Outgroup mobility threat – how much intergenerational integration is wanted?

ABSTRACT

The descendants of immigrants have become increasingly successful in Western European education systems and labor markets, which in principle supports Richard Alba’s new assimilation theory (NAT). Still, acceptance and recognition on the part of the mainstream has not kept pace. We have suggested outgroup mobility threat (OMT), the fear of being overtaken by intergenerational mobility, as one possible explanation. However, Richard Alba has argued that demographic changes widely lead to a non-zero-sum mobility and do not necessarily cause competition threat. In this paper, we relate Richard Alba's reasoning to the concept of OMT, focusing on the German immigration context. We bring together the arguments by distinguishing between realistic and symbolic threat and by differentiating between occupations and groups. We measure OMT and test our hypotheses using a vignette study that we recently conducted as part of the kick-off survey of the new National Discrimination and Racism Monitor (NaDiRa) in Germany. Our results confirm that Richard Alba's argument is superior to the simple competitive threat perspective. However, we also show that it is still insufficient, and OMT comes into play when immigrant groups move up into value-based occupations that are assumed to normatively shape society. Moreover, we demonstrate that this threat is felt particularly when Muslims ascend to these types of occupations.

KEYWORDS:

• New assimilation theory
• anti-Muslim attitudes
• integration paradox
• symbolic threat
• racism

Introduction

Looking at recent trends in contemporary Western European migration societies, research reveals an interesting and somewhat puzzling pattern of developments: On the one hand, there is strong, not to say overwhelming, evidence that integration, in its structural, cultural, and social components, is clearly progressing over time and especially across generations. This is true for different countries and for very different aspects of integration (Alba and Foner 2015; Kalter et al. 2018), including the structural domains of education systems and labour markets (Heath and Cheung 2007; Heath and Brinbaum 2014), which are widely considered to be of central importance. On the other hand, anti-immigrant attitudes, essentially based on the narrative that immigrants and their descendants would be unwilling or unable to integrate, appear to be remarkably persistent (Czaika and Lillo 2018; Dennison and Dražanová 2018), if not even on the rise (Meuleman, Davidov, and Billiet 2018; Küpper, Krause, and Zick 2019). This is especially true for anti-Muslim attitudes (Bell, Valenta, and Strabac 2021; Bertelsmann Stiftung 2015; Decker et al. 2022, 72; Gorodzeisky and Semyonov 2019), which have played a dominant role in European discourse in recent decades. Hostile attitudes toward Muslims are often most pronounced when compared to those toward other minority groups (Zick and Krott 2021).

These trends and observations are also ambiguous in the light of new assimilation theory, as put forward very prominently by Richard Alba. Basically, it is argued that assimilation as a revised and refined analytical-descriptive concept, understood as the decline of ethno-racial distinctions and their impacts, is a very likely process, especially in the structural realm (Alba and Nee 1997; 2003). This is widely supported by research (Drouhot and Nee 2019; Waters and Pineau 2015; White and Glick 2009). However, Richard Alba and his co-authors have also emphasised the two-sided nature of the process and that it also includes acceptance on the side of mainstream society (Alba and Foner 2015). The fact that intergenerational mobility and structural assimilation trends are not accompanied by respective recognition tendencies on the part of the dominant society thus poses a challenge also to new assimilation theory.

In previous papers, we have suggested outgroup mobility threat (OMT) as a mechanism that can contribute to the list of possible explanations for recognition gaps, especially for anti-Muslim attitudes (Foroutan et al. 2019; Foroutan and Kalter 2021). OMT reflects a variant of the ‘integration paradox’ (Verkuyten 2016) and means that the dominant society, or what Alba calls the ‘mainstream’, may develop negative attitudes toward immigrant and minority groups precisely because they are increasingly better structurally integrated. Intergenerational mobility can lead to a threat of being overtaken. But, as Richard Alba has also emphasised in much of his recent work, things are not so simple. It is questionable whether the structural mobility of the descendants of immigrants really causes competition and threat. Overall demographic changes are leading to what Richard Alba has called ‘non-zero-sum’ mobility (Alba 2020; Alba and Maggio 2022). But this makes the persistence of hostile attitudes towards minorities even more puzzling.

In this paper, we confront threat-theoretical arguments with Richard Alba's non-zero-sum argument and seek to develop these concepts and our construct of OMT further to understand the puzzle outlined. Building on the distinction between realistic and symbolic threat (Stephan and Stephan 2000) and related theoretical approaches, we argue that it can make a difference whether minority mobility is only about relative resources or whether it also affects values, norms, and habits, i.e. the shape of society and its ‘rules of the game’.

In the following, we explain and develop the theoretical ideas in more detail. We then derive hypotheses about which occupational characteristics might have an impact on OMT and for which minority groups this impact might be stronger. We measure OMT and test the hypotheses using a vignette study that we recently implemented in a submodule of the new National Discrimination and Racism Monitor (NaDiRa) in Germany, which is conducted by the German Center for Integration and Migration Research (DeZIM Institute). We present our main findings and discuss their implications for Richard Alba’s reasoning.

Theory and past research

New assimilation theory and its evidence

New assimilation theory as proponed by Richard Alba (Alba and Nee 1997; 2003) has made many fundamental clarifications and corrections to classical assimilation theory (e.g. Park 1950; Gordon 1964). Above all, it has emphasised that assimilation – in its core defined as ‘the attenuation of an ethnic or racial distinction and the cultural and social differences that are associated with it’ (Alba and Nee 1997, 834) – is to be understood as an analytical, descriptive concept, and that two accentuations are especially important: First, assimilation is no longer seen as an inevitable, law-like quasi-linear process, but as contingent, and outcomes may vary across contexts, groups, and dimensions of assimilation. However, it is still argued that, given general mechanisms of action and interaction (Nee and Alba 2013; Kalter 2022), there is much reason to believe that assimilation is very likely under the conditions of most modern migration societies, especially with respect to the important structural dimension. Second, it is stressed that assimilation is to be understood as a two-way process and not as a unilateral adaptation on the side of the immigrant minorities only.

In later work, Richard Alba has also occasionally switched to the concept of integration. He and Nancy Foner have argued that there is considerable overlap between the revised assimilation concept and the concept of integration, which is commonly favoured in the European context (Alba and Foner 2015). Within this outline, they have suggested the following definition:

‘Integration’, as we understand it, refers to the processes that increase the opportunities of immigrants and their descendants to obtain the valued ‘stuff’ of a society, as well as social acceptance, through participation in major institutions such as the educational and political system and the labor and housing market. (Alba and Foner 2015, 5)

This accentuates the fact that the second point above, the two-sidedness of the process, also entails recognition by the dominant society. This provides a potential theoretical link to approaches that emphasise worth and cultural membership in inequality research (Lamont 2018). Acceptance and recognition indeed seem indispensable for a contemporary understanding of integration. We have suggested conceptualising integration in a wider sense as the individual, social, political, economic, legal and symbolic processes of establishing recognition, equal opportunities and participation for all citizens of a society, regardless of their cultural, ethno-racial, religious or national origin (Foroutan and Kalter 2022, 153).

It is safe to say that the first key view of new assimilation theory is strongly supported by research on mobility processes among disadvantaged minorities in Western societies (Alba and Foner 2015; Kalter et al. 2018), especially with respect to the structural areas of education and the labour market (Heath and Brinbaum 2014; Heath and Cheung 2007). Intergenerational trends are not least driven by high aspirations (Kao and Tienda 2012; Salikutluk 2016) and positive choice effects (Jackson, Jonsson, and Rudolphi 2012; Dollmann 2017) that contradict the narratives of the unwilling, non-integrable minorities who would finally end up as fiscal burdens, as employed by the populist right. At the same time, however, these right-wing populist parties have successfully built election campaigns on anti-immigrant strategies, and negative attitudes against immigrants have risen in many countries rather than declined (Meuleman, Davidov, and Billiet 2018; Küpper, Krause, and Zick 2019). So the recognition processes on the side of the mainstream population have by no means kept pace with the structural mobility processes on the side of most minorities.

Outgroup mobility threat and non-zero-sum mobility

There are many factors and arguments to explain anti-immigrant attitudes and some exogenous trends could be natural confounders to the recognition of the structural successes of the minorities. Explanations refer, for example, to economic crises (Billiet, Meuleman, and Witte 2014; Kuntz, Davidov, and Semyonov 2017; Meuleman, Davidov, and Billiet 2018) or growing polarisation due to the dynamics of social or traditional media (Štětka, Mihelj, and Tóth 2021). An interesting line of argumentation regards the trends as something that might partly be endogenous to the processes of structural integration. Some scholars argue that expectations on both the migrant and non-migrant side increase as adaptation progresses (Dixon et al. 2010; Tolsma, Lubbers, and Gijsberts 2012; Foroutan 2012; Ten Teije, Coenders, and Verkuyten 2013; Canan and Foroutan 2016; El-Mafaalani 2018). The migrants and their descendants perceive lack of recognition more strongly when they are better integrated and react more sensitively to devaluations and a lack of equality which can lead to a kind of ‘integration paradox’ when they withdraw into ethnic enclaves (Verkuyten 2016; De Vroome, Martinovic, and Verkuyten 2014; Schaeffer 2014; El-Mafaalani 2018) or use strategies of identity politics (Attia 2014; Dotson 2018). But there might also be another understanding of ‘integration paradox’ redirecting it towards the non-immigrant majority society that reacts with aversion and fear of status when immigrants integrate and ascend into power (Sutterlüty 2010; Foroutan 2019, 145f; El El-Mafaalani 2018). In other words, some members of the majority population may develop negative attitudes toward immigrant and minority groups, not because they are unable or unwilling to integrate, but precisely because they succeed in doing so.

This connects to traditional theories of ethnic competition and threat (Sherif and Sherif 1969; Blalock 1967; Blumer 1958), which assume that there is a basic conflict over scarce resources. In its classical understanding, however, the threat is usually related to new immigrants entering a society. To address the specific threat that might stem from the intergenerational mobility of longer settled immigrant groups, we suggested the concept of outgroup mobility threat (OMT). OMT, in other words refers to threats that are caused by the growing structural success of migrant and minority groups. The concept relates to Honneth and Sutterlüty (2011), who have already addressed the idea that the rise to power of formerly marginalised groups is only accepted to a certain extent. We were able to show that OMT helps us to understand anti-Muslim attitudes in Germany independently of other factors (Foroutan et al. 2019; Foroutan and Kalter 2021).

However, in much of his recent work Richard Alba has emphasised that there should be a strong demographic force working against competition threat mechanisms in most Western societies (Alba 2020: Alba and Maggio 2022). As the dominant cohorts of the baby boomers are about to leave the labour market and as the younger cohorts have become increasingly diverse, many vacancies will appear, fostering ample opportunities for structural assimilation. This holds true also for the labour market in Germany (Baumann et al. 2019). Richard Alba speaks of ‘non-zero-sum mobility’, emphasising that there is no longer tight competition as the formerly largely White mainstream will not be able to refill these positions anyway.

Given these seemingly conflicting lines of reasoning – OMT vs. non-zero-sum mobility –, it is useful to refine and develop the arguments and concepts further. A fruitful starting point is to ask whether the logics of outgroup mobility threat or non-zero-sum mobility apply in principle to all kinds of occupations, or whether the arguments apply only to some but not to others. Here, the distinction between realistic and symbolic threat within the general framework of threat theory (Stephan and Stephan 2000) becomes important. Roughly speaking, realistic threat refers to the material situation and results from competition for scarce resources, while symbolic threat concerns cultural values, shared norms and beliefs, and agreement on the ‘rules of the game’ in a society (Campbell 1965; Stephan, Ybarra, and Morrison 2009; Stephan and Stephan 2017). From a more general theoretical perspective, one could say that realistic threat is about relative access to resources, while symbolic threat is about social production functions that determine the value of resources (Lindenberg 1989; Kalter and Granato 2002; Kalter 2022). This also relates to the juxtaposition of the ‘economic insecurity perspective’ and the ‘cultural backlash thesis’ (Inglehart and Norris 2017; Chung and Mau 2014; Lengfeld and Ordemann 2017) to explain the recent rise of the populist right.

While both realistic and symbolic threat usually contribute to explaining patterns of anti-immigrant/minorities attitudes, symbolic threat is often seen as being especially important in wealthy Western European countries (Davidov et al. 2020; Uenal 2016), where the competition is not mainly about shelter, food or personal security but more about personal freedom, expression and the quality of democracy (Inglehart and Norris 2017; Mudde 2021). Here competition between norms and lifestyles (Hochschild 2017; Kaschuba 2016) leads to a struggle for cultural sovereignty and a polarisation of society along the lines of pro- and anti-plurality (Foroutan 2019).

There is reason to assume that realistic threat might indeed be reduced by the rising number of job vacancies due to demographic change. The fact that upward-mobility of formerly underprivileged groups improves their material situation does not necessarily come at the cost of the already privileged mainstream members, at least not in absolute terms. And, as the pandemic has shown, a number of highly ‘system-relevant’ occupations are already over-proportionally filled by people with immigrant backgrounds (Khalil, Lietz, and Mayer 2020). Many high-ranking positions, e.g. physicians, are obviously system-relevant too. So, there is much room for win-win rather than non-zero-sum. However, the story is likely to be different for symbolic threat. Even if one’s material situation is not directly threatened by minority groups climbing up the social ladder, the fact that they might enter occupations that are involved in shaping and maintaining the values and rules of a society might feel threatening, as these values and rules have ensured one’s position and privileges in society so far.

The German context and the specific role of anti-Muslim sentiment

The threat posed by different minority groups may be perceived differently, depending on historical conditions or the context of their immigration, their demographic and socioeconomic selectivity, or perceived cultural distance. Minority group boundaries can be based on various characteristics, such as country of origin, religion, or skin colour, and are inherently blurred and fluid (Brubaker 2002; Wimmer 2008).

In Germany, people with a Turkish background are the largest group that is often distinguished. In the late 1960s and early 1970s, Turkey became the main country of origin for the recruitment of guest workers in Germany, especially for low-skilled labour. Due to the dynamics of family reunions, transnational marriages, and other events, Turkey was still the most common country of origin (more than 2.7 million) among people with a migration background as defined by the Federal Statistical Office in 2021 (BMI/BAMF 2022, 172). Turkey is by far the leading country of origin among those who belong to the second generation in Germany.

Available estimates based on survey data suggest that about 90% of people with a Turkish background in Germany are Muslims (including Alevis) (Kalter and Heath 2018, 74; Schührer 2018, 25).¹ The importance of this religion-based categorisation has increased enormously, as in many other countries, at the latest since September 11 (Mudde 2021). After Turkey, Kazakhstan (just under 1,3 million) and Syria (just under 1,1 million) account for the largest Islamic-shaped countries of origin in Germany (BMI/BAMF 2022, 172), with most Syrians having arrived as war refugees only since 2014. Other Islamic-shaped countries of origin with notable shares are, for example, Iraq, Afghanistan, Iran and Morocco. According to projections, between 5.3 and 5.6 million Muslims lived in Germany in 2019 which accounts for 6.4%–6.7% of the population (Pfündel, Stichs, and Tanis 2021).

In recent decades discussions about the claims of cultural, ethnic, religious, and national minorities have focused heavily on Islam and Muslims. Not only in Germany, but also in other European countries, Muslim immigration has increased (Pew Research Center 2017), coupled with high levels of anti-Muslim hostility (Bell, Valenta, and Strabac 2021; Gorodzeisky and Semyonov 2019). While Muslims invoke fundamental rights in almost all European constitutions that grant them religious freedom and the right to visibly practice their rites, they shape everyday life in public schools, cities, and workplaces by making concrete demands, such as eating halal food, building mosques, or going to work wearing headscarves or other visible signs of religiosity. In largely secular Western Europe, they challenge the existing habits and norms of society by demanding not only representation but also the right to reshape the society in which they live. And they are better able to do so the higher they are on the social ladder. In doing so, they contest the privileges of the established majority society that sets the ‘rules of the game’, which has led to constant debates about whether Islam and Muslims ‘belong’ to Germany, France, the Netherlands, etc. (Korteweg and Yurdakul 2014; Olgun 2015; Thomas 2011). In parallel, however, this has also led to Muslims fighting for their rights not only in the (pop)cultural and public spheres, but also in the political sphere, resulting in Muslims being the largest minority group in German parliaments at the federal and local levels (Hughes 2016; Schönwälder, Sinanoğlu, and Volkert 2013; Wüst and Bergmann 2023). As a result, people who represent ‘(…) sectors that were once culturally dominant in Western Europe may react angrily to the erosion of their privilege and status’ (Inglehart and Norris 2016, 3).

The data we will describe and use in the following will allow us to compare the threat felt from the social mobility of Muslims in comparison to other groups that are relevant in the German context. In addition to Islamophobia, the German government’s National Action Plan against Racism cites anti-Semitism, antiziganism, and anti-Black racism as particularly virulent forms of racism in Germany (Bundesregierung 2017). In the pandemic, anti-Asian racism also gained prominence and attention (Suda, Mayer, and Nguyen 2020), challenging the perception of Asians as a ‘model minority’. Due to the Nazi past, anti-Slavic racism in Germany deserves special attention, and immigrants from Eastern Europe dominated German immigration in the decades following the recruitment of guest workers. After Turkey, Poland (nearly 2.2 million) and Russia (about 1.3 million) are the second and third countries of origin for people with a migration background in 2019 (BMI/BAMF 2022, 172) and Romania is sixth (about 1 million).

Research questions and hypotheses

In this paper we study the acceptance of structural integration by analyzing outgroup mobility threat (OMT) with respect to different minority groups when these enter specific job positions in the German labour market. We focus on three different set of questions:

1. First, we are interested in group differences, i.e. whether OMT is felt more strongly when it relates to specific minority groups. More precisely, we are interested in how far the evidence for OMT is stronger when it comes to Muslims. As the discussion in the last section has shown, there are theoretical reasons and empirical hints that lead one to expect that this is the case in the context of the German discourse. We formulate this as the first hypothesis:

   ▪

   H1: OMT is perceived more strongly in relation to Muslims than in relation to other minority groups in Germany.

2. Second, we are interested in differences between occupations. As outlined in the theoretical discussion above, not all kinds of job mobilities might pose the same amount of threat nor are they contested in the same way. The different theoretical perspectives lead to three different basic hypotheses:

   	⚬	Mere-competition hypothesis: Realistic threat theory and the general ‘economic insecurity perspective’ would lead to the conclusion, that minorities are welcome in the labour market as long as they do not challenge one’s own status. Accordingly, one would expect: ▪ H2: OMT is stronger for occupations with higher socio-economic status.
   	⚬	Non-zero-sum hypothesis: The structural assimilation perspective of Richard Alba suggests that the acceptance of minorities is not so much connected to status per se, but that it should be higher in occupations where a shortage is felt due to the general demographic changes. In other words: ▪ H3: OMT decreases with the perceived shortage in occupations.
   	⚬	Value-shaping hypothesis: Symbolic threat theories, the emphasis on values in the concept of social production functions and the ‘cultural backlash’ perspective, emphasise that social mobility is not only about the overall economic welfare but might also be considered as a threat because it challenges the ‘rules of the game’ and thus the value of one’s own resources and privileges. Accordingly, one would expect that: ▪ H4: OMT increases in those professions that play a role in shaping and upholding society's values and institutional rules.

3. Third, we are interested in whether there are interactions between minority groups and occupations; in particular, we ask whether H4 holds especially strongly for Muslims.

   ▪

   H5: OMT is highest when Muslims enter professions that play a role in shaping and upholding society's values and institutional rules.

Data and methods

The data for our analyses come from the kick-off study (DeZIM 2022) of the National Discrimination and Racism Monitoring (NaDiRa) in Germany which is run by the German Center for Integration and Migration Research (DeZIM). It is based on a representative CATI survey in Germany, which was conducted between April and August 2021. The fieldwork of the study was carried out by the BIK Aschpurwis + Behrens GmbH. The target population of the survey is the German-speaking population living in private households aged 14 and over. The sample was drawn according to the usual ‘dual frame’ approach, which combines landline and mobile numbers (Häder and Häder 2014). The data set contains 5,003 interviews (for more details see: BIK 2021).

To capture group- and occupation-specific outgroup mobility threat (OMT), a factorial design module was implemented. The respondents were asked to evaluate vignettes of the following type combining three factors:

Different groups are more or less well represented in different areas of society. How good or bad would you find the following developments? Would it be very good, fairly good, neither good nor bad, fairly bad or very bad if more #factor1(minority groups) #factor2(gender specification) worked in/as #factor3(occupations).

Factor1 varies seven different groups, which are especially important in the German context: ‘Asian’, ‘Jewish’, ‘Muslim’, ‘Roma’, ‘Eastern European’, ‘Black’, ‘Turkish’. Note that in addition, two further categories were included in the design of the survey: One is ‘German without migration background’ which is not used in the analyses here.² The other is an empty category with no group specification at all; the corresponding vignettes are used to measure the general demand/need of occupations (see below), but not as cases in the main analyses. Factor2 consists of three different gender specifications: ‘women’, ‘men’ and just ‘people’. Factor3 varies 12 different occupations which are summarisd in Table 1.

Table 1. Occupational categories chosen for the vignettes.

	Private production and services	Medical sector	Public sector	Political sector
High Status	Executive boards (70) Engineers (73)	Physicians (85)	Judges (85) Professors (77)	Higher political offices (77)
Medium/Low Status	Manual trades (34) Office clerks (45)	Nurses (25)	Teachers (68) Police officers (56)	Local politicians (51)

In brackets: ISEI score.

We covered four different areas of the labour market, namely private production and services, medical sector, public sector, and political sector. In each we chose both higher status and also lower or medium status occupations which results in the eight cells of Table 1. In four of these eight cells we considered two different occupations, so that we included 12 occupations in total.

All respondents received eight vignettes, each of them covering a different cell of Table 1 for factor3. In the cells containing two occupations, one of these was selected randomly. Minority groups (factor1) and gender specification (factor2) were randomly selected. For the analyses we only use the vignettes in which one of the seven minorities is inserted. These are 30,276 vignettes in total.

Note that the sample representing the (German-speaking) population in Germany naturally contains a number of respondents who, depending on the operationalisation, fall into one or more of the specified minority group categories. For instance, 100 respondents self-identify as ‘Asian’, 42 as ‘Jewish’, 243 as ‘Muslim’, 24 as ‘Sinti or Roma’, 250 as ‘Eastern European’, 63 as ‘Black’, and 127 were born in Turkey or have at least one Turkish-born parent. Overlaps, i.e. multiple group memberships, are possible. In total there are 736 vignettes in the analyses where the respondent is a member of the group specified in the vignette formulation. This is only a very small proportion (2.4%) of all vignettes. We decided to leave them in the analyses; we checked and found that dropping them would not lead to any substantive change of the results in this paper.³

We measure OMT by the share of those who evaluate a vignette either by ‘bad’ or ‘very bad’. In general, the evaluation answers to the vignettes show an uneven distribution with a clear break between the first three answer categories (very good, good, neither good nor bad) and the remaining two (bad, very bad).

To analyze the effect of vignette factors we run linear probability models estimating robust standard errors accounting for the clustering along respondents (see Appendix Table A1). To ease the interpretation of the coefficients, we multiply the dependent OMT variable by 100; thus the coefficients of the independent variables can be read as a change in percentage.

To measure the status of an inserted occupation we use the International Socio-economic Index of Occupational Status (ISEI) (Ganzeboom, De Graaf, and Treiman 1992). Table 1 contains the corresponding values of the occupations in brackets.⁴ To measure the societally perceived shortage in occupations, we make use of the fact that the data also contains vignettes with no minority group specification at all, but just insert an empty category for factor2. Accordingly, the vignettes reduce to, for example: ‘Would it be very good, fairly good, neither good nor bad, fairly bad or very bad if more people (woman, men) would work as nurses’. We use the percentage of ‘very good’ answers as a measure for the subjectively perceived shortage in an occupation. The respective percentages for the 12 occupations range from 14.9% (office clerks) to 41.9% (nurses). We use a dummy variable to indicate whether an occupation is one of the six in Table 1 belonging to either the public or the political sector, as these are especially involved in the shaping or maintenance of societal norms, values and rules.

Results

Group differences

Figure 1 analyzes the differences between the minority groups over all gender specifications and occupations. The first thing to notice is that, overall, the share of aversive answers is relatively low. Summing up all minority groups included, 8 percent of all vignettes are evaluated with either ‘bad’ or ‘very bad’.⁵ However, there are clear differences between the minority groups.

While there is only little aversion to Black (4.6%) or Jewish (5.0%) people in the presented occupations, there is a significantly higher disapproval of the other five groups. The highest OMT shares are found towards Muslims (11.1%), followed by Sinti and Roma (10.4%), Turkish (9.4%), East European (8.8%) and Asian (8.2%).⁶

Figure 1 confirms hypothesis H1 which expects the OMT with respect to Muslims to be especially strong. As the confidence intervals in the figure suggest, and as Model 1 (Appendix Table A1) confirms, the difference to almost all other groups is highly significant, Sinti and Roma being the only exception.⁷

Figure 1. OMT differences between groups.

Data: NaDiRa kick-off study.

Differences between occupations

Turning to the differences between the occupational categories, Figure 2 shows the OMT shares summing up all minority groups combined. The occupations are ordered by their ISEI score which is given in brackets.

Figure 2. OMT differences between occupations.

Data: NaDiRa kick-off study.

We find that there are huge differences in OMT between the different occupations, with the largest resistance against the minority groups assuming the functions of local politicians or judges. At the other extreme, only a tiny share of respondents dislikes them in manual trades or as engineers.

Testing the mere-competition hypotheses

According to hypothesis H2 there should be higher OMT in occupations with higher ISEI score. But this does not really seem to hold true when checking Figure 2 by eye. Most strikingly, OMT is high for ‘local politicians’ and low for ‘engineers’, ‘professors’, and ‘physicians’. There is also no clear pattern when comparing low or medium status occupations to high status occupations in the separate areas of the labour market. ‘Engineers’ receives even somewhat lower OMT than ‘manual trades’ or ‘office clerks’, ‘physicians’ even less than ‘nurses’, ‘professors’ less than ‘teachers’, and ‘high political offices’ less than ‘local politicians’.

Still, in a regression using all vignettes, controlling also for the group and gender specification of the vignette, the ISEI score of the occupation given in the vignette has a positive effect (see Table A1, Model 3 in the Appendix). This means that there is a tendency for OMT to be somewhat larger for higher status occupations. The effect is significant on a high level (p < .001); however, the number of observations (vignettes included in the analyses) is very large, and the actual effect size is rather weak: the probability of OMT rises by 0.03% with one unit of the ISEI score, which for the full range of ISEI scores included (25–85) can only account for a difference of 1.8%. Including the ISEI score also does not increase the explanatory power of Model 1 without it. In comparison, including the occupations as dummies in Model 2 increases the R²-value by more than a factor of 3.

In sum, there is only weak support for hypothesis H2. We conclude that a mere realistic competition view is far from sufficient to understand the patterns of resistance against minority groups being more integrated in different occupations.

Testing the non-zero-sum hypothesis

To test hypothesis H3, which is derived according to the non-zero-sum argument by Richard Alba, we add the perceived shortage to the regression (Model 4 in Appendix Table A1). The explanatory power of the model rises. The variable has a negative effect of −31.2, which is significant on a high level. This means that an increase of one unit in perceived shortage would decrease OMT by 31.2%. Note that perceived shortage is measured as a percentage, so a unit reflects 100%. Still, the range of the variable (from 14.9% to 41.9%) covers 27% which yields a difference of 8.4% in the predicted OMT value.

Figure 3 illustrates the OMT differences between occupations again, but now ranked according to the perceived shortage. As compared to Figure 2, a kind of trend is more clearly detectable. Figure 3 also shows the values predicted from the regression including the perceived shortage (Model 4 in Appendix Table A1). The graph confirms that the model can capture some of the differences between the occupations. It also shows, however, that the actual OMT patterns partly still deviate greatly from the prediction.

Figure 3. OMT differences between occupations ranked by perceived shortage.

Data: NaDiRa kick-off study.

We conclude that the perceived shortage in occupations explains the differences in OMT for minority groups much better than the status, and that thus the demographic non-zero-sum argument expressed in H3 is generally supported. However, it still does not seem to grasp some major OMT differences between the occupations sufficiently.

Testing the value-shaping hypothesis

As a next step in our regression analyses of the vignettes, we add a dummy variable indicating whether an occupation is one of the six in the public or political sector (see Table 1), and thus involved in the shaping and consolidating of societal values and rules. We find that this adds significantly to the explanation (see Model 5 in Appendix Table A1). The coefficient tells us that, controlling for the other variables, OMT is on average 5.7% higher for value-related occupations as compared to the others. Including this variable also leads to a substantive increase in the R²-value.

Thus, hypothesis H4 is basically supported, and it seems that the value-shaping argument adds to the non-zero-sum argument. The inclusion of the value dummy reduces the effect of the perceived shortage variable, but the latter still contributes significantly to Model 5.

Interactions between groups and occupations

In the last step of our analyses, we address hypothesis H5 and investigate to what extent there is a specific OMT posed by Muslims when it comes to value-shaping occupations. To assess this, we add an interaction of the dummy variable for the public and political sector occupations with the dummy for the Muslim category and also with the dummy for the Turkish group to the regression (Model 6).⁸ We find that next to the positive main effect of the public and political sector, both interactions have a highly significant positive effect, the one for Muslims being stronger than the one for Turkish.⁹ This means that there is a specific OMT when respondents think of Turkish people, and even more so if they think of Muslims in these kinds of occupations.

Figure 4 shows this in more detail. It compares the vignette answers for the different occupations between respondents asked about Muslims, Turkish people and the other groups. In addition, the figure contains the predicted OMT values from Model 4 (Appendix Table A1), i.e. based on a regression on the occupations’ ISEI-scores and the perceived shortages.

Figure 4. OMT relating to Muslims and Turkish people compared to OMT relating to the other groups and to model predictions.

Data: NaDiRa kick-off study.

Interestingly, perceived OMT relating to Muslims and Turks does not differ significantly from that perceived with respect to the other groups in five of the six non-value-shaping occupations. There is a notable difference between Muslim and the others in the occupation ‘engineers’, but OMT is on a comparably low level here and the estimate for Muslim is still in the range of the value predicted by the ISEI score and the perceived shortage. The estimate for the category Turkish is in line with the estimate for the category ‘other’ even in all six non-value-shaping occupations.

Things are different in five of the six value-shaping occupations, professors being the only exception. There is a clearly higher OMT for Muslims in the occupation ‘teachers’, while the estimates for Turkish and other are in the range of the predicted value. For judges, the estimates for both Muslim and Turkish not only by far exceed the predicted values, but also the estimate for the group ‘others’. When it comes to police officers, there is no OMT higher than predicted for the ‘others’, but the category Turkish differs significantly, and the category Muslim is also (though not significantly) on a much higher level. For the occupations of high political offices and local politicians, we observe a slightly higher OMT than predicted for the others, a clearly higher value for Turkish, and a much higher value for Muslims.

There is, thus, a specific threat attached to Muslims, and to a large degree also to the overlapping group of people with a Turkish background, when it comes to entering positions in the public and political sector, positions that are related to the shaping and maintenance of societal values. For the other groups, the estimated threat differs only slightly from the values predicted by a mere competition (ISEI score) and non-zero-sum (perceived shortage) argument.¹⁰

Conclusions and discussion

Western societies have become increasingly diverse, and demographic processes will reinforce this in the years ahead. Richard Alba has rightly argued that in many countries the age structure of the workforce will offer the descendants of formerly socioeconomically disadvantaged immigrants unique opportunities for social mobility and thus for effective structural integration. As research on ethnic inequality has shown, this view is strongly supported by the increasing success of immigrant children in education systems and labour markets in many countries. But acceptance and recognition are also necessary components of sustainable societal integration, and it is less clear how the members of the previous mainstream or dominant society will perceive and deal with these processes.

Richard Alba’s view is also quite optimistic in this regard. He has emphasised that the game is basically a non-zero-sum game and that, given the demographic structure, the intergenerational gains of the immigrant groups are not necessarily at the expense of the former mainstream. Thus, there should be no serious reason for a threat perception as would be expected under traditional theories of ethnic competition. Our analysis in the German context provides some support for this argument: in our vignette study, we presented twelve different occupations and demonstrated that dislike of immigrant minorities varied across these occupations. In an attempt to explain the patterns of aversion and non-aversion by key characteristics, the perceived shortage in the occupation, which is related to the non-zero-sum argument, performs much better than the status of the occupation, which would indicate competition.

Nevertheless, attitudes of rejection against immigrant groups, especially against Muslims, persist controlling for the demand of different jobs. And in the vignette analyses perceived shortage does a much poorer job in explaining the aversion regarding occupations that are in the public and political sector and are thus involved in the shaping and maintenance of societal norms and values. And this holds, again, especially for Muslims. We argue that a refined version of our concept of outgroup mobility threat (OMT) can contribute to understanding this. OMT captures the specific fear of being overtaken by intergenerational integration successes. In this general formulation, this seems challenged by and in conflict with the non-zero-sum argument. However, it is helpful to explicitly make the meanwhile common distinction between realistic and symbolic threat. More precisely, we have argued that the non-zero-sum argument is plausible for realistic threat, but that the logic is likely to be different for symbolic kinds of threat. If immigrant minorities enter occupations that shape societal norms and rules, and thus the value of the resources that the so-far mainstream members possess, it is not a non-zero-sum game anymore. OMT and the non-zero-sum argument are thus necessary corrections to each other, and the seemingly conflicting perspectives can be integrated via this differentiation.

But given this differentiation, the long-term consequences for societal integration are less straight-forward. New assimilation theory is based on the idea that in immigration countries more and more people from different backgrounds form a continuously changing mainstream, and ethno-racial distinctions will lose their influence on crucial aspects of life over time (Alba and Nee 1997, 863; Alba and Foner 2015), driven by strong general mechanisms, supported by structural forces such as the demographic ones. Many assimilation and integration theories speak out to these blurring boundaries (Wimmer 2008; Lamont and Molnár 2002) and without doubt there is much empirical evidence for this in the U.S. and many other immigration contexts (Alba 2005; Drouhot and Nee 2019).

However, post-colonial theory and critical race theory point to the continuity of othering based on perceptions of racial inferiority and white supremacy even though racialized people ascend into the mainstream (Embrick and Moore 2020; Paré 2022). Theories of ambivalence (Bauman 2016), hybridisation (Bhabha 2000; Spivak 1990), and orientalism (Said 1978) are less optimistic about long-term recognition of minorities suggesting that it could be precisely the blurring of boundaries that leads to irritation and anxiety on the part of the white majority, because the blurring boundaries dissolve the idea of whiteness as purity and the Western world as more civilised. Questioning established positions and privileges (Craig and Richeson 2014; Craig, Rucker, and Richeson 2018; Davis and Wilson 2021) and addressing structural racism (Bailey et al. 2017; Bonilla-Silva 2021) is rather leading to polarised debates about culture wars and cancel culture (Norris 2023) than to a new relaxed hybrid mainstream. Philosophical approaches explain the reluctance toward blurring boundaries and an ever more diverse mainstream through the comfort we find in the binary logic in which we as humans are trained to position ourselves towards ‘the other’ (Buber 1995).

The possible explanation for the persistence of non-recognition that we have analyzed in this paper is based on a different and clearer mechanism, but touches on some of these elements. In contemporary Germany, as in most other European countries, ‘Muslim/Islam’ is the brightest boundary, which is confirmed by our results. As a ‘religious’ marker, the external attribution contains not only generalisations and stereotypes about socioeconomic background, relevant skills and resources, or cultural habits, but also about basic norms and values that are perceived as different from those of the largely secular mainstream. Many people feel that these norms and values are at odds with the general rules of how society works or should work. This leads to a particular danger of an ‘either/or’ struggle over the constitution of society, rather than gradual learning and adaptation. And at the same time, it is precisely this struggle that makes the boundary brighter.

It is an open empirical question whether OMT posed by Muslims in value-shaping occupations will nevertheless decline, albeit perhaps with a slower gradient, if the perceived shortages in theses occupations continue to increase. Figure 4 above gives the impression that this is not the case, but the design and data do not allow us to do analyse this in more detail; one would need a much larger number of occupations and greater variance in their shortages to say this with greater certainty.

Richard Alba has already pointed to the fact that ethno-racial differences in access to positions and resources will not disappear completely, but he takes a fundamentally optimistic stance, betting on learning and adapting to diversity and counting on strong demographic structural forces. Although much points in this direction, the continued opposition of a significant portion of the mainstream to immigration and certain ethno-racial groups remains puzzling, and the rise of anti-immigrant and anti-Muslim political parties poses a challenge to the idea of directional progress. Consequently, Western immigration societies also face increasing cultural, ethnic, religious, and racial self-emphasis that contradicts the notion that ethnic boundaries automatically fade out if structural integration succeeds. In Germany and elsewhere in Europe, pronounced identity politics is also a signature of the current time, particularly because, despite all the intergenerational progress, the promises of equity and equality – including recognition – have yet to be convincingly realised (Foroutan 2019).

Acknowledgements

We thank Steffen Shah for his extremely dedicated help and work in the development of the vignette items and in the setup of the survey. We thank Nina Bühler and Caroline Simpson for their excellent support in preparing the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data of this study will be made available for scientific use via the DeZIM Research Data Center (https://fdz.dezim-institut.de/en). Estimated date of publication is the end of 2023.

Notes

1 Precise data from the official statistics is not available as the information on religious affiliation is not included in the German Microcensus.

2 The category was included in the survey for other purposes, but it makes no sense to include it in the analyses here. The formulation of the vignette referring to Germans without migration background triggers an anti-diversity framing of the question and a ‘bad’ or ‘very bad’ answer probably represents a pro-diversity attitude rather than a feared threat by this group.

3 We refer to this again in footnote 6.

4 The occupations were coded according to the classification scheme for the German Microcensus suggested by GESIS (https://www.gesis.org/missy/files/documents/MZ/isco_isei.pdf).

5 Weighting the vignettes with the redressment weight that is contained in the data set only marginally increases this share from (8.0 to 8.3). As in this paper we are interested in differences between minority groups and occupations and not in characteristics of the respondents, we do not use the redressment weight throughout the analyses.

6 Here, it is important to emphasize once again that the results would remain unchanged even if we excluded those 736 cases where the respondent belongs to the minority group which is specified in the formulation (see 'Data and methods'). More precisely, the percentages would not deviate by more that 0.1% in each case. Therefore, the racial hierarchy depicted in Figure 1 is not influenced by the actual number of individuals from each group in the sample.

7 For the group ‘Turkish’ this holds on a 1%-level, for all other groups even on a 0.1%-level.

8 Most, but not all people with a Turkish background in Germany are Muslim and among Muslims those with a Turkish background constitute the largest subgroup, so there is a large but not perfect overlap between these two categories.

9 We also tested whether the two dummies (Muslim or Turkish) interact with the perceived shortage or with the ISEI of the occupations, but this is not the case (on a level of at least p < .05).

10 This holds true even more if one takes out the category ‘Sinti and Roma’, a group that has also been related to symbolic threat mechanisms in Germany (Jedinger and Eisentraut 2020), from the group ‘other’. Then, only in the occupation local politicians would there be a small significantly higher OMT as predicted by ISEI score and perceived shortage.

Appendix

Table A1. Linear probability models of OMT on vignette factors and occupational characteristics.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Factors
Minority groups
Ref. Muslim
Asian	−2.98 ***	−3.17 ***	−2.98 ***	−3.01 ***	−3.12 ***	−0.87
	(0.64)	(0.63)	(0.64)	(0.64)	(0.64)	(0.67)
Jewish	−6.08 ***	−6.16 ***	−6.09 ***	−6.08 ***	−6.10 ***	−3.87 ***
	(0.60)	(0.60)	(0.61)	(0.60)	(0.60)	(0.63)
Sinti & Roma	−0.76	−0.90	−0.78	−0.78	−0.88	1.36
	(0.69)	(0.68)	(0.69)	(0.69)	(0.69)	(0.71)
East European	−2.36 ***	−2.33 ***	−2.37 ***	−2.33 ***	−2.35 ***	−0.13
	(0.67)	(0.67)	(0.67)	(0.67)	(0.67)	(0.69)
Blacks	−6.53 ***	−6.69 ***	−6.54 ***	−6.60 ***	−6.67 ***	−4.43 ***
	(0.60)	(0.60)	(0.60)	(0.60)	(0.60)	(0.63)
Turkish	−1.72 **	−1.83 **	−1.71 **	−1.76 **	−1.87 **	−0.93 **
	(0.65)	(0.64)	(0.65)	(0.65)	(0.64)	(0.72)
Gender specification
Ref. no spec.
Men	1.59 ***	1.47 ***	1.48 ***	1.47 ***	1.54 ***	1.52 ***
	(0.40)	(0.41)	(0.41)	(0.41)	(0.40)	(0.40)
Women	−1.75 ***	−1.79 ***	−1.79 ***	−1.77 ***	−1.77 ***	−1.79 ***
	(0.36)	(0.37)	(0.37)	(0.37)	(0.36)	(0.36)
Occupations
Ref. Nurses
Manual trades		−1.38 **
		(0.54)
Office clerks		0.27
		(0.59)
Local politicians		9.85 ***
		(0.63)
Police officers		3.86 ***
		(0.71)
Teachers		3.63 ***
		(0.71)
Executive boards		3.94 ***
		(0.72)
Engineers		−1.86 ***
		(0.52)
High political offices		6.90 ***
		(0.58)
Professors		−0.39
		(0.57)
Physicians		−1.15 **
		(0.43)
Judges		8.85 ***
		(0.83)
ISEI of occupation			0.03 ***	−0.01 *	−0.04 ***	−0.04 ***
			(0.01)	(0.01)	(0.01)	(0.01)
Perceived shortage				−31.21 ***	−17.24 ***	−17.28 ***
				(2.13)	(2.17)	(2.17)
Value-related					5.67 ***	4.64 ***
					(0.32)	(0.36)
Value-rel ## Muslim						4.56 ***
						(1.00)
Value-rel ## Turkish						2.55 ***
						(0.92)
Constant	11.24 ***	8.29 ***	9.43 ***	20.78 ***	15.52 ***	13.82 ***
	(0.59)	(0.62)	(0.65)	(1.10)	(1.09)	(1.11)
N	30276	30276	30276	30276	30276	30276
R^2	0.010	0.033	0.010	0.017	0.025	0.026

Data: NaDiRa kick-off study.

Coefficients represent changes in probability (dependent variable OMT multiplied by 100).

Standard errors in brackets (robust, accounting for clustering by respondents).

* p < 0.05; ** p < 0.01; *** p < 0.001.