How critical thinking, multicultural education and teacher qualification affect anti-immigrant attitudes

Abstract

Previous studies identify a relationship between education and anti-immigrant attitudes. There is, however, uncertainty regarding the underlying explanations linking education to attitudes. In this article, we examine whether a relationship exists between exposure to teaching about critical thinking as well as multiculturalism (measured as religions/cultures as well as xenophobia/racism), and anti-immigrant attitudes among adolescents. In addition, we examine whether teacher qualification matters for attitudes. The analysis is based on survey data collected from high school students in Sweden. The results show an association between exposure to teaching about critical thinking as well as multiculturalism (both indicators) and anti-immigrant attitudes among students, i.e. higher exposure is related to lower levels of anti-immigrant attitudes. However, we find that teaching about xenophobia/ racism affects attitudes, but not when simultaneously controlling for teaching about critical thinking and religions/cultures. In terms of teacher qualification, we find that students in schools with a high proportion of certified teachers tend to have lower levels of anti-immigrant attitudes.

Keywords:

• anti-immigrant attitudes
• education
• critical thinking
• multicultural education
• teacher qualification
• Sweden

Introduction

It is well known that high educational attainment reduces negative out-group attitudes (Hello, Scheepers, & Sleegers, 2006; Hello, Scheepers, Vermulst, & Gerris, 2004; Hjerm, 2001; Schuman, 1997; Weil, 1985). While several studies have established that there is a strong relationship between an individual’s level of education and their attitudes towards out-groups, less is known about what links education to attitudes. In this article, we focus on the instructional content of education as a potential explanation, and examine whether two types of knowledge acquired in school, namely critical thinking and multiculturalism, contribute to the understanding of the relationship between education and anti-immigrant attitudes. In addition to the role of critical thinking and multiculturalism, we also examine whether there is a relationship between teacher qualification and adolescents’ anti-immigrant attitudes.

Previous research identifies several possible explanations for the relationship between level of education and attitudes about immigrants. For example, level of education may be a proxy for social status. A higher level of education implies a more privileged position in society, and consequently less exposure to competition from ethnic minorities in the labour market, housing market, and schools (Hello, Scheepers, & Gijsberts, 2002; Meeusen, de Vroome, & Hooghe, 2013). Another explanation commonly referred to in the literature is the selection effect, i.e. individuals who have a parent with a high level of educational attainment tend to be less prejudiced due to their social background and more likely to continue with further studies themselves. Lancee and Sarrasin (2015), in fact, show, by using longitudinal data, that higher educational attainment in itself seems to have no liberalizing effect in Switzerland. Further, a socialization effect may also explain the relationship between education and attitudes towards immigrants. From this perspective, the school acts like a socializing agent promoting the dominant norms of democratic societies (Bowles & Gintis, 1976). In addition to the socializing force of schools as institutions, schools expose students to other social influences such as when interacting with other students. The longer an individual spends within the educational system, the more likely he or she is to develop social contacts with ethnic minorities (Hello et al., 2004), which according to the contact theory (e.g. Pettigrew & Tropp, 2006), tend to reduce anti-immigrant attitudes. Finally, social desirability might also partly account for the relationship between education and anti-immigrant attitudes. However, even though social desirability has been empirically shown to matter via bogus pipeline experiments (Sigall & Page, 1971) or inducing answer-related consequences (Michaelis & Eysenck, 1971), the extent to which social desirability affects survey responses is overall rather limited.

While these previous explanations help shed light on the relationship between education and out-group attitudes, they fail to consider what is taught in school and by whom. This analysis departs from previous research by specifically focusing on what is going on inside the schooling system in terms of instruction and content knowledge. From this perspective, the relationship between education and attitudes towards immigrants concerns the actual knowledge acquired in school and includes what individuals learn based on curriculum guides and other instruments used by teachers (Banks, 2001; Gay, 2004). Curricula and instruction that prioritize critical thinking skills and multiculturalism are likely to promote a higher level of awareness about stereotypes rendering negative out-group attitudes inert. Further, as predicted by social learning theory, a better understanding of different religions and cultures should further reduce negative out-group attitudes.

The two central questions raised in this article are: ‘How does teaching about critical thinking and multiculturalism influence adolescents’ anti-immigrant attitudes?’ and ‘How does teacher qualification influence adolescents’ anti-immigrant attitudes?’ Critical thinking is understood as a general type of knowledge involving reflective and rational thinking that permeates the school curriculum, and consequently is present within every subject taught in schools. Multiculturalism and multicultural education refers to a specific type of knowledge about different cultures and groups of people, often aimed at reducing anti-immigrant attitudes. Finally, we examine the potential relationship between teacher qualification and adolescents’ anti-immigrant attitudes.

Theoretical background

We now turn to a more in-depth discussion about these specific aspects of knowledge and education. Critical thinking is a cognitive process that we can learn and develop (Daly, 1998; Paul & Heaslip, 1995; Walsh & Paul, 1986) which affects the way we think about and understand the world. Critical thinking is: ‘… reasonable and reflective thinking focused on deciding on what to believe or do’ (Ennis, 1993, p. 180). The cognitive process includes promotion of initiative and independent judgement as well as problem solving and conflict resolution (Hello et al., 2004). Furthermore, critical thinking is a method of thinking and reasoning that includes developing metacognitive skills and applying the problem-solving structure, learnt in some context, to a new situation (Pithers & Soden, 2000).

The body of research on critical thinking shows that developing critical thinking skills decreases prejudice by increasing the awareness that group members can do things as individuals, and that these acts are not acts performed by the group as a whole (see e.g. Abrami et al., 2008; Garside, 1996). Critical thinking thus challenges our stereotypes, understood as cognitive, learned constructs about other groups or group-members (Lippman, 1922; Stangor, 2009). Critical thinking skills should therefore enable individuals to identify stereotypes, understand the limitations of reliance on stereotypes, and analyse their own prejudices. Previous research on stereotyping has shown that prejudice does not rely on an individual’s awareness of stereotypes but rather their endorsement of specific stereotypes (Devine & Elliot, 1995). It is reasonable to assume that students having received education about critical thinking should be less prejudiced, since they have been provided with tools and a higher level of cognitive capacity to think about other groups and cultures in a more reflective and reasonable way. Critical thinking is a deeply grounded and valued skill, and education about critical thinking permeates the course curriculum of most schools in modern societies, therefore acting as some latent curricula. While critical thinking fosters reflective and reasoned thinking about different groups, cultures, and religions, it may not act directly to decrease the prevalence of stereotypes held by an individual. Rather critical thinking enhances an individual’s ability to question the validity of those stereotypes.

The second instructional component that we focus on is multicultural education. This component revolves around different sets of knowledge required by citizens in a democratic society in order to get along with people from other ethnic or cultural groups than their own (Banks & Banks, 2009). Multicultural education is a broad concept and there is little consensus in the literature regarding the boundaries of multicultural education (Banks, 1993; Sleetner & Grant, 1987). In this study, multicultural education refers to what Castagno (2009) views as educating for cross-cultural competence. First, it refers to education and knowledge relating to facts and values about different religions and cultures. From this point of view, multicultural education aims to increase knowledge about different religions and cultures in order to foster cross-cultural competence. Second, multicultural education also incorporates teaching students about xenophobia and racism.

The reason multicultural education should reduce negative out-group attitudes comes from social learning theory, which explains varying degrees of prejudice by focusing on the learning processes of attitudes and behaviour during early childhood and adolescence. Social learning theory is often approached within the framework of symbolic racism theories (Henry & Sears, 2002; Meertens & Pettigrew, 1997; Sanders & Kinder, 1996; Sears, Laar, Carrillo, & Kosterman, 1997) which assume that racism and prejudice are guided by a set of assumptions about what minority ethnic groups deserve and how they should act. Social learning theory suggests that by acquiring more knowledge about different groups, individuals will understand and learn to respect other cultures, thereby decreasing negative out-group attitudes.

Previously, intervention studies have assessed the potential effect of multicultural education on prejudice (e.g. Wright & Tolan, 2009). Both training in diversity and multicultural thinking (Paluck & Green, 2009), as well as cognitive programmes reduce prejudice (Aboud & Levy, 2000). However, other studies indicate a more limited effect (Beelmann & Heinemann, 2014). It is also unclear whether such interventions matter over time, and whether they are feasible in an educational setting where schools must achieve a multitude of tasks in addition to the reduction of negative out-group attitudes (Banks, 1991). However, recent non-experimental research by Verkuyten and Thijs (2013) shows that multicultural education has a positive effect on inter-ethnic attitudes, specifically in terms of two mechanisms: improving cultural knowledge and the establishment of anti-racism norms within the classroom. Still, more research set within the schooling system in different contexts is called for to assess the impact of multicultural education on attitudes. We expect, according to social learning theory, that multicultural education in the form of increasing knowledge about different religions and cultures, as well as about racism and xenophobia, will reduce negative attitudes towards immigrants among adolescents.

Furthermore, it might not just be the content of instruction in schools that matters, but also the quality of instruction. In relation to anti-immigrant attitudes among adolescents there are two potentially important aspects related to the teachers. First, previous research has shown that students generally tend to learn more from qualified teachers than from non-qualified teachers. For example, studies show a relationship between students’ grades and teachers’ level of education (Rockoff, 2004). Given that students generally learn more from qualified teachers, it is safe to assume that students with greater contact with credentialed teachers will receive more and better instruction about critical thinking and multiculturalism. Second, credentialed teachers are more likely to incorporate the societal values and norms into their teaching because the training of qualified teachers is both subject-specific, and designed to promote the liberal and democratic values stipulated by the Swedish Education Act (2010:800). This training includes actively dealing with racism and intolerance, whenever confronted with such sentiments in schools. Therefore, we expect that qualified teachers have better overall pedagogical skills as well as more awareness of the societal values and norms stipulated in the Education Act and curricula, and should consequently be more effective in teaching critical thinking and multiculturalism. Students exposed to qualified teachers should thus display lower levels of anti-immigrant attitudes compared to students exposed to teachers who are not qualified. We now set out to examine whether these aspects of education matter for adolescents’ anti-immigrant attitudes.

Data

To answer our research questions, we use survey data collected from Swedish high school students. In the Swedish school system, children start primary school at the age of seven and attend a minimum of nine years of obligatory schooling. The clear majority (about 98%) of primary school pupils continue with senior high school, typically for another three years of schooling. Swedish high schools offer several different national programs, divided into two main tracks. The theoretical track prepares students for further studies at the university level, while a vocational (‘practical programs’) track offers training for a variety of professions. The vocational high school programs also offer students opportunities to continue with certain types of higher education at the university level.

Statistics Sweden administered the survey and collected the data during the school year 2009/2010. The students (aged 16–18 years, mean = 17.1) in the sample attended the first and last years in senior high school, and the number of respondents is 4675 (girls = 50.8%). Most high school students in the sample are born in Sweden (91.0%) and 70.6% have two Swedish-born parents. 10.1 per cent of the students born in Sweden have one parent born in another country and 10.3% of the students born in Sweden have two parents born in another country.

The students in the survey attended 154 high schools in Sweden. We used stratified sampling in order to ensure representation of schools with a high number of students with foreign background. We excluded schools having too few students to merit consideration from the sample procedure. Forty-six schools from the initial sample declined participation, because their students had recently participated in several surveys. Forty per cent of the sample comes from schools in the three largest cities of Sweden, Stockholm, Gothenburg and Malmö. From within the sample, we draw a new unbound random sample of classes with at least one class per school resulting in 328 classes. The response rate of the survey was 75%. Most of the 25% who did not participate in the survey were absent from the school at the time of the survey due to, for example, internships, while a few students declined to participate in the survey (Löwander & Lange, 2010). We added register data of the participating schools and socio-demographic information about the areas where the schools are situated to the survey data.

Measures

The dependent variable: anti-immigrant attitudes

The dependent variable, anti-immigrant attitudes, is an additive index based on the following two survey items: ‘Do you think it is good for Sweden’s economy that people from different countries come here to live?’ and ‘Do you think it is good for Sweden’s cultural life if people from other countries come here to live?’ The high school students indicated on a five-point scale to what extent they agreed or disagreed with these statements: (1) No; (2) Doubtful; (3) Yes, to a certain extent; (4) Yes, absolutely; (5) Have no opinion. We exclude the ‘Have no opinion’-category and use the remaining responses to create an additive index. We use the Spearman–Brown test of validity (see e.g. Eisinga, Grotenhuis, & Pelzer, 2013) to test whether these items are suitable for constructing a two-item scale. The resulting value of .76 confirms the appropriateness of using the two items to construct an additive index. The range of the index is between 0 and 10, where higher values indicate negative attitudes towards immigrants. The items used to construct the index in this study are identical to those used by the European Social Survey to denote negative out-group attitudes (Hjerm, 2009; Schneider, 2008).

It is important to note that measuring negative attitudes towards immigrants is not an entirely straightforward task and there is disagreement among scholars over how to operationalize such concepts. In the European context, scholars have used different concepts such as xenophobia, anti-immigrant attitudes, threat, and prejudice, when attempting to capture negative attitudes towards immigrants and minority groups (Ceobanu & Escandell, 2010). In general, it is understood among researchers who study prejudice that the associations between a variety of independent variables and different concepts such as xenophobia and anti-immigrant attitudes tend to be robust regardless of the choice of outcome variables.

Previous research suggests that the two items used in this study function well as a measure of anti-immigrant attitudes. However, each item alone does not necessarily have face validity. For example, it is possible to think that immigrants are a burden to the economy while at the same time believing that immigration is something a society should support despite the economic burden. The items in principle measure stereotypes (empirically testable statements about the world) and not attitudes (value judgements about a subject). However, as an operationalization of the latter, the items are functional given that the index correlates highly with other comparable attitudinal concepts. There is, for example, a .6 correlation in the data-set between our index and related attitudes such as, wanting to decrease immigration from poorer countries as well as unfavourable attitudes towards Muslims. Moreover, we find similar patterns of correlation in other data-sets, further strengthening the validity of our index measuring anti-immigrant attitudes. For example, in the European Social Survey the correlation in Sweden between our dependent variable and the willingness to decrease or increase immigration of non-ethnic immigrants is .7.

Independent variables

There is a scarcity of existing survey data about different aspects of education relating to critical thinking and multiculturalism, and even less data which links these to anti-immigrant attitudes. Due to limitations in the data, we use single items to measure the amount of exposure students have to teaching about critical thinking as well as xenophobia and racism. We use the following survey item to measure student exposure to critical thinking: ‘Do you think your teachers are trying to make you think critically?’ The high school students used the following five-point scale to answer this question: (0) ‘No nothing’; (1) ‘Yes, a little’; (2) ‘Yes, quite a bit’; (3) ‘Yes, very much’; and (4) ‘Do not remember’. We here code the response (4) ‘Do not remember’ as (0) ‘No nothing’ and thereafter rescale the variable to vary between 0 and 10, where high values represent a high level of exposure to teaching about critical thinking. To reduce potential bias associated with self-reports, we use the aggregated responses for each class (mean score) as our measure capturing the extent of exposure individual students have to teaching about critical thinking. If individuals’ self-reports about exposure to teaching about critical thinking are affected by their attitudes towards immigrants we substantially reduce some of the potential bias by measuring exposure to teaching about critical thinking at the aggregated class level. This approach is in line with the ‘wisdom of crowds’ argument (Surowiecki, 2004) where it has been shown that the group makes a better evaluation than single experts. The wisdom of crowds has been validated in a number of different fields and has not been restricted to single events or number guessing but also in relation to more difficult problem-solving issues (e.g. Yi, Steyvers, Lee, & Dry, 2012). Even though self-reported data contains bias, research shows that even self -reported grades, although biased in different groups, nonetheless predict outcomes as well as non-self-reported grades (Kuncel, Credé, & Thomas, 2005). By aggregating individuals’ attitudes to the classroom level we do substantially reduce biases.

We measure exposure to multicultural education, in two different ways. First, we operationalize it as exposure to specific teaching and knowledge about the five big religions of the world: Judaism, Christianity, Islam, Buddhism and Hinduism. Learning about these religions also includes learning about the cultures related to them. Here, we use one survey item for each religion asking students: ‘Have you some time during your time in school been taught about [Religion]?’ The five-point response scale consisted of the following options: (0) No nothing; (1) Yes, a little; (2) Yes, quite a bit; (3) Yes, very much; and (4) Do not remember. We code the response (4) ‘Do not remember’ as (0) ‘No nothing’ and thereafter create an additive index. We rescale the index to vary between 0 and 10, wherein high values represent a high level of exposure to teaching about religions and cultures. As previously discussed, we aggregate the scores on the index to the class level to reduce bias associated with self-reports. Thus, the scale measures the extent of exposure an individual student has to teaching about religions and cultures by the mean index score of the student’s class.

Second, we also define multicultural education as teaching students about xenophobia and racism. We use the following survey item to measure this aspect of multicultural education: ‘Have you, during your time in school, been taught about racism and xenophobia?’ The five-point response scale consisted of the following options: (0) No nothing; (1) Yes, a little; (2) Yes, quite a bit; (3) Yes, very much; and (4) Do not remember. We code the response (4) ‘Do not remember’ as (0) ‘No nothing.’ Again, we re-scale the variable to vary between 0 and 10, where high values represent a higher level of exposure to teaching about xenophobia and racism, and use the aggregated responses for each class (mean score) to reduce bias associated with self-reports.

Finally, the third educational component that we focus on is qualifications of teachers. The variable ‘proportion of certified teachers’ captures this aspect and measures the proportion of teachers in each school that have earned relevant teaching credentials (i.e. they hold an exam in teaching from a university). The information about the proportion of certified teachers in each school comes from school registers. Unfortunately, the data do not permit us to link a specific teacher to a specific classroom setting. Therefore, we must treat this variable as a school-level variable, rather than a class-level variable. However, we assume that the higher the proportion of certified teachers in a school, the higher is the probability of exposure to a higher quality of teaching.

In addition to our central independent variables measuring students’ exposure to instruction about critical thinking and multiculturalism as well as the qualifications of the teachers, we control for a number of individual-level and school-level variables in order to take into account other relevant factors potentially affecting adolescents’ anti-immigrant attitudes. The individual-level controls include the following: gender, parental level of education (tertiary vs. other), immigrant status (4 categories), and type of high school programme (practical vs. theoretical). To ensure that any relationships between our central independent variables and anti-immigrant attitudes are not related to the socio-economic character of the schools, we also control for the level of income in the school area (per cent highest income quintile in school area). The level of income in the school area is a reasonable proxy for the socio-economic composition of different schools, since the vast majority of high school students in Sweden attend a high school in the area where they live.

Method

To account for the three-level structure of the data, where students (level 1) are nested within classes (level 2), which are in turn nested within schools (level 3), we apply multilevel analysis (MLA). This type of regression technique is suitable when simultaneously using information and data collected from individuals as well as their contexts, i.e. when modelling variables that belong to different hierarchical levels (Hox, 2010; Steenbergen & Jones, 2002). Multilevel models are, basically, extensions of linear regression models; they provide accurate estimates of uncertainty and statistical significance when dealing with dependent observations. OLS regression models demand independent observations and since our data violate this assumption, there is a risk of underestimating standard errors if using standard regression techniques. Multilevel models thus allow us to simultaneously examine the relationship between individual-level characteristics and contextual-level characteristics in relation to our dependent variable, anti-immigrant attitudes, which is measured at the individual-level. Hence, we can simultaneously study the impact of both class-level variables (critical thinking and multiculturalism) and school-level variables (the proportion of certified teachers) on high school students’ attitudes towards immigrants, net of relevant individual-level and school-level controls. MLA further allows us to decompose the variation in anti-immigrant attitudes into individual-, class-, and school-level variance components. This decomposition provides a measure of the relative importance of factors at each level in accounting for the overall variation in adolescents’ anti-immigrant attitudes.

Results

We now turn to the empirical analysis to examine the potential relationships between students’ exposure to instruction on critical thinking and multiculturalism, as well as teacher qualifications, and the levels of anti-immigrant attitudes. We begin by examining the descriptive statistics for the dependent variable as well as the central independent variables reported in Table 1. In terms of anti-immigrant attitudes, we find that the average score is 4.61 (SD = 3.01) on the 0–10-scale. We also find that the mean score on the variable measuring exposure to teaching about critical thinking is 4.05 (SD = 1.86) on the 0–10-scale across the 328 classes. Further, the mean score on the variables measuring exposure to teaching about ‘religions and cultures’ and ‘xenophobia and racism’ are 5.68 (SD = 1.21) and 4.62 (SD = 1.08), respectively. In terms of teacher qualifications (school-level variable), we find that the mean proportion of certified teachers across schools is .72 (SD = .19). There are, however, schools in the sample where this proportion reaches only 4%, as well as schools where nearly all (97%) the teachers are certified. The difference between the net sample size and the number of cases in our analysis is due to internal non-responses and the unability to classify individuals into classes.

Table 1. Descriptive statistics.

	No. units	Min	Max	Mean	S.D.	Skewness
Anti-immigrant attitudes (i)	3702	0	10	4.61	3.01	.25
Teaching about critical thinking (c)	328	0	10	4.05	1.86	.33
Teaching about religions and cultures (c)	328	1.33	10	5.68	1.21	–.43
Teaching about xenophobia and racism (c)	328	0	8.33	4.62	1.08	–.23
Proportion of certified teachers (s)	154	.04	.97	.72	.19	–1.13

Notes: (i) = individual-level variable; (c) = class-level variable; (s) = school-level variable. Min and max values represent the existing values and not the theoretical end-points.

Having examined the descriptive statistics, we continue with the multi-level regression models presented in Table 2. We first estimate a baseline (null) model to establish the amount of variation in anti-immigrant attitudes that can be attributed to each analytical level, i.e. individuals, classes, and schools. This baseline model (M0) provides information about the relative importance of variables at the individual-level, class-level, and school-level in accounting for variation in anti-immigrant attitudes among adolescents. The variance components in the baseline model (M0), presented in Table 2, indicate that differences between classes account for almost 10% of the variance in anti-immigrant attitudes whereas approximately 8% of the variance can be attributed to differences between schools. From this initial baseline model we can conclude that factors at the class-level and school-level seem to be quite important for explaining variation in adolescents’ attitudes towards immigrants. We now set out to examine whether this class-level and school-level variation in anti-immigrant attitudes are related to high school students’ exposure to teaching about critical thinking and multiculturalism, as well as certified teachers.

Table 2. The relationships between the content of education, teacher qualification and anti-immigrant attitudes. Multilevel regression. Dependent variable: anti-immigrant attitudes (0–10).

	M0	M1	M2	M3	M4
Individual-level variables
Girl (ref.: boy)		–.66**	–.61**	–.68**	–.60**
Parental level of education: tertiary (ref.: other)		–.16	–.15	–.17	–.15
Practical high school programme (ref.: theoretical programme)		–.09	–.12	–.11	–.10
Immigrant status (ref.: student and both parents born in Sweden)
Born outside Sweden		–1.86**	–1.87**	–1.90**	–1.85**
Born in Sweden, both parents born in another country		–1.00**	–1.02**	–1.03**	–1.00**
Born in Sweden, one parent born in another country		–.44*	–.49*	–.48*	–.47*
Class-level variables
Teaching about critical thinking 0–10		–.24**			–.18**
Teaching about religions and cultures 0–10			–.37**		–.27**
Teaching about xenophobia and racism 0–10				–.19**	–.02
School-level variables
Proportion of certified teachers 0–1		–1.43**	–1.46**	–1.64**	–1.40**
Per cent high level income in school area		–.02	–.03*	–.04**	–.02
Intercept	4.70**	10.48**	11.93**	11.03**	11.75**
Individual-level variance	7.53	7.16	7.15	7.16	7.16
Class-level variance	.89	.52	.55	.60	.47
School-level variance	.72	.24	.25	.30	.23
–2 Log Likelihood	18,327	16,472	16,468	16,489	16,458
n	3702	3375	3374	3375	3374

^** p < .01

^* p < .05

To examine these potential relationships, we now turn to Model 1–3 in Table 2. These models examine the relationships between each instructional component and anti-immigrant attitudes (net of controls). When examining Model 1, it becomes evident that the relationship between student exposure to teaching about critical thinking and their attitudes towards immigrants is statistically significant. High school students that are exposed to more teaching about critical thinking (as indicated by the aggregated mean score in their class) tend to report lower levels of anti-immigrant attitudes. The regression coefficient associated with ‘teaching about critical thinking’, indicates that scores on the anti-immigrant index (0–10) decrease by .24 scale points for each point increase on the scale (0–10) measuring exposure to teaching about critical thinking. There is clearly a strong association between belonging to a school class where students on average report higher scores on the scale measuring critical thinking and significantly lower scores on the anti-immigrant index. Moreover, there is a substantial reduction from .89 to .52 in the class-level variance component when including the variable ‘teaching about critical thinking’ in Model 1.

Turning to Model 2 and 3 it also becomes evident that there is a relationship between exposure to teaching about multiculturalism and anti-immigrant attitudes among adolescents. Model 2 indicates that a one-point increase on the scale measuring ‘teaching about religions and cultures’ is associated with a decrease on the anti-immigrant attitudes index by .37 scale points. Similarly, in Model 3 we find that a one-point increase on the scale measuring exposure to ‘teaching about xenophobia and racism’ corresponds to a decrease in index scores by .19 scale points. It is also evident from Model 2 and 3 that the unexplained variance at the class-level is substantially reduced when including these variables. Finally, we simultaneously include all three variables in Model 4, and the results here suggest that teaching about critical thinking and religions/cultures seem to be of particular importance for anti-immigrant attitudes, while the coefficient associated with xenophobia and racism is not statistically significant when controlling for ‘teaching about critical thinking’ and ‘teaching about religions and cultures’. It is important to note that the correlations between the three variables varies between .30 and .36 and the test for collinearity (VIF < 1.3) does not indicate any problems when including all three measures in the same model.

Regarding the other educational aspect examined here, teacher qualification, we find a statistically significant relationship between the proportion of certified teachers in schools and anti-immigrant attitudes among students, as indicated by Model 1–3. The coefficient associated with this variable shows that students on average score 1.43–1.64 scale-points lower on the anti-immigrant index in schools where all teachers are qualified compared to in schools where no teachers are qualified. Hence, each percentage point increase in the share of certified teachers in schools is associated with a decrease in index scores by about .014–.016 scale-points. Considering that the proportion of certified teachers varies substantially across schools (see Table 1), the relationship between teacher qualification and students’ anti-immigrant attitudes is clearly non-negligible. The variance components in Model 1–3 show that including the variable measuring ‘proportion of certified teachers’ substantially reduces the variance between schools in terms of students’ anti-immigrant attitudes. This indicates that variation in the proportion of certified teachers partly explains differences in adolescents’ anti-immigrant attitudes across schools. It should be emphasized here that this relationship is net of the socio-economic background of the area the school is situated in. We have also tested interaction models (results available from the authors) between qualified teachers and the content of education, but none of these interaction terms are statistically significant.

In terms of the control variables, we find that girls and students with immigrant background, on average, report lower levels of anti-immigrant attitudes. We have also controlled for age and the number of years spent in high school (not shown) and although the relationship between both age and/or the number of years spent in high school and anti-immigrant attitudes is statistically significant, inclusion of these variables does not affect the relationships between instructional exposure and anti-immigrant attitudes. There is, furthermore, no statistically significant relationship between parental level of education and students’ anti-immigrant attitudes. While the latter observation is somewhat surprising, it is nevertheless consistent with previous findings on the link between parental prejudice and children’s attitudes, according to which the relationship is not as immediate as could be expected (Aboud & Doyle, 1996), but depends on factors such as the level of attachment (Sinclair, Dunn, & Lowery, 2005) or the saliency of the issue at home (Levy & Hughes, 2009). Also, when testing for potential random effects associated with the individual-level variables, we find that the effects of these variables on adolescents’ anti-immigrant attitudes do not vary statistically significantly across classes or schools. Therefore, we choose to report the most parsimonious models without random effects.

Finally, we note here that we have tested different ways of coding the variables measuring exposure to teaching about critical thinking and multiculturalism (including, for example, binary measures collapsing ‘Yes, quite a bit’ and ‘Yes, very much’ vs. ‘No nothing’ and ‘Yes, a little’). However, the alternative coding of the variables produced similar results.

Discussion

Although previous research has established that there is a robust relationship between education and anti-immigrant attitudes, uncertainty nevertheless prevails regarding the role various aspects of education play in this relationship. In this article, we therefore examined the relationship between specific contents of education, in terms of exposure to teaching about critical thinking and multiculturalism, and adolescents’ anti-immigrant attitudes. In addition to the instructional content, we also examined whether there is a relationship between teacher qualification and adolescents’ anti-immigrant attitudes. We examined the potential relationships between these aspects of education and attitudes by using multi-level analysis and survey data collected from Swedish high school students. Specifically, we examined whether students who have received more exposure to teaching about critical thinking, different religions and cultures, as well as xenophobia and racism tend to report lower levels of anti-immigrant attitudes than students who have received less exposure to these topics. Further, we examined the potential significance of teacher qualification on attitudes by analysing the relationship between the proportion of certified teachers in schools and the anti-immigrant attitudes among students.

Our empirical findings suggest that both exposure to teaching that promotes critical thinking and introduces multicultural content, and the qualifications of the teachers, are of importance for high school students’ attitudes towards immigrants. Individual students attending classes where students on average report that they have received more teaching about critical thinking clearly report lower levels of anti-immigrant attitudes compared to students in classes where teaching about critical thinking is sparse. In line with seminal research on activation of stereotypical thinking (e.g. Devine, 1989; Devine & Elliot, 1995), we interpret this finding as support for the notion that developing critical thinking skills better equips students to overcome stereotypical thinking. This result thus suggests that promoting critical thinking in schools is important when attempting to reduce anti-immigrant attitudes among adolescents.

In addition, our findings show that exposure to multicultural teaching also matters for adolescents’ anti-immigrant attitudes. Our findings indicate that students receiving more instruction about religions and cultures (the effect associated with education about xenophobia and racism is not statistically significant when controlling for the other aspects of education) tend to report lower levels of anti-immigrant attitudes. This finding is in line with previous research based on ‘social learning theory’, showing that the content of the education in schools can help students develop more positive attitudes towards out-groups (Banks, 1991). While previous research on multicultural education mainly has focused on intervention studies, the present study adds to this body of literature by using survey data to demonstrate a general relationship between multicultural education and anti-immigrant attitudes.

In addition to the relationships found between the content of education and anti-immigrant attitudes, another important finding in this study is that teacher qualification affects adolescents’ anti-immigrant attitudes. High school students who attend a school with a high proportion of certified teachers generally report lower levels of anti-immigrant attitudes than students who attend a school with a low proportion of certified teachers. In line with previous findings which suggest that certified teachers generally uphold a higher quality of teaching (Darling-Hammond, 2000) we assume that as a result of their teacher training, certified teachers are likely to possess a better understanding of didactics, the curricula, societal norms, and values, as well as the Education act. It remains difficult to ascertain which characteristics of the certified teachers or aspect of their training are particularly important when it comes to shaping adolescents’ attitudes. Further, as we were unable to measure students’ exposure to certified teachers at the class level, some caution should be used in interpreting these results. While our findings suggest that there is a relationship, further research is needed to assess exactly ‘what’ it is about qualified teachers and their way of teaching that encourages students to be less averse to immigrants.

Our study has some limitations that are worth reviewing. One limitation concerns the self-reported nature of the measures which capture exposure to teaching about critical thinking and multiculturalism. While there is a risk that such self-reports are biased, we reiterate that we measured exposure to teaching about critical thinking and multiculturalism as aggregated class-level measures. Thus, the relationship between exposure to teaching about, for example, critical thinking and anti-immigrant attitudes is not based on individual students’ assessments of their exposure to this topic, but on the aggregated assessments of their classes. We therefore argue that the issue of biased self-reporting should not be of serious concern. Another potential limitation of this study is a reliance on single items when measuring exposure to teaching about critical thinking as well as racism and xenophobia. However, given that we find robust relationships when using single items, we might have found even stronger associations if we would have had access to multiple indicators for each aspect. In relation to this, and of course in general when using cross-sectional data, we nevertheless need to be somewhat cautious when discussing whether the associations found represent causal relationships. Further, there are limitations to the generalizability of this study with regard to other national contexts. We are aware that curricula differences do exist across different countries, which might explain, for example, why previous studies have shown mixed results regarding the relationship between multicultural education and anti-immigrant attitudes. However, it is unlikely that the relationship between exposure to teaching about critical thinking and anti-immigrant attitudes found in this study is subject to such contextual differences, indeed, critical thinking is a universal and non-biased skill.

In this article, we have shown that teaching about critical thinking and multiculturalism are of importance if schools want to foster liberal and tolerant citizens. Not only does critical thinking skills promote better academic results, it also clearly helps to fulfil the other important task of the schooling system, namely to foster citizens that can participate in the increasingly heterogeneous societies of today. Critical thinking skills clearly provide students with the ability to move beyond stereotypes and prejudice when confronted with out-group members. Even though some scholars have had difficulty directly connecting multicultural education and attitudes (e.g. Castagno, 2009), our findings indicate that instruction about religions and cultures as well as about xenophobia and racism warrant further consideration. In addition to the tendency of certified teachers to produce better outcomes in terms of students’ school performance (see e.g. Hatfield, Burchinal, Pianta, & Sideris, 2015), we argue that promoting qualified teachers might have the additional benefit of producing less prejudiced students. Overall, our results suggest that teaching about critical thinking as well as different religions and cultures and the level of teacher qualification in schools contribute to the understanding of the underlying link between schooling and anti-immigrant attitudes.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Mikael Hjerm is a professor of Sociology at Umeå University. His research interests include prejudice, nationalism and related attitudes.

Ingemar Johansson Sevä is an associate professor of Sociology at Umeå University. His research interests include political attitudes and environmental sociology.

Lena Werner is a PhD student at the department of Sociology, Umeå University. Her dissertation focuses on the relationship between different aspects of the educational system and prejudice.

Funding

This research was supported by Marianne and Marcus Wallenberg Foundation [grant number 2014.0019], Swedish Foundation for Humanities and Social Sciences [grant number P14-0775], and Swedish Research Council for Health, Working Life and Welfare [grant number 2016-07177].