobs. In general and in accordance with literature, Male respondents are more likely than female respondents to disagree with the previous statement regarding jobs. There is a 5 to 10 percentage point difference between male and female in the United States, France and Germany. For Germany this difference remains and even gets bigger in the full model. However, for France there is negative effect where male respondents are about 6.5 percentage points less likely than 2 http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/migration_and_migrant_population_statistics 22

female respondents to disagree with the statement that immigrants take jobs from natives and show a more negative attitude towards immigrants. Urban residence has a positive and statistically significant relationship with the Immigrants Take Jobs variable in the first model for the United Kingdom yet a negative statistically significant relationship for Italy. Lastly, as expected, the employment variables do not have a statistically significant impact on the probability of disagreeing with immigrants taking your job. The only statistically significant relationship can be found in Italy where in both models unemployed individuals are about 10 percentage points less likely than individual who are not unemployed (students, retired, employed) to disagree with the statement that immigrants take jobs from natives. ECONOMIC CONTROLS As hypothesized, the economic controls, in general, do not have a statistically significant impact on the probability of disagreeing that immigrants take jobs from natives. I showed this already for the employment variable where there is only a minor effect present for the unemployment variable. For four of the six countries the results show that Satisfaction with Economy has a negative and significant relationship with the attitudes towards migrants: Germany, Italy, Spain and the United States. A more negative assessment of the economy is associated with an increased agreement that immigrants take jobs from natives which indicates a more negative perception of immigrants. After controlling for the cultural variables, the statistical significant relationship holds only for Italy where a negative assessment of the economy is associated with a 23

decrease of about 6 percentage points in the probability of disagreeing that immigrants take jobs from natives. The statistical significance of the Satisfaction with Personal Finances variable is only present in the first model and this for two countries: the United States and Spain. The sign of this relationship is very surprising. When people indicate that their personal financial situation has not improved in the previous year, this is associated with an increase in probability of disagreeing that immigrants take jobs from natives indicating a more positive view on immigration. The belief that either the Economy or Unemployment are the most important issues today only have a statistical significance in model 1 and as expected are no longer statistically significant when controlling for immigrant perception and cultural controls. An interesting relationship is present for these two variables. In the United Kingdom and Spain, the individuals who think that either unemployment or the economy are the most important issues of today are between 8 and 15 percentage points more likely than people who think that some other issue is more important to disagree with the statement that immigrants take jobs from native. This relationship is surprisingly associated with a more positive attitude. On the other hand and in line with what was expected, in Germany and Italy we see an opposite direction for the Unemployment Most Important variable. Here we see that the individuals who think that unemployment is the most important issue today are about 7 percentage points less likely than people who think that some other issue is more important to disagree with the statement that immigrants take jobs from native. 24

When we assess the statistical significance of the economic variables, in accordance with my hypothesis, when cultural variables are added to the model the economic variables no longer have a statistically significant effect on the attitudes towards migration. The only country that retains some statistically significant economic variables is Italy where individuals who are unemployed and less satisfied with the economy are less likely to show a positive attitude towards migration. CULTURAL CONTROLS AND IMMIGRANT PERCEPTION All cultural and immigrant perception controls show a statistically significant effect on the attitudes towards immigration. The variables that are highly statistically significant for all countries are Immigrant as Neighbor and Cultural Effect Migration. Individuals who are comfortable with immigrants as neighbors are between 12 to 32 percentage points more likely than people who are not comfortable with immigrants as neighbors to have a more positive view and to disagree that immigrants take their jobs. Individuals who think that immigration has an enriching effect on the country's culture are between 9 to 23 percentage points more likely to disagree with the statement that immigrants take jobs from natives. A few cultural variables have a negative and statistically significant relationship with the dependent variable Immigrants Take Jobs. Individuals who think a precondition for Citizenship should be having lived in country for most of one's life or 'sharing cultural values and individuals who would give preference to polices that allow highly Educated immigrants with No Job Offer to enter the country are less likely (with about 10 percentage points) to disagree with the statement that immigrants take jobs from migrants which indicates a more negative attitude. 25

For individuals who prefer migrants with no job offer to enter the country over those who do have a job offer, this increase in the perception that immigrants take jobs from natives is counterintuitive. Individuals who think that Most Immigrants are Legal and those who think immigrants are not required to Act like Natives are about 7 percentage points more likely than people who think the opposite to disagree that immigrants take jobs from natives. For the Most Immigrants are Legal variable this effect is not statistically significant for Italy and United Kingdom which is somewhat odd since Italy is confronted with a lot of the inflow of immigrants. In France, the variable Immigrant should act like native does not have a statistically significant effect. Lastly, the variable Immigration Most Important issue is statistically significant in the regression for the United States and Italy. Surprisingly, Italians who think immigration is the most important issue are 13 percentage points more likely than Italians who think some other issue is important to have positive views on immigration. However, for Americans this relationship has a negative sign and has a larger effect indicating a decrease in the likelihood of being positive towards immigrants when an American thinks that immigration is the most important issue (which is the expected sign). To demonstrate that economic variables have less explanatory power regardless of the perceived consequences of immigration, I ran the same models on the dichotomous variables Cultural Effect Immigration. Tables 8a and 8b demonstrate that demographics and cultural variables have the most statistically significant relationship with the cultural consequence variable and can explain better what determine people s attitudes towards immigrants. 26

Table 8a: Attitudes towards immigrants -estimation results for each country Dependent variable Cultural Effect Immigration Country USA UK France Model (1) (2) (1) (2) (1) (2) Independent variables Ideology -.112*** (.021) -.082*** (.023) -.103*** (.023) -.049** (.024) -.166 (.020) -.104*** (.022) Citizen.051 (.133).038 (.136) -.019 (.142).010 (.132).118 (.125).160 (.138) Foreign Born.074 (.084).023 (.088).103 (.089).055 (.091).038 (.086).057 (.095) Foreign Born Parents.018 (.053).018 (.057).074 (.066).086 (.071).076 (.052).072 (.055) Male.023 (.033).027 (.037).013 (.036).025 (.039) -.094*** (.032) -.113*** (.035 Age -.003** (.001) -.002** (.001) -.000 (.002).002 (.002).001 (.002).002 (.002) Education.153*** (.034).095** (.038).154*** (.034).091** (.037).125*** (.029) -.035** (.059) Employed.031 (.044) -.046 (.048).094** (.050).121** (.051).034 (.054) -.035 (.059) Unemployed.046 (.054).046 (.058).034 (.060).075 (.065).045 (.070).016 (.075) Urban.067** (.034).033 (.037).072* (.037).063 (.040).002 (.036) -.0001 (.038) Satisfied Personal Finances.042 (.017).021 (.019).039** (.018) -.006 (.019).014 (.019) -.004 (.021) Satisfied Economy -.025 (.027).022 (.029).023 (.032).051 (.034) -.038 (.034) -.041 (.035) Unemployment Most Important.088*.061.085*.049.056.036 (.046) (.053) (.046) (.052) (.037) Economy Most Important.066*.031.135***.025.096** (.040) (.048).041 (.047) (.041) Immigrant Education - Job.070* -.053 (.038) (.039) Immigrant as Neighbor.400*** -.053*** (.058) (.039) Most Immigrants Legal.172***.143*** (.039) (.042) Citizenship: Lived in country.038 -.109** (.060) (.050) Citizenship: Same cultural values.004 -.023 (.050) (.048) Immigrant should act like native.073***.088*** (.018) (-.183) Immigration Most Important -.198** -.183*** (.090) (.060) Constant.416*** -.127.138 -.492***.536*** (.196) (.220) (.228) (.238) (.220) Observations 770 544 726 551 747 547 Adjusted R².098.244.125.309.142.302 Notes: Significant at the ***1, **5, *10 percent level. Standard deviations are in parentheses. (.041).028 (.045) -.008 (.046).390*** (.069).188*** (.042) -.070 (.069) -.072 (.044).088*** (.024) -.140* (.085).014 (.248) 27

Table 8b: Attitudes towards immigrants -estimation results for each country Dependent variable Cultural Effect Immigration Country Germany Italy Spain Model (1) (2) (1) (2) (1) (2) Independent variables Ideology -.110*** (.019) -.071*** (.020) -.117*** (.021) -.050** (.023) -.123*** (.019) -.076*** (.021) Citizen -.071 (.079) -.0007 (.079) -.084 (.216) -.176 (.238).089 (.117).125 (.126) Foreign Born -.076 (.064).-027 (.065).084 (.111).012 (.123).218** (.101).139 (.111) Foreign Born Parents.096** (.048).043 (.049) -.010 (.126).041 (.130).092 (.102).078 (.115) Male -.063** (.030) -.022 (.031) -.033 (.036) -.024 (.039) -.047 (.033) -.028 (.036) Age -.001 (.001) -.001 (.001) -.001 (.001) -.001 (.001).001 (.001).001 (.001) Education.092*** (.022).072*** (.023).128*** (.028).068** (.030).118*** (.024).062** (.027) Employed -.031 (.035) -.022 (.036) -.013 (.043) -.008 (.047) -.020 (.045).001 (.051) Unemployed -.101*** (.039) -.088** (.040) -.075 (.058) -.092 (.063) -.056 (.052) -.053 (.059) Urban.046 (.030).029 (.031) -.006 (.036) -.001 (.039).028 (.033).019 (.041) Satisfied Personal Finances.000 (.018) -.010 (.019).021 (.022).049** (.024).057*** (.018).041* (.021) Satisfied Economy -.050*** (.018) -.044** (.019) -.020 (.044) -.001 (.046).025 (.039).019 (.041) Unemployment Most Important -.008 (.040) -.0001 (.041) -.032 (.044) -.126** (.055).065 (.057).004 (.073) Economy Most Important.036 (.034).008 (.035) -.001 (.045) -.073 (.054).069 (.058).011 (.074) Immigrant Education - Job -.015.031 -.097** (.048) -.169*** (.044) Immigrant as Neighbor.413*** (.050).328*** (.053).378*** (.030) Most Immigrants Legal -.060 (.046).080* (.043).082** (.037) Citizenship: Lived in country -.082 (.074) -.076 (.070).014 (.049) Citizenship: Same cultural values -.100** (.048) -.116** (.057) -.087* (.048) Immigrant should act like native.095*** (.020).097*** (.022).097*** (.023) Immigration Most Important -.226*** (.081) -.297*** (.079) -.089 (.152) Constant.999*** (.137).418*** (.160).811*** (.274).496 (.313).279 (.178) -.157 (.237) Observations 869 697 648 473 794 579 Adjusted R² 0.105.236 0.079.243.124.233 Notes: Significant at the ***1, **5, *10 percent level. Standard errors are in parentheses 28