Introduction
Colour blindness is a concept that is established in the US context (Bonilla-Silva 2006; Neville et al. 2000) that has gained increased attention among European scholars (Bonnet 2014; Burdsey 2011; Jansen et al. 2016). Colour blindness can be understood as “the belief that race should not and does not matter” and such a belief thereby diminishes the significance of racial group membership and its influence on life outcomes (Daughtry et al. 2020: 2). Colour blindness is understood by different scholars as a racial ideology or form of racism (Bonilla-Silva 2006); a discourse (Moras 2008) or a social attitude (Neville et al. 2000). This paper focuses on colour blindness as a social attitude. Studies over the past decade have shown the prevalence of colour-blind attitudes in Western societies and its relation to increased levels of racial prejudice (Bonilla-Silva 2006; Forman & Lewis 2015; Loya 2011; Richeson & Nussbaum 2004), and this has resulted in a growing body of literature that problematises the denial of race and racism in Europe (e.g. Goldberg 2015; Simon 2017) and the investigation of existing colour blindness in different European contexts (e.g. Rissanen 2021 investigating school principals colour-blind ideology in Sweden and Finland). There exist also studies in the European context that measure specifically colour-blind attitudes and show its prevalence among different groups. For example, Meeussen, Otten and Phalet (2014) and Jansen et al. (2016) studied colour-blind attitudes in Dutch organisations and its effect on employees’ work satisfaction and work group functioning. Moreover, Hachfeld et al. (2015) investigated colour-blind attitudes among German teachers with a focus on different aspects of professional competence when teaching immigrant students. Despite this growing body of literature that investigates colour-blind attitudes in the European context, more studies are needed that can show the prevalence of colour-blind attitudes in European countries and at the same time offer a better understanding of this concept in the European context.
Among Western countries, Sweden is known for its “exceptionalism” (Schierup & Ålund 2011). With large investments in social welfare, Sweden is often described as a universal and inclusive country with equal rights and a decent standard of living for all citizens and residents (Borevi 2013). Swedish welfare institutions play a key role in providing services to all residents, including migrants with the right to reside in Sweden. However, this tradition of socio-cultural inclusion and public welfare is increasingly questioned with more people arguing that immigration is a threat to the Swedish welfare state (Schierup, Ålund & Neergaard 2017). This larger ideological climate might also shape how welfare professionals perceive migrants (Gaucher et al. 2017). This makes Sweden and the workforce of its welfare institutions an important case to study the prevalence of colour-blind attitudes and its association to anti-immigration attitudes among welfare professionals.
This study thereby aims to understand the complex relationship between colour-blind attitudes and anti-immigration attitudes among welfare professionals. Thereby, this paper contributes to the less studied research stream of colour-blind attitudes in European societies. This article builds on existing studies by surveying two of the main Swedish welfare organisations – the Public Employment Services (PES) and the Swedish Social Insurance Agency (SSIA) – to examine the prevalence of colour-blind attitudes among welfare professionals’ and how colour-blind attitudes are associated with anti-immigration attitudes.
The article first provides a theoretical overview of colour blindness and outlines the main theoretical points on which the hypotheses are based on. In the methods section, the operationalisation of the concepts and the research design is described. In the results section, the outcomes of the regression models are discussed. Eventually, the article concludes by discussing the contributions made to the migration and ethnic and racial studies literature.
Understanding Colour-blind Attitudes
In the following section, I first provide conceptual understandings of colour blindness. Second, I account for existing studies on the implications of colour-blind attitudes in Europe. Third, the relationship between colour-blind attitudes and racial prejudice and social desirability are discussed.
Colour Blindness – An Overview
There exists a vast complexity in how the literature speaks about colour blindness. Research from the social psychology literature suggests that colour blindness might be used in good faith to advance racial harmony and equality (Babbitt, Toosi & Sommers 2016). It is put forth that people truly believe that not talking about race leads to equality and is a sign of being fair and is a way to reduce prejudice (e.g. Knowles et al. 2009). People might also avoid talking about race to not risk saying something offensive, or it might simply be a way to preserve one’s status quo when it comes to their own benefits (Babbitt et al. 2016). The latter is also adopted by critical race scholars that put forth that the adoption of colour-blind attitudes aims to sustain the status quo by ignoring racial inequality (Bonilla-Silva 2006; Forman 2004). Neville et al. (2000) define colour-blind attitudes as the ‘denial of racial dynamics.’ Their understanding and measurement of colour-blind attitudes are influenced by Frankenburg’s (1993) concepts of colour evasion and power evasion. Colour evasion is about emphasising sameness, and power evasion is about the idea that ‘everyone has the same opportunity to succeed’ (Neville et al. 2000: 60). Others also suggest that colour blindness is used as a strategy to appear unbiased and not racist (Penner & Dovidio 2016). Eduardo Bonilla-Silva (2006) argues that colour blindness is best understood as a form of racism that offers an ‘ideology that explains contemporary racial inequality as the outcome of non-racial dynamics’ (2). Forman (2004) offers a symptomology for colour blindness, arguing that ‘individuals are likely to not express their prejudices toward racial minorities explicitly but rather are more likely to express their negative feelings in ways that are subtler or covert’ (46). Other scholars link colour-blind ideology to post-raciality, arguing that it allows the illusion of equal treatment by advocating for ‘social harmony by proposing elimination of focus on [racial/ethnic/cultural] group differences’ (Goldberg 2015; Karafantis et al. 2010: 690). Therefore, the problem with adopting a colour-blind perspective is that it fails to acknowledge the significance of prejudice in society (Jones 1997). Scholars from the European context, like Hachfeld et al. (2015) emphasise equal treatment in their definition of colour blindness despite what they refer to as one’s ‘background.’ This leads us to the next section, which presents existing empirical studies in the European context.
Implications and Consequences of Colour-Blind Attitudes in Europe
Regional and historical context is important for understanding the nature of colour-blind attitudes. For historical reasons, many in Europe argue that the best way to get past racism is to simply stop using the racial lens to make sense of our daily lives. For example, Goldberg (2009) argues that the denial of race in Europe is a result of World War II and the Holocaust. For many, the use of racial terminology is associated with notions of biological inferiority and is considered dangerous and immoral (Hesse 2004). This historical context is part of the reason for the practise of “not seeing or acknowledging race” in Sweden, for example (Hübinette et al. 2012; Osanami Törngren 2015). In Sweden, colour blindness is part of an anti-racist vision that has developed after World War II, together with the view that Sweden is a proponent of social justice and gender quality (Hübinette & Tigervall 2009; Lundström & Hübinette 2020). Historically, the term race has slowly been eliminated from the public discourse after World War II to be replaced by terms like ‘ethnic group’ and ‘foreign background.’ Brännström (2016) further states that in the current societal context, referencing race is perceived as ‘suspect and uninformed’ (50).1 Therefore, in the Swedish context, colour blindness is understood as a wish to not see colour and to avoid the term race. Race should simply have no meaning in Sweden, and racism is portrayed as an issue that ‘other countries’ have (Lundström 2021). One example is the removal of the term ‘race’ from Swedish anti-discrimination legislation and the Ministry of Integration and Gender Equality (2009) positing that including the term race in the legal text could be perceived as legitimising racist ideas (Reisel 2015). This shows how much colour blindness is used as the default approach in society to then be picked up through social norms (Apfelbaum, Norton & Sommers 2012).
While some believe that colour blindness in Europe has a more benign and historically unique function and interpretations, there are also studies that suggest it may operate the same in both the United States and Europe (e.g. Hachfeld et al. 2015; Meeussen et al. 2014). Several studies exist that help to understand colour blindness in different European contexts. Existing studies show that colour-blind logic is present in policies and laws that assess racial and ethnic discrimination (see for instance, Simon 2010; Simon 2017 for the French context and Brännström 2018 for the Swedish context). Another study shows that French police officers and security guards make use of colour-blind speech norms when reporting on their policing practises with minority youths. However, the interviewees in that study also show race consciousness in their work suggesting that a colour-blind discourse does not prevent people from engaging in race-conscious practises (Bonnet 2014). Existing studies also investigate diversity policies in Belgian schools to explore the meaning of colour blindness in an educational context. Celeste et al. (2019) show that a colour-blind approach in schools is related to a lower feeling of belonging and lower educational outcomes among majority and minority group students. Empirical studies from the Swedish context, for example, show the use of colour-blind language when talking about mixed marriages and the migration experiences of Swedish expats. By using colour-blind language, privilege is expressed while downplaying whiteness, which leads to further ‘us’ and ‘them’ dichotomies in society (Lundström 2021; Osanami Törngren 2019). While these studies describe how colour blindness might look ‘on the ground’ in various European countries, fewer studies assess the prevalence and consequences of colour-blind attitudes in Europe. One study of Dutch leaders’ diversity perspective examined the role of colour-blind attitudes opposed to multiculturalism showing that colour-blind attitudes among leaders predicted minority group members distancing from the work group (Meeussen et al. 2014). In line with that study, Jansen et al. (2016) investigate the role of colour-blind attitudes and multiculturalism in work satisfaction in Dutch organisations. The authors find that colour-blind attitudes are equally prevalent among majority and minority groups, but it only increases work satisfaction for the majority group, indicating that colour-blind attitudes do not result in equal outcomes in work life. Another study investigates colour-blind attitudes among German teachers with a focus on different aspects of professional competence when teaching immigrant students (Hachfeld et al. 2015). This study finds that colour-blind attitudes are negatively related to adapting teaching to ethnically and culturally diverse students. Overall, existing studies show that colour-blind attitudes can have negative implications for individuals in different European contexts.
Colour-Blindness, Social Desirability and Prejudice
Existing studies suggest that there exists a link between self-reported colour-blind attitudes and becoming more socially desirable. Social desirability reflects the ‘tendency to give overly positive self-descriptions’ that fit current social norms (Paulhus 2002: 50). Professionals and people with a higher educational level tend to be more aware of current norms and tend to feel more pressure to express themselves in a more socially desirable way (Krysan 1998). People tend to indicate an unrealistically positive self-description in order to feel more confident, but also for the pursuit of power and control. Knowles et al. (2009) show that white people who are more egalitarian are more likely to endorse colour-blind attitudes. Appearing more egalitarian is also associated with preserving a positive self-image, and therefore colour blindness and social desirability likely co-occur, as both represent ways to fit within the norms of society. Moreover, Tourangeau and Yan (2007) put forth that people are more likely to answer in socially desirable ways when responding to sensitive questions concerning their beliefs, and therefore social desirability can also be linked to holding prejudice. This means that the more socially desirable someone wants to come across the less prejudice they will express.
While reporting colour-blind attitudes is socially desirable, studies from the US context show that colour blindness is associated with holding racial prejudice (e.g. Bonilla-Silva 2006; Neville et al. 2013). For instance, Richeson and Nussbaum (2004) show that when white college students were exposed to vignettes that primed them to think about being colour-blind, they reported higher scores on the implicit racial attitude test. Ayford and Zaaimans’ (2021) study also confirms this association for the South African context through a survey study examining students’ colour-blind attitudes and racial prejudice. Moreover, several studies show that people often use colour-blind scripts as part of a strategy to appear unbiased in social interactions (e.g. Apfelbaum, Sommers & Norton 2008; Babbitt et al. 2016). That is, appearing colour-blind or unbiased is part of an effort to ‘not come across as racist’ and keeping a positive self-image (Cabrera 2014: 3).
Research Hypotheses
Based on the theoretical understandings of colour-blind attitudes, in the following section, I propose the research hypotheses for this study. Given the need to better understand colour-blind attitudes among welfare professionals and its link to anti-immigrant attitudes, this study tests for associations between social desirability, colour-blind attitudes and anti-immigration attitudes among Swedish welfare professionals. Given the literature on the positive association between colour blindness and prejudice (Apfelbaum et al. 2008; Richeson & Nussbaum 2004), it is expected that those who report more colour blindness would also harbour anti-immigrant attitudes. It is also suggested that greater levels of social desirability are associated with more colour blindness but less anti-immigration attitudes (Knowles et al. 2009; Tourangeau & Yan 2007).
This study focuses on welfare professionals, and therefore the specific hypotheses are
Welfare professionals who hold more colour-blind attitudes express more anti-immigration attitudes compared to those who hold less colour-blind attitudes.
Welfare professionals who express higher levels of social desirability express more colour-blind attitudes compared to those who express less social desirability.
Welfare Work and Migration in Sweden
Before introducing the data and method of the study, I present the Swedish context in relation to migration, racial prejudice and welfare work.
A vast body of literature demonstrates that welfare professionals hold prejudice against welfare recipients that are part of a minority group across different contexts (e.g. Arai et al. 2016; Park et al. 2011; Pedersen, Stritch & Thuesen 2018). When it comes to the Swedish context, there exist qualitative as well as quantitative studies that show that welfare professionals hold anti-immigration attitudes that in turn lead to less support for minorities than for the dominant in-group (Arai et al. 2016; Eliassi 2014; Johansson & Molina 2002; Larsson 2015; Schütze 2020). Despite existing studies that link welfare professionals’ racial attitudes to the provision of welfare services, we know little about the prevalence and relation of colour-blind attitudes and anti-immigrant attitudes among welfare professionals.
Immigration is an important part of the Swedish welfare state and welfare work. Immigrants constitute a substantial portion of the Swedish population, with more than 20% being foreign born in 2020 (Statistics Sweden 2021). Sweden has over 70 years of immigration history. Many refer to the years following World War II, 1950s and 1960s, as the labour immigration period. After this period, from the 1970s until about 1985, immigration in Sweden transitioned from labour migrants to refugees and their families, who were coming mostly from Asia and South America (e.g. Turkey, Iran, Lebanon, Syria, Chile and from Eastern European countries; Lundh & Ohlsson 1999). In the 1990s, due to the conflicts in the former Yugoslavia, Sweden received many refugees from the Balkan region. Recent immigration is marked by the war in Syria. In 2020, the immigrant population in Sweden consisted mostly of people coming to Sweden from Syria, Finland, Iraq, Poland and Somalia (Statistics Sweden 2021). In 2015, Sweden’s welcoming migration policy changed. In the late summer of 2015, Sweden opened its borders to 162,877 asylum seekers, representing 12.4% of all applications in the European Union. This lasted for almost six months (Barker 2017; Emilsson 2018). This short period of increasing numbers of asylum seekers resulted in dramatic changes in Swedish migration policies, shifting from one of the most welcoming countries to being overly restrictive, including closed borders, more temporary resident permits, and limited opportunities for family reunification (see also Schierup et al. 2017).
Methods
Survey data was collected from Swedish welfare professionals working in two of the main welfare organisations in the country – the PES and SSIA. The study was approved by the Regional Ethical Board in Lund, Sweden in 2016. The PES and the SSIA both offer aid to the entire Swedish population, including foreign-born legal residents. Both the PES and SSIA are state run. The PES has the general mandate to give assistance to the unemployed while collaborating with the SSIA on work-oriented solutions for people who have reduced working capacity. The PES is also responsible for providing newly arrived refugees who have legal resident permits2 with training and competence on integrating into the Swedish labour market through the “establishment program.” The SSIA provides financial security at various stages of life for everyone who lives and/or works in Sweden. This includes benefits and allowances for families with children, people who are ill and people with disabilities (Mathias 2017; SSIA 2017).3 The SSIA also distributes benefits to those taking part in PES’s labour market and establishment programmes (SSIA 2017). Both organisations had about 14,000 employees at the time of the data collection with offices throughout the country, and about half of the workers are members of a union that served as the sampling frame for the study (Kalton 1983).
The sampling frame for the survey came from one of the main public unions in the country. The union includes 50% of the employees at both organisations (Kalton 1983).4 In November 2016, I sent a 36-question web survey5 to a random sample of 6,650 welfare workers using employee email accounts. A pilot study and meetings with representatives of both welfare organisations helped to identify the everyday concepts used by welfare professionals and their work with migrants. A total of 1,617 respondents answered, resulting in a response rate of 24.3%. This response rate is normal for web surveys, which are usually between 20% and 30% (Dillman, Smyth & Christian 2014; Kaplowitz, Hadlock & Levine 2004). Respondents who did not qualify as welfare professionals were dropped from the sample (e.g. union members working with IT issues), resulting in a final working sample of 1,319 respondents (the procedure of removing respondents was also used by Liozu & Hinterhuber 2013). The population of PES and SSIA employees in the union was 28% men and 72% women with an average age of 47.7 (SD 11.1) years 6 at the time the survey was conducted, and 14.9% were foreign born, and this reflects the survey respondent data almost identically.
Measures
Dependent Variable
Anti-immigration attitudes
Anti-immigration attitudes are assessed as a latent construct composed of six items that have been used in previous studies (e.g. Hjerm 1998): (a) Immigrants improve Swedish society by adding new ideas and cultures, (b) immigrants increase crime rates, (c) immigrants take jobs away from people who were born in Sweden, (d) immigrants are good for the Swedish economy, (e) Sweden should take stronger measures to exclude illegal immigrants and (f) immigrants are a burden for Sweden. In a Swedish context, racial categories like ‘Black, White, Latino’ do not apply to societal stereotypes. Instead, one refers to ‘immigrants’ when marking a minority group that the larger population holds stereotypes against (Andersen & Guul 2019). The four response options of the items range from fully disagree to fully agree (1 = less negative attitudes; 4 = more negative attitudes) and items a and d were reverse coded so that all items were in the same direction. As a robustness test, an exploratory factor analysis was conducted that resulted in a one-factor solution with an eigenvalue of 3.25 that explained 55% of the variance in the variables. The construct has a Cronbach’s alpha of 0.81, which means it has good reliability and the item had a mean of 1.8 (SD = 0.6). With regards to missing data, 6.6% of the values of this item were missing.
Independent Variables
1. Colour-blind attitudes
The measure of colour-blind attitudes is adapted from the ‘Colour-blind racial attitude scale’ (CoBRAS), which was developed and validated by Neville et al. (2000). The scale contains 20 items and is divided in 3 sub-scales: racial privilege, institutional discrimination, and blatant racial issues. CoBRAS is influenced by Frankenburg’s (1993) earlier mentioned concepts of colour and power evasion. The latter, the idea that ‘everyone has the same opportunity to succeed’ (Neville et al. 2000: 60) and if ethnic and racial minorities fail to do so, it is due to their individual characteristics, is in line with the principle of equality of the Swedish welfare state (Barker 2017). Based on that, three items are adapted reflecting the power evasion concept. The measures are adopted for the Swedish context to be more context specific and to assure that respondents can relate to the items (Bandura 1986) by replacing the term “Whites” with “Swedes” and the term “race” with “ethnic background.” Colour blindness is operationalised as a latent variable composed of the following items: (a) Swedes have certain advantages in society because of their ethnicity, (b) migrants have not the same possibilities as Swedish people and (c) the colour of one’s skin influences people’s possibility in society. The response options range from fully disagree to fully agree, and all items were reverse coded so that all items were in the same direction (1 = less colour-blind; 4 = more colour-blind). As a robustness test, an exploratory factor analysis was conducted that resulted in a one-factor solution with an eigenvalue of 2.09 that explained 70% of the variance in the variables. The construct has a Cronbach’s alpha of 0.78, which means it has good reliability. The latent construct has a Cronbach’s alpha of 0.78, which means it has good reliability and the item had a mean of 1.9 (SD = 0.7). With regards to missing data, 5.9% of the values of this item are missing. Moreover, to ensure that the measurements for colour-blind attitudes and anti-immigration attitudes are empirically unique constructs, an exploratory factor analysis was conducted with all items of the colour-blind attitude measurement and the anti-immigration attitude measurement that resulted in a two-factor solution with respective factor loadings belonging to anti-immigration attitudes and colour-blind attitudes.
2. Social desirability
The concept social desirability is assessed as a latent construct composed of four items that captured the exaggeration of positive qualities and understate negative qualities (Kemper et al. 2012)7: (a) Although I feel stressed, I am always kind and courteous to others, (b) it has happened that I have taken advantage of someone, (c) in an argument, I always remain objective and adhere to the facts and (d) when I talk to someone, I always listen carefully to what the other person has to say. Social desirability is included because respondents are more likely to show a social desirability bias when answering questions concerning their values and beliefs (Tourangeau & Yan 2007). Five answer options range from very true to not correct at all. Items a, c, d, e are reverse coded so that increases reflect a greater interest in desirability (1 = less socially desirable; 5 = more socially desirable). As a robustness test, an exploratory factor analysis was conducted that resulted in a one-factor solution with an eigenvalue of 1.87 that explained 47% of the variance in the variables. The construct has a Cronbach’s alpha of 0.62, which means it has good reliability and the item had a mean of 4.14 (SD = 0.5). With regards to missing data, 1.2% of the values of this item were missing.
Control Variables
Control variables include demographics such as age, gender (female = 1), nativity (foreign born = 1); education, which is measured in the four following categories: (a) elementary and high school, (b) vocational training, (c) single university courses and (d) university, where the category “university” served as the reference category when creating three dummy variables and organisational context (PES = 1) (Chen, Moskowitz & Shue 2016; Park et al. 2011; Watkins-Hayes 2009). Overall, among the control variables, few values are missing among the items (minimum 0.15% to maximum 1.29%).
Empirical Strategy
The data is analysed by making use of ordinary least squares (OLS) regressions with anti-immigration attitudes as the dependent variable. The descriptive results of the variables used in the analysis can be found in Table A1 in the appendix. A correlation matrix between the dependent variable and the independent variables is also conducted (see Table 1). In the next step, three sets of OLS regression models are applied to elicit which independent variables are associated with anti-immigration attitudes with a focus on the relation between colour-blind attitudes and anti-immigration attitudes. A hierarchical linear regression model is applied. This is a framework whereby several regression models are built by adding new variables to each model at each step (Kim 2016). The first set included the control variables (Model 1). The second set added colour-blind attitudes (Model 2) and the third set added social desirability (Model 3). The results are shown in Table 2.
Table 1
Correlation matrix of all variables.
| 1. | 2. | 3. | 4. | 5. | 6. | 7. | |
|---|---|---|---|---|---|---|---|
| 1. Female | – | ||||||
| 2. Age | –0.09** | – | |||||
| 3. Foreign born | 0.01 | –0.12*** | – | ||||
| 4. PES | –0.23*** | 0.09** | 0.07* | – | |||
| 5. Social desirability | –0.03 | 0.09* | 0.09** | 0.06* | – | ||
| 6. Colour-blind attitudes | –0.02 | 0.18*** | 0.00 | 0.02 | 0.18*** | – | |
| 7. Anti-immigration attitudes | –0.04 | 0.09** | –0.01 | –0.01** | –0.01 | 0.29*** | – |
[i] Note: * p < 0.05; ** p < 0.01; *** p < 0.001.
Table 2
OLS regression.
| ANTI-IMMIGRATION ATTITUDES | MODEL 1 | MODEL 2 | MODEL 3 |
|---|---|---|---|
| Control variables | |||
| Age | 0.004* | 0.001 | 0.002 |
| Male | Ref | Ref | Ref |
| Female | –0.086* | –0.083* | –0.084* |
| Non-foreign born | Ref | Ref | Ref |
| Foreign born | 0.018 | 0.03 | 0.015 |
| University | Ref | Ref | Ref |
| Elementary_Highschool | 0.100* | 0.056 | 0.064 |
| Single University courses | 0.106 | 0.086 | 0.089 |
| Vocational training | 0.019 | –0.021 | –0.120 |
| Organisational Context | |||
| SSIA | Ref | Ref | Ref |
| PES | –0.144*** | –0.153 | –0.149 |
| Colour-blind attitudes | 0.233*** | 0.242*** | |
| Social desirability | –0.077*** | ||
| N | 1160 | 1160 | 1160 |
| Adj. R2 | 0.02 | 0.10 | 0.11 |
[i] Note: * p < 0.05; ** p < 0.01; *** p < 0.001.
Results
Table 1 shows the pairwise correlation between all variables. The descriptive data (in Table A1 in the appendix) shows that the mean of the anti-immigration attitudes as well as the colour-blind attitude are close to the midpoint (two) on a scale from one to four. It is also shown that colour-blind attitudes are associated with anti-immigration attitudes. It furthermore shows that colour-blind attitudes and social desirability are associated. The bivariate correlation shows no association between social desirability and anti-immigration attitudes.
Hypothesis 1 states that welfare professionals who hold more colour-blind attitudes, express more anti-immigration attitudes than those who hold less colour-blind attitudes, this hypothesis is supported. Colour-blind attitudes are positively associated with anti-immigration attitudes. A one unit increase in the colour-blind index was associated with an increase of 0.24 (p > 0.001) units on the anti-immigration attitudes index when controlling for background variables and including social desirability into the model. Meaning, welfare professionals that express more colour-blind views also hold more prejudice against migrants.
Hypothesis 2 states that welfare professionals who express more social desirability, express more colour-blind attitudes than those who express less social desirability, this hypothesis is supported. Social desirability is positively associated with colour-blind attitudes as seen in the correlation matrix in Table 1. The simple regression shows that a one unit increase in the social desirability index was associated with an increase of 0.18 (p > 0.001) units on the colour-blind attitudes index without including any control variables. Meaning, welfare professionals that answered the survey in a more socially desirable way also expressed more colour-blind views.
The findings also show that social desirability is negatively associated with anti-immigration attitudes. A one unit increase in the social desirability index was associated with a small decrease of 0.07 (p > 0.001) units on the anti-immigration attitudes index when controlling for background variables and colour-blind attitudes. Meaning welfare professionals that answered the survey in a more socially desirable way expressed less prejudice towards migrants.
The findings furthermore suggest that adjusting for social desirability is important when considering the relationship between colour-blind attitudes and anti-immigration attitudes. While there is no association between social desirability and anti-immigration attitudes when conducting the bivariate correlation, this association is significant in the full model when colour-blind attitudes are included. This finding suggests that colour blindness and social desirability co-occur in relation to anti-immigration attitudes.8
Table 2 shows that model 3 has the best model fit with an adjusted R2 of 0.11 (11%). In comparison, Model 1 shows an adjusted R2 of 2%, which increases to 10% when colour-blind attitudes are introduced into the model and increases further to 11% when social desirability is introduced into the model, indicating that these two associations can be identified as important.
Discussion
The aim of this article is to examine the prevalence of colour-blind attitudes among welfare professionals’ and how these attitudes are associated with anti-immigration attitudes but also social desirability with large-scale quantitative data. This study thereby tests how colour-blind attitudes, a concept from the US context, can be applied to a Swedish welfare institutional context and finds convergent results. The growing numbers of studies about colour blindness in Europe show the increasing importance to study colour-blind attitudes in European societies (e.g. Jansen et al. 2016; Meeussen et al. 2014; Simon 2017) but more empirical studies are needed. This study confirms previous research in the US context showing that holding colour-blind attitudes is associated with elevated levels of prejudice (Apfelbaum et al. 2008; Bonilla-Silva 2006; Richeson & Nussbaum 2004). Thereby, this study also expands this research to a new national context. Focusing on welfare professionals in Sweden, this study shows that professionals with more colour-blind attitudes have an increased bias towards migrants. The study thereby contributes to the cross-national and cross-contextual validation of the existing research on colour blindness but also offers a novel contribution by empirically showing an association between colour-blind attitudes and social desirability.
The empirical results show that colour-blind attitudes are associated with anti-immigration attitudes. This means that welfare professionals who hold more colour-blind attitudes also hold greater bias towards migrants. This confirms previous studies from the US context (e.g. Bonilla-Silva 2006; Richeson & Nussbaum 2004). It is suggested, as others have proposed (e.g. Penner & Dovidio 2016), that when people work to present themselves as colour-blind, in most cases, this is best understood as a way to appear egalitarian and unbiased even while also holding negative attitudes towards migrants. In their role as representatives of the state, welfare professionals might think it is important to be perceived as extensions of the Swedish social and organisational norms, which include the ideal of “colour blindness” where racial markers, such as the skin-colour, should not influence one’s chances to succeed in society (Bonnet 2014). In that sense, colour-blind attitudes represent “interactive social norms” that one deploys to be in line with societal norms (Jackson, Wilde & Goff 2016: 130).
The results presented here also show that social desirability is associated with colour-blind attitudes, even though the association is rather weak. This means that welfare professionals that portray themselves in a more socially desirable way also hold more colour-blind attitudes. This finding is important since this association has not been tested empirically previously and thereby adds new insights into understanding the concept of colour-blind attitudes. This result suggests that appearing colour-blind and emphasizing equal opportunities for everyone despite their ethnic and racial background co-occurs with wanting to convey a positive self-image (Cabrera 2014). This finding suggests a specificity of understanding colour blindness in Sweden, a way to communicate a positive self-image that fits in its anti-racist discourse and positive self-image as a country championing human rights and equality (Lundström & Hübinette 2020).
While the findings of this study improve our understanding of the relationships between colour blindness, anti-immigration attitudes and social desirability for the European context, this study has several limitations that research might address in the future. While one of the strengths of this study is the original survey data, there is room for improvement with respect to the study measures. For example, colour-blind attitudes were not measured using the full CoBRAS of Neville et al. (2000) and the measurement was adjusted to the Swedish context. This means that racial categories were partly replaced by non-racial categories such as ethnicity and Swedish (skin colour was, however, referred to) to avoid non-responses due to a language that respondents might be unfamiliar with or perceive as offensive (Simon 2010). This of course in itself shows the colour-blind norms present in Sweden and might have influenced the outcome of the analysis. One can speculate that the relationship between colour blindness and anti-immigration attitudes was amplified due to the non-racial terminology in the measurement, but only further studies that test measurements of colour blindness using racial and non-racial terminology could confirm this empirically. Moreover, since very little is known about how colour blindness plays out for different ethno-racial groups in Sweden, a more nuanced operationalisation of the category foreign-born could be revealing and productive. Additionally, one might reason that countries like the United States, with a long history of ethnic and racial diversity might differ in their pattern of interactions when it comes to racial attitudes compared to a country like Sweden, where diversity has increased rather recently. Yet the findings of this article show converging results similar to Hachfeld et al. (2015), which examined colour-blind attitudes in the German context as well as Meeussen et al. (2014), which examined colour-blind attitudes in the Dutch context. Therefore, I suggest that the underlying relationship between colour blindness and elevated levels of prejudice ‘travels well’ and seems to be an important association in need of further investigation beyond the US context. Therefore, future studies should consider additional settings.
Beyond its empirical findings, the article also has implications for practical welfare work. Namely, welfare professionals’ colour-blind attitudes might influence interactions with minority groups seeking these services. For example, Jackson et al. (2016) suggest that avoiding race might lead to less friendly behaviour, but also that endorsing colour blindness might reduce our ability to recognise racial discrimination. Earlier research also implies that this effort might result in decreased quality of welfare services for minorities due to a preoccupation with “seeming unbiased” – rather than focusing on the needed service at hand – resulting in appearing more anxious during interactions (Penner & Dividio 2016; West & Schoenthaler 2017). Future studies should measure colour-blind attitudes among welfare professionals and put these into relation with measurements capturing interactions and concrete outcomes of the day-to-day welfare work with minority groups.
Conclusion
Despite its limitations, this study contributes to a nuanced discussion of colour-blind attitudes in the European context. This study contributes to research on colour blindness by confirming existing assumptions that colour blindness is associated with elevated levels of anti-immigration attitudes. It also offers direct evidence of a previously speculated association between social desirability and colour blindness, suggesting that both concepts are used as a strategy to promote a positive self-image that is in line with societal norms and standards. Since it appears that colour blindness has negative implications for minorities and intergroup relations, given the increasing preference for thinking of oneself as colour-blind among people living in multiracial societies, we need more research into the prevalence of colour-blind attitudes across different contexts and the relationship between colour blindness and the practical welfare work with minority groups as these trends might shape their access to welfare benefits.
Appendices
Appendix
Table A1
Descriptive statistics.
| MEAN (SD) | |
|---|---|
| Age | 47.7 (11.1) |
| Female | 71.8% |
| Foreign born | 14.9% |
| University | 58.1% |
| Elementary/High school | 26.3% |
| Vocational training | 5.0% |
| Single university courses | 10.6% |
| PES | 56.3% |
| Colour-blind attitudes (1–4) | 1.9 (0.7) |
| Anti-immigration attitudes (1–4) | 1.8 (0.6) |
| Social desirability (1–5) | 4.14 (0.5) |
| Total number in the sample | 1,319 |
Notes
[3] See Brännström 2016 for a comprehensive overview of the development of the term race in Sweden after World War II.
[5] In 2016, 42% of the registered clients at the PES were foreign born (PES 2017). Similar estimates are not available for the SSIA, but since the SSIA gives services to clients registered at the PES, we can assume that both organisations have a considerate number of migrants as clients.
[6] In 2016, approximately 82% of all civil servants in the public sector in Sweden were members of a union (Kjellberg 2017). This means that other welfare workers within these two institutions are part of other unions. In Sweden, union membership is classified as sensitive information and therefore there exists no exact estimation of how many welfare workers within these two institutions were union members. Yet, we can assume that most welfare workers are union members. The union that served as the sample frame is also the largest union for PES and SSIA union members. Moreover, in Sweden, most employees are unionised.
[9] The social desirability scale by Kemper et al. (2012) consists of a two-factor item model; exaggeration of positive qualities and minimisation of negative qualities. We made use of the former factor item only.
Acknowledgements
I want to thank Mosi Ifatunji for his generous engagement in the initial stage of this project and continuous support and inspiration when writing this article.
Competing Interests
The author has no competing interests to declare.