Who supports whom? Citizens’ support for affirmative action policies in recruitment processes towards four underrepresented groups

ABSTRACT Despite the divisive character of Affirmative Action Policies (AAP)s, we still know very little about why some people oppose, while others support such regulations. We provide new evidence from a survey experiment in Germany, where we asked respondents to what extent they would support the introduction of a hypothetical regulation favouring – if equally qualified – members of an underrepresented group in the recruitment process for a management position. We randomly varied the APP’s target group between women, persons with an immigrant background, native East Germans and persons from a non-academic household. Our study shows that being a member of the AAP-targeted group significantly and substantially increases support for such a regulation. We also find evidence for increased support for AAP if the targeted group is perceived as disadvantaged. By contrast, our results do not back up the idea that prejudice affects AAP support.


Introduction
Identity politics have recently gained salience in the public debates of most Western European countries.Recently, this debate has focused on the extent to which traditional left-wing parties should orient their political agenda towards measures targeting structural inequality faced by particular social groups (Abou-Chadi et al., 2021).Proponents of such a political (re)orientation conceive group-based rights as a necessary and complementary tool to individual-based rights for redressing group-based unequal participation in various societal arenas.By contrast, opponents argue that traditional left-wing parties should strive to fight social inequality primarily within the whole population regardless of group memberships.Accordingly, allocating particular rights to individuals because of common group membership is assumed to lead to a fragmentation of the society into subgroups defined by particular identities which are perceived as constitutive and ultimately weaken social cohesion.
This heated debate around the relevance of identity politics for the traditional left-wing parties tends to discuss various instruments to redress inequality, including gender-neutral restrooms in public buildings, the implementation of gender-sensitive language, or the introduction of quotas in favour of underrepresented groups in leading positions.The latter instrument is an example of an affirmative action policy (AAP): AAP imply that members of underrepresented groups are given preference over others if equally qualified in selection processes for leadership positions and/or public offices, and are thus group-based policies aimed at equality of outcomes (Harrison et al., 2006).As AAP are often perceived as a zerosum game, they are an essential topic of contestation between proponents and opponents of identity politics.
In Europe, the public debate on identity politics discusses women, LGTBIQ+ and persons with an immigrant background as the main underrepresented groups (Abou-Chadi et al., 2021).In the German context, two other underrepresented groups have been receiving attention in this debate.First, persons born in the former GDR (hereafter 'East Germans') are still largely underrepresented in key positions in various societal arenas more than thirty years after reunification (see Mau 2019;Rennefanz 2019).Accordingly, this might help explain the feeling of being left behind and the perception of a lack of representation shared by some of the East Germans.Second, social origin is a particularly important determinant in educational and professional achievement in Germany (Consiglio & Sologon, 2022), leading to an underrepresentation of persons from a non-academic background (i.e., with parents without tertiary education degrees) in leading positions (see Table 3).
The public debate has not only been discussing the relevance of identity politics instruments in the political agenda of traditional left-wing parties but also which underrepresented groups should be entitled to benefit from regulations aiming at improving their representation in influential positions.Against this backdrop, the lack of empirical studies on citizenś attitudes toward AAP for underrepresented groups is striking.On the one hand, various studies have highlighted the relatively high level of tolerance toward underrepresented groups among the German population (e.g., Mau et al., 2020;Teney & Rupieper, 2021).Moreover, public opinion towards redistributive politics (e.g., Reeskens & van der Meer, 2019;Schwander & Vlandas, 2020) and perceived social inequality (e.g., Mijs, 2021) has received much attention from scholars.On the other hand, we know very little about citizenś support for concrete policies targeting underrepresented groups for redressing group-based inequality apart from support for gender quota regulations for supervisory boards, which only concern a very small group of individuals in the highest ranking positions of specific large companies (Möhring et al., 2019;Möhring & Teney, 2020).
Investigating citizens' support for AAP for underrepresented groups is not only of societal relevance but also provides for interesting scientific inquiry.Analysing support for AAP can help us understand the interaction between key attitudinal dimensions underpinning the support and real-life policies.Previous research focusing on equal outcome policies for African Americans in the US suggests that support for AAP is based on the interplay of identification with the target group, one's own level of prejudice against this group, and the perception of unequal opportunities faced by the target group (Krysan, 2000).However, as most previous research has largely focused on the US context and AAP aimed at equal opportunities for African Americans, results cannot be generalised to other contexts and other underrepresented groups.With our study, we can test these previous results in another context and for diverse target groups which are not only defined by ethnicity.
Against this background, we investigate the determinants of support for implementing a hypothetical AAP in recruitment processes for leading positions.Based on preregistered survey experiments (Teney et al., 2022) conducted with a YouGov online panel among the working population in Germany, we can compare the level of support for AAP for women, persons with an immigrant background, persons born in East Germany, and persons from a non-academic background (i.e., with parents without tertiary education degree). 1 More specifically, we designed a survey experiment composed of vignettes introducing a hypothetical regulation favouring one of the four target groups if equally qualified in the recruitment process for a leading position.The target groups were randomly assigned to the vignettes.Respondents were first asked to evaluate four vignettes in total.Then, they were invited to answer several attitudinal items that might explain support for and opposition to AAP measures.For our survey, we oversampled women, persons with an immigrant background, East Germans and persons from a non-academic background (hereafter 'sampling groups').This sampling design enables us to investigate support for AAP among members of AAP target groups and to incorporate three dimensions of analysis.First, we can compare differences in support for AAP depending on the vignetteś target group.Second, we can compare the level of support for AAP of respondents belonging to these target groups.Third, we can investigate the power of various attitudinal factors in explaining these differences in the level of support for AAP.We focus on the factors that have been shown to play a key role in the support for policies aimed at redressing ethnic inequality: group-based interest and group identity, prejudice and belief about discrimination and source of inequality (Dixon et al., 2017;Sniderman et al., 1999;Weeks & Baldez, 2014).We assess the extent to which these factors can explain (1) vignette target group differences in support for AAP and (2) differences in support for AAP between respondents belonging to these target groups.
We find that being a member of the AAP-targeted group significantly and substantially increases support for the introduction of a quota regulation.Additionally, we find group-based differences in the rationales for supporting a specific AAP: While women in particular tend to show solidarity with members of other underrepresented groups, East Germans and persons from a non-academic household respond according to group competition theory.We also find evidence for increased support for AAP if the targeted group is perceived as disadvantaged.By contrast, our results do not back up the idea that prejudice affects AAP support.
This paper is structured as follows: First, we present our hypotheses to explain (1) differences in support for AAP among respondents belonging to different target groups and (2) differences in support for AAP depending on the vignetteś target groups.Then, we describe the design of our survey experiment before presenting our results and concluding.

Group-based support for AAP
How do (perceived) group differences explain support for AAP?A rational choice approach is frequently used to explain policy support by individuals, whereby own interest is assumed to follow an individual's socio-demographic characteristics, i.e., their membership of socially defined groups such as gender, age and origin.Subsequently, individuals evaluate their personal economic opportunities in relation to the economic position of their community.People grow more optimistic about their future when their group's economic position improves, even if their individual economic situation stays the same (Becker, 2021).According to this group-based interest perspective, individuals who perceive the benefit of a policy for members of their own group are more likely to support it.Consequently, members of the target groups tend to favour AAP policies that are believed to help their own demographic group (Harrison et al., 2006).The body of empirical research on support for AAP has consistently provided evidence confirming this group interest-based explanation (e.g., Harrison et al., 2006;Möhring et al., 2019;Scarborough et al., 2019).
Thus, we expect respondents belonging to a target group to support AAP for their own target group to a significantly larger extent than respondents not belonging to the target group (Hypothesis 1).
Second, members of a target group might differ from members of the nontarget group in their support for AAP for other target groups.On the one hand, members of an underrepresented group might be more likely than non-members to identify and feel solidarity with other underrepresented groups.They might indeed perceive similarities in their disadvantaged position with members of other underrepresented groups.This, in turn, would lead members of an underrepresented group to support AAP for other target groups to a greater extent than non-members.We know from US studies that members of underrepresented groups are more likely to support AAP targeted at other underrepresented groups than members of the majority group (i.e., white men) (Bolzendahl & Coffé, 2020;Kane & Whipkey, 2009;Scarborough et al., 2019).However, findings on support for AAP from the US context might not be generalisable to other countries.Indeed, most US studies on support for AAP focus on African Americans as an AAP target group (Harrison et al., 2006;Krysan, 2000).However, the socio-structural situation of Afro-Americans is hardly comparable to the socio-structural case of underrepresented groups in Germany or other European countries.This leads us to formulate our second hypothesis: Respondents belonging to one target group are more likely to support AAP for the other target groups than respondents not belonging to any target group (Hypothesis 2a).
Alternatively, we could also expect respondents of a target group to oppose AAP for other target groups to a larger extent than respondents not belonging to any target groups.Indeed, based on group competition theory (Blumer, 1958;Bobo, 2000), respondents might perceive members of AAP target groups as competitive threats for valued (but scarce) social resources, statuses and privileges.Indeed, a recent survey among East Germans and German Muslims showed that East Germans perceive social upward mobility of German Muslims as a threat against their own status (Foroutan et al., 2019).As AAP tend to be considered a zero-sum conflict, members of a target group might evaluate the implementation of an AAP for another target group as a threat for their own group (see, for a detailed discussion, Różycka-Tran et al., 2015).As the number of leading positions is limited, there is not only a conflict between target and non-target group members but between members of different target groups.Potential conflicts between members of underrepresented groups might even be more intense as they are outsiders competing for scarce vacant job positions that established non-target group members do not yet occupy.Thus, the introduction of AAP for various target groups in recruitment processes for leading positions implies that the absolute benefit from such a regulation to the members of one target group directly depends on the total number of underrepresented groups included in the regulation (and their demographic size) concerning the number of vacant job positions.This leads us to formulate a competing hypothesis to Hypothesis 2a: Respondents belonging to one target group are less likely to support AAP for the other target groups than respondents not belonging to any target groups (Hypothesis 2b).
How do perceived disadvantage and prejudice explain support for AAP?
Support for AAP depends on the perception that members of a target group are unfairly disadvantaged in the respective societal arena.Following studies on perceptions of group inequality and support for redistribution, we can discuss two primary motivations leading to redistributive preferences: The literature distinguishes between normative and economic motives for redistribution.Income inequality is positively correlated with the demand for redistribution, ascribed to norms and values-based inequality aversion.Becker (2020), for example, demonstrates that when informed about objective discrepancies between individuals with different characteristics ascribed at birth (e.g., gender, race or family background), US citizens modify their redistribution preferences, which may serve as markers of economic justice.In the same vein, the literature submits that the perception that group members are unfairly treated or structurally disadvantaged is indeed a strong determinant of support for AAP for ethnic minorities (Harrison et al., 2006;Krysan, 2000) and diversity policies in the labour market for women and ethnic minorities (Möhring & Teney, 2021;Scarborough et al., 2019).
Yet, people are not opposed to a proportion of inequality caused reasonably by inequalities in individual merits (Ahrens, 2022).Addressing Ahrens's (2022) argument, respondents are less likely to support AAP if they consider members of a target group not to be disadvantaged or to be responsible for their disadvantaged situation.That is why, in addition to the perception of disadvantage faced by members of the target groups, we will also assess the extent to which respondents evaluate members of target groups to be as committed to their job as non-members of target groups.The perceived job commitment (operationalised as perceived competence and performance) of members of target groups, in turn, might also explain the different levels of support for AAP depending on the target groups.Indeed, respondents might perceive members of a target group as disadvantaged in the labour market but consider them responsible for their situation (i.e., because of a perceived lack of commitment).Such respondents are likely to oppose AAP for the corresponding target group, as they would consider members of the target group to control their labour market situation.Such an argument has been made in the debate on the introduction of a quota for women on the boards of DAX-listed companies (e.g., Terjesen & Sealy, 2016): qualified women are perceived to be, on average, less willing to hold such leading positions due to lower career aspirations.This leads us to derive our third hypothesis: Differences in support for AAP according to vignette target groups are partly explained by different levels of perceived disadvantage and perceived commitment of the target groups in the labour market (Hypothesis 3).
Ethnic and racial prejudices are significantly related to lower levels of support for relatively informal solidarity towards immigrants (van Oorschot, 2006) and welfare programmes targeted at ethnic minorities (Harell et al., 2016).This argument might be extended to other underrepresented groups.An assumed lack of competency to execute leading positions ascribed to specific groups might drive disapproval of AAP targeted at these groups.Möhring and Teney (2021) point to the relevance of prejudice against women and against persons with an immigrant background in explaining target group differences in support for AAP with prejudice being negatively related to support for AAP.Thus, the fact that prejudice varies across the AAP target groups might explain differences in support for AAP for different target groups.We operationalise prejudice against the four target groups by evaluating members of the target groups as not possessing the necessary competencies and qualities for holding a management position.More precisely, we asked respondents to assess the competence of a hypothetical supervisor if they were a member of one of our four target groups compared to a supervisor not belonging to any target groups.Accordingly, we formulate our last hypothesis as follow: Differences in support for AAP according to vignette target groups are partly explained by different levels of prejudice against the target groups (Hypothesis 4).

Sample
We used original data that we collected with a YouGov online panel in July 2021.The sample consists of 2,676 individuals registered in the labour force in Germany at the time of the survey.As we focused on attitudes towards AAP of respondents belonging to one of four groups that would benefit from an AAP, we oversampled respondents of those groups, in the following labelled as sampling groups (SG).In analogy with our AAP target groups, we defined four sampling groups: women, persons with a migration background, native East Germans (i.e., born in one of the former GDR states), and respondents from a non-academic background.We defined respondents as having a migration background if they are foreign-born, have at least one foreign-born parent, or have non-German citizenship.We classified respondents as having a non-academic background if neither of their parents graduated from university or technical college.

Experimental design
To assess support for AAP towards our four target groups, we developed vignettes on a hypothetical AAP regulation (Teney et al., 2022).We varied four dimensions of the vignettes: In the first dimension, we randomly assigned the target group benefitting from the hypothetical AAP (individuals with a migration background, women, native East Germans or persons from a non-academic background).Second, we varied the level of the labour market segregation faced by the target group (highly underrepresented or underrepresented).Third, the vignettes differed in the sector of activity in which the AAP will apply (private sector or public service).Lastly, we randomised the management level of the position targeted by the regulation (senior management, middle management, or senior positions).The vignettes read as follows: [Members of target group] are [segregation level] in executive positions [sector of activity] in Germany.Therefore, an independent commission proposes the following regulation: Employers should be legally obliged to give preference to [members of target group] in application procedures for [position's management level] if equally qualified.
Our outcome variable refers to the degree to which respondents support or oppose such a regulation.We measured support for an AAP on an 11-point Likert scale ranging from 0 (fully oppose) to 10 (fully support).We assigned each respondent randomly to four versions of the vignette.In total, we assessed the support for 10,704 vignettes (each of the 2,676 respondents evaluated four vignettes).We used a full factorial design.Given our four vignette dimensions (target group, segregation level, sector of activity, and position's management level) with their different levels (see description above and section S1 of the Supporting Information), the size of the vignette universe is 4 × 2×2 × 3 = 48.The factorial design is 'full' as we considered the entire universe of the 48 possible vignettes when randomly drawing four vignettes per respondent.

Operationalisation
To test our hypotheses, we used three further questions as moderator variables.First, we asked to what extent respondents thought that people from our four target groups were disadvantaged in applying for management positions in Germany.The respondent could indicate their opinion on a fivepoint scale ranging from 'not at all' to 'very strongly' for each group.Second, we asked the respondents to rate the leadership competence and presented an antagonistic pair of individuals.We presented four antagonistic pairs (woman vs man; a person with vs without migration background; a native East vs West German; a person from a non-academic vs academic household) and asked on an 11-point Likert scale whether they thought that the first person (a member of a target group) was more competent (value of −5) or the member of the majority (value of +5).A value of 0 corresponded to the opinion that neither the members of the target group nor the members of the majority are considered more competent.The values in-between could be used to nuance the rating.Third, in analogy, we asked how respondents rated the performance of the four pairs, asking whether they thought that the first person (a member of a target group) performed better in the same position (value of −5) or the member of the majority (value of +5).Perceived competence and level of performance (or lack thereof) jointly measure prejudice.
We furthermore controlled for respondents' age, educational level, employment sector, whether they fulfil a managerial role, whether they are married/live in a civil partnership, and whether they have children or not.We included these controls mainly to capture the non-random assignment of the moderators in our analysis.We provide summary statistics in the Supporting Information Tables S1 and S2 for the whole sample and by survey group, respectively.

Analytic strategy
The units of analysis are the vignettes (with respondents' evaluation of the vignette as the outcome).We employed ordinary linear least squares regression to regress the rating outcomes on sets of vignette-specific indicators and individual-level controls and cluster the standard errors by respondent.In concrete terms, we regressed the support for AAP on the group the AAP is targeting (women, persons with a migration background, native East Germans or persons from a non-academic household), the segregation level faced by this group (high or very high), the sector of activity in which the AAP will be implemented (private or public), and the managerial position (senior, middle or lower).Additionally, we included the perceived disadvantages a target group faces, competence and performance as moderators, and controlled for respondents' age, educational level, employment sector, whether they fulfil a managerial role, whether they are married/live in a civil partnership, and whether they have children or not.

General support for AAP
Figure 1 depicts the overall support for our four hypothetical AAP (the corresponding regression coefficients can be found in Table S3, Model 1.).AAP targeting women and persons from a non-academic household find significantly more support than those aimed at individuals with a migration background or native East Germans.Note that the predictions plotted in Figure 1 represent the tendency to support an AAP by respondents not belonging to any of our target groups.
We gain a more precise insight by subdividing our analysis according to our sampling groups.Figure 2 illustrates the differences in support for our four hypothetical AAP by the sampling groups.We plotted the predicted probabilities for respondents belonging to a specific target group against respondents not belonging to this target group (i.e., male vs female, individuals with vs without migration background, native East vs West Germans and persons from an academic vs non-academic household).
Figure 2 shows that the primary overall trend from Figure 1 is replicated within all sampling groups: Non-members of the target group (in yellow) support AAP aimed at women and individuals from non-academic households to a greater extent than regulations for migrants or East Germans.For members of a target group (in blue), we find some variation to this pattern; most importantly, we see that members of a target group differ substantially from non-members in their support for the AAP targeting their own  S3 of the Supporting Information).
group.However, this preference for their own group is less pronounced for persons from a non-academic household.

Differences in support for AAP between respondents belonging to different target groups
Table 1 provides the main effects for supporting the four proposed AAPs separately.It shows that being a member of the AAP-targeted group significantly and substantially increases support for such a regulation (see regression coefficients for 'sampling groups (SG)').This finding aligns with the group-based interest perspective and our Hypothesis 1.Additionally, we note that women are more likely to support any AAP proposed.To a lesser extent, this also holds for respondents with a migration background: They support all AAP to a significantly larger extent except the one aimed at native East Germans.In contrast, native East Germans and individuals from non-academic households support only AAP targeting their own group to a significantly larger extent. 2  We can interpret these different rationales for supporting a specific AAP according to our Hypotheses 2a and 2b: While women and interviewees with a migration background tend to show solidarity with members of other underrepresented groups, East Germans and persons from a non-academic household respond according to group competition theory.S4 of the Supporting Information.
Notably, for all four AAP groups, the vignette dimensions cannot explain any differences in support for the AAP.Solely the support for an AAP targeting native East Germans decreases if the regulation affects the private sector (as opposed to public service).This effect is statistically significant but small in magnitude (0.26 on an 11-point scale).In the same vein, our controls are not predictive for respondents' support towards a specific AAP.Only age is negatively correlated with support for an AAP aimed at migrants, and higher levels of education slightly decrease support for AAP targeting women. 3

Perceived disadvantage and prejudice
We hypothesised that the support for an AAP might be moderated by the perceived disadvantage members of an underrepresented group face and by prejudice.The latter is measured as the attested competence and the willingness to perform at a high level or a perceived lack of those components (see section on 'Operationalisation').With respect to perceived disadvantage, we first give an overview of how disadvantage of different underrepresented groups is evaluated by the respondents, and then use data from the German Socio-Economic Panel (SOEP) to assess how these evaluations match the real representation of underrepresented groups in leading positions in Germany.Table 2 gives the averages of the perceived disadvantage scores divided by SG for the different target groups.The second column includes the values for respondents not belonging to any target group, i.e., West German men from academic households without migration background.These respondents perceive women and persons with a migration background as slightly disadvantaged when applying for managerial positions (recall: we measure perceived disadvantage on a five-point scale; 1 = not disadvantaged at all, 5 = very strongly disadvantaged).
In contrast, native East Germans and individuals from non-academic households are not seen as greatly disadvantaged.Overall, we see only little variation in the perception of group-based disadvantages in the recruitment process for managerial positions (Table 2).
We can back-up these observations by comparing the individual perceived disadvantage with objective measurements of disadvantage.The SOEP entails different measurements that quantify whether and to which degree different groups are excluded from managerial positions.Following Holst and Friedrich (2017), Table 3 summarises different ways of identifying employees in the private sector with managerial functions.The first two make use of the self-reported occupational position.Further, we define a manager according to the ISCO-08 and the ISCO-88 scales.The Classification of Occupations (KldB, 2010) developed by the Federal Employment Agency offers a third possibility for operationalisation.Lastly, we operationalise managers with the help of the Erikson-Goldthorpe-Portocarero scheme (Brauns et al., 2000).This scheme is based on the ISCO-88 or ISCO-08 occupational classification; the occupations are scaled using the ISEI index (Ganzeboom & Treiman, 1996) based on their socio-economic occupational status.This results in 11 classes, the first two of which identify managerial and highly skilled workers.
To comprehend the extent to which members of a minority are disadvantaged in the labour market when it comes to positions with a managerial function, we can compare the proportion of individuals belonging to one of our target groups with a managerial role with the ratio of these groups in the full-time employee population over 18. Independently of the measurement used in Table 3, we observe that the percentage of target group members who are full-time employees is always higher than the ratio with a managerial role.This fact indicates clearly that members of the target groups are underrepresented in managerial positions.Compared to the assessment of our respondents not belonging to any target group, we observe that native East Germans and persons from a non-academic household are also underrepresented in positions with a managerial role, even though this is at a lower level than for people with a migration background and for women.In this sense, the respondents to our survey did a good job mapping the hierarchy of underrepresentation, placing women and migrants at the top and native East Germans and individuals from non-academic backgrounds at the bottom.How does the perceived disadvantage (or lack thereof) of the four target groups affect the support for AAPs targeted at these groups?
On the basis of Table 1, we can essentially state that the support for AAP is higher when a group is perceived as disadvantaged when applying for managerial positions.This effect is sizeable and significant for all four target groups and aligns with our theoretical assumptions (Hypothesis 3).Next to the perceived disadvantage, we also hypothesised that prejudice against a target group might moderates the support for an AAP.We measured prejudice as the attested competence and the willingness to perform at a high level or a perceived lack of these components.Descriptively, we observe only little variation for these two measures (see Table 4. Note: a 0 means that the two groups have the same competency or willingness to perform.A positive value means that a member of a target group is attested to have more competence or willingness to perform; a negative value points to the perception that a member of the majority group has more competence or willingness to perform).We see that all values are slightly negative for our control group (i.e., respondents not belonging to any target group).However, they are very close to 0, so that we can imply that the majority of respondents do not consider there to be varying levels of competence or performance between social groups.
In our multivariate model (see Table 1), only the AAP targeting women is affected positively by the competence ascribed to members of our target groups.In other words: If women are seen as more competent in a leadership position than men, this leads to higher support for a regulation aimed at women.Nonetheless, this effect is relatively small in magnitude.The perceived level of performance affects only the support for AAP towards migrants: If respondents perceive migrants to perform better than nonmigrants, they support AAP aimed at migrants to a higher extent.Again, this effect is statistically significant but small in magnitude.Thus, we can only partly confirm our theoretical assumptions formulated in Hypothesis 4. 4  Overall, the lack of variation in means over the different groups in both measures of perceived competence and performance probably explains the non-significant effect between these attitudinal measures and support for the corresponding AAP.
We have illustrated the differences in the predicted support for a specific AAP by the respective sampled groups in Figure 3.We plotted results from two models, with and without the inclusion of our three moderating variables (i.e., perceived disadvantage in the labour market, perceived level of performance and perceived level of competence).Graphically, we see again that these three measurements combined only partly explain the differences in support for AAP.

Discussion and conclusion
This study examined support for AAP targeting four underrepresented groups in the German labour market.We exploited a survey experiment randomly varying the target group for a hypothetical regulation in the recruitment process for a management position favouring either women, persons with an immigrant background, native East Germans or persons from a Note: OLS regressions with clustered standard errors for respondents.The figure is based on the models in Table 1; with and without the inclusion of the moderating variables.
non-academic household.We argued that the extent to which individuals support an AAP is contingent first on their (ascribed) group membership, second on the level of perceived disadvantage of the target group, and third on the level of prejudice held against the target group.
Generally, our results show that AAP targeting women and persons from non-academic households are supported to a larger extent than regulations aimed at East Germans and persons with an immigrant background.This could hinge on the fact that being East German or having a migration background touches upon regional membership in a broader sense, whereas being a woman or from a non-academic household are ascriptions which are more broadly distributed in the general population.Besides, this finding highlights that persons from a non-academic household comprise a salient social category in Germany.
Furthermore, welfare chauvinism could explain low support for regulations aimed at persons with a migration background.Another plausible explanation points to a more general problem of identity politics: the requirement for clear-cut criteria for group membership.In the case of East Germanand to a lesser degree for persons with a migration backgroundthe definition of who should belong to the target group is not uncontested and clear-cut, as boundaries between some categories of distinctions tend to blur (i.e., native East and West Germans).
Our finding pinpoints consistently that being a member of the AAP-targeted group increases the support for such a regulation, aligning with the group-based interest perspective.Notably, we find that while women in particular are more likely to support any AAP proposed, native East Germans and individuals from a non-academic household support only AAP targeting their own group.Additionally, the support for AAP is higher if a group is perceived as disadvantaged when applying for managerial positions.Lastly, we find only partial evidence for the hypothesised link between prejudice against an underrepresented group and the support for an AAP aimed at this group.
One important question that arises from these results is why women show more solidarity towards other underrepresented groups than members of other minorities.We can enqueue this finding in line with a 'modern gender gap' in political attitudes (Goossen, 2020).According to this gender gap, women are generally more in favour of left-wing parties and policies, including support for a comprehensive redistribution of resources and welfare provision (e.g., Inglehart, 2018;Inglehart & Norris, 2003;Shorrocks & Grasso, 2020).From a socio-psychological perspective, another plausible explanation could be found in the higher level of empathy women generally show compared to men (Mestre et al., 2009), especially towards ethnic and cultural minorities (Cundiff & Komarraju, 2008).
These findings have clear implications regarding inequality and its perception in society.Following Mijs (2021) and Mijs and Savage (2020), the rise in inequality and belief in meritocracy have gone hand in hand.The authors argue that individuals consent to inequality as they are convinced that societal success reflects a meritocratic process.The more unequal a society, the more individuals tend to explain success in meritocratic terms and less through non-meritocratic factors such as ascribed characteristics.In this sense, the underrepresentation of minorities in managerial positions is coupled with a feeling of deservingness (as also backed up by our results) legitimising the underrepresentation itself.AAP can help overcome this misperception and actively contribute to an equality of outcomes.
Clearly, more work is needed to identify individual motivations to support AAP in other contexts and for other minorities.This study focused on one particular case of AAP (recruitment process for leading positions).However, there are many ways in various societal areas and sectors of activity where AAP could be implemented.Further work is also necessary to better ascertain the mechanisms through which support for AAP is amplified or reduced and to assess whether the underlying mechanism we find here also applies when looking at support towards other types of AAP.

Notes
1. We refrain from including LGTBIQ+ as the fifth target group in our vignette experiment.The introduction of quotas favouring members of the LGTBIQ+ group if equally qualified in recruitment processes for leading positions would require candidates to indicate their sexual orientation in their application.The implementation of such a regulation would therefore be much more challenging for the LGTBIQ+ group than the other underrepresented groups of our survey experiment.In order to provide vignettes that are as realistic as possible, we decided against the inclusion of LGTBIQ+ as the fifth AAP target group.2. These results are robust when considering that the sampling group categories are not mutually exclusive, meaning that a member of one sampling group can, at the same time, also be a member of another sampling group.Figures S1-S4 of the Supporting Information also illustrate that these findings persist when including interactions among the different sampling groups.3. The results are not altered by the introduction of indicators for the sequence with which we presented the vignettes to the respondents (see Table S5-S7 in the Supporting Information).4. In Tables S8-S12 of the Supporting Information we introduced summarising scales for our three moderating measurements.We constructed these scales using principal component analysis.The results are not altered when we specify our models with the summarising scales instead of the separate items.
comments on an earlier version of this article.Last but certainly not least, the authors would like to thank the anonymous reviewers and editors at JEPP for their constructive feedback and support.

Figure 1 .
Figure 1.Main effects support for AAP.(The figure is based on Model 1, TableS3of the Supporting Information).

Figure 2 .
Figure 2. Main effects support for AAP by sampling group.Note: The figure shows results for fully interacted models based on TableS4of the Supporting Information.

Table 1 .
Main results by AAP group.

Table 2 .
Means in perceived disadvantage for the different AAP groups.
Means of the control group (e.g., respondents not belonging to any target group) are bold.(Seealsodescriptive statistics, TableS1and S2 of the Supporting Information).

Table 3 .
Managers in the private sector according to different operationalisations.

Table 4 .
Means in ascribed competency and performance for the different AAP groups.Means of the control group (e.g., respondents not belonging to any target group) are in bold.(See also descriptive statistics, TableS1and S2 of the Supporting Information).