Sage Journals: Discover world-class research

Abstract

Discriminatory behavior is often assumed to partly be a result of implicit or unconscious biases. However, to my knowledge, no published experimental studies exist on the presence of such attitudes among public officials such as teachers or police officers and the related effects on real-life client discrimination. This study attempts to fill this research gap by combining a field experiment that captures ethnic discrimination among principals with an implicit attitude test (IAT) capturing their implicit attitudes toward Arab parents. The results suggest that Swedish elementary school principals have a moderately negative implicit bias toward Arabs according to the IAT. However, their implicit associations do not appear to drive the differential treatment of clients to a significant degree. Hence, implicit biases as captured by the IAT may possibly not be the most significant drivers of ethnic discrimination today among public officials.

Keywords

Discrimination implicit bias field experiment public officials implicit attitude test prejudice

Introduction

Recently, we have witnessed a rapidly growing body of experimental evidence in political science and public administration demonstrating that public officials discriminate against ethnic minorities, even in established democracies (for a review, see Costa, 2017.²). These findings are disturbing, as public officials are expected to uphold fundamental democratic principles such as equal treatment. Additionally, ethnic minorities who experience discrimination and prejudice may be less likely to trust governmental institutions and less likely to participate in politics (cf. Crowley, 2001). Alongside this line of empirical research, the search for theories explaining why public officials discriminate has increased.

The wider discrimination literature in economics, sociology, and political science has mainly focused on taste-based explanations as well as statistical/strategical discrimination. However, inspired by social psychology research, recent highly cited articles have also highlighted implicit biases where discrimination is assumed to be a result of automatic, spontaneous, implicit, or impulsive processes operating outside of our awareness (Bertrand et al., 2005; Quillian, 2006). Such implicit biases are also mentioned by political scientists as one of several possible mechanisms behind discrimination among public officials (see, e.g., Olsen et al., 2020: 3; Jilke et al., 2018: 425; Hemker and Rink, 2017: 801). In line with this, substantial research in social psychology performed with students in laboratory conditions has shown that automatically activated stereotypical perceptions can potentially influence a wide range of social judgments, decisions, and behaviors (Kurdi et al., 2019; Greenwald et al., 2020).

To draw valid conclusions regarding the role of implicit biases in explaining client discrimination based on ethnicity, however, we also need real-world evidence, preferably related to public officials. The results based on laboratory studies of anonymous students are not necessarily generalizable to real interactions between, for example, teachers/police officers and their clients, where the perceived costs of discrimination and social desirability bias are likely much higher. Field experiments on the impact of automatic biases on real-world discriminatory behavior are still extremely rare in the social sciences (especially in comparison to laboratory studies; see Kurdi et al., 2019: 10; Greenwald et al., 2020: 50; Oswald et al., 2013: 172).³ Notable exceptions are the published study by Rooth (Rooth, 2010) finding a relationship between negative implicit association and ethnic discrimination among recruiters and the study by Alesina et al. (2018) finding a relationship between implicit stereotypes among teachers and grade discrimination.⁴ However, the former does not focus on public officials/client discrimination, while the latter is an unpublished working paper. Differences between private sector employees and public officials, for example, in terms of public service motivation (cf. Perry et al., 2010), may decrease generalizability. It is hence still an open question whether public officials actually hold implicit biases and whether these biases affect their behavior.

This article attempts to fill this research gap by presenting one of the first studies on the presence of implicit discriminatory attitudes and their effects on real-life ethnic discrimination⁵ among public officials. Differential treatment is captured using a correspondence experiment. A total of 3430 elementary school principals in Sweden were randomly contacted via email by fictional parents interested in placing their children in the school and inquiring about the school and the admission procedure. The parents’ ethnicity (signaled with Swedish- and Arabic-sounding names), professions, and gender were randomized. Both the degree of responsiveness to the emails from the aliases and the friendliness of the replies are studied. The latter dimension is important, as more informal communication may be more likely to be affected by implicit biases.

Implicit biases are captured using an implicit attitude test (IAT; Greenwald et al., 2003), which was emailed to all the principals after the field experiments. The IAT is a computer-based test designed to measure individual differences in relative associations between concepts. More specifically, the principals had to rapidly categorize two concepts (Arab- and Swedish-sounding names) with unpleasant and pleasant words, with easier pairings (faster responses) interpreted as being more strongly associated in memory than more difficult pairings (slower responses), thereby indicating the presence of implicit associations. Even though the test is not without its drawbacks (for a review of its strengths and weaknesses, see De Houwer et al., 2009), it is still by far the most established method in the social sciences of capturing implicit discriminatory attitudes (Greenwald et al., 2020). Several large meta-studies have found small but arguably important correlations between IAT test scores and discriminatory behavior, even after taking into account different study characteristics and potential publication bias (Kurdi et al., 2019; Greenwald et al., 2020; for a more critical meta-study, see Oswald et al., 2013). That the IAT could potentially capture more unconscious than conscious attitudes is supported by evidence showing that IAT effects are more difficult to control (fake) than traditional questionnaire responses and that the IAT can register attitudes even when participants do not know the origin of those attitudes (see De Houwer et al., 2009).

The study focuses on elementary school principals, as schools constitute important settings for public service provision involving frequent engagement with the general public, and the decisions made by principals have potentially large consequences for their clients. This is also a context in which it is reasonable to expect to find a relationship between implicit biases and the differential treatment of clients. It is often argued that humans are more likely to be influenced by unconscious prejudice in situations that are very demanding of cognitive resources (cf. Bertrand et al., 2005), and school principals are generally subject to a much higher degree of stress than the population in general (Läraren, 2020). In line with this, the principals studied in our sample reported that they often felt stressed and frequently could not cope with all the things that they had to do. The principals are employed at publicly financed private or public schools in Swedish municipalities, which have the administrative responsibility for organizing and financing the school system (for more information, see Appendix F).

The focus is on an interesting empirical setting that has rarely been studied before: a Scandinavian welfare state (Sweden). Most of the studies on implicit discriminatory attitudes have been conducted in the US (cf. Kurdi et al., 2019; Oswald et al., 2013). Sweden has a reputation for being an immigration-friendly welfare state. Nineteen percent of the population is foreign born (Statistics Sweden, 2019). Sweden is often described as a European outlier, with public opinion showing particularly positive views toward ethnic minorities in comparison with those in other countries (World Value Survey, 2014). From this perspective, we may be less likely to find (implicit) discrimination in Sweden than in the US and the rest of Europe. However, the general discourse toward migrants from the Middle East has hardened in recent years, and this group is now politicized and often subject to discrimination in many societal arenas (e.g., Rooth, 2010; Taghizadeh, 2022). Hence, Sweden may be more representative of northern Europe than is typically assumed.

Methods

Differential treatment of clients is captured through a correspondence experiment.¹ Every elementary school principal heading a unique school in Sweden⁶ was contacted via email by one randomly assigned fictional parent (N = 3430) between 11 and January 14, 2020. Almost all the principals who were emailed (96%) had Swedish-sounding names.

Treatments

A factorial design is employed based on ethnicity, socioeconomic status (SES), and gender.⁷ An advantage of factorial designs is their efficiency with respect to the number of experimental subjects; for a given number of treatments, factorial designs require fewer experimental subjects than alternative experimental designs to maintain the same level of statistical power (cf. Collins et al., 2009). For ethnicity treatment, immigrants with a background in the Middle East were chosen because they constitute an important and politicized ethnic minority group in Sweden.⁸ The male name Mahmoud and the female name Fatimah were chosen, both with the surname Hassan, which is rather common in Sweden (Statistics Sweden, 2021). Names usually carry with them not only an ethnic association but also a certain SES association, with the names of individuals belonging to ethnic minorities more strongly associated with low educational attainment and low income. As a result, most previous experiments on the presence of discrimination and implicit biases may have confused ethnic discrimination with SES discrimination. To avoid the possibility that the results are driven by SES associations connected to the chosen names, Swedish-sounding names associated with SES levels similar to those of the Arabic names were used in both the experiment and the IAT (e.g., Kevin and Melissa; see Appendix D and Tables 6 and 7 for information regarding the pretest on names). To provide additional controls for SES and to increase generalizability, the study also includes treatments for high-skilled professions (dentist) in half of the e-mails and low-skilled professions in the other half (care assistant).

Content of the emails and coding scheme

The letter sent to the principals can be found in Figures 3 and 4 in Appendix E. The emails were written as if they were sent by someone considering moving to the municipality to avoid arousing suspicion among principals in very small municipalities. They contained three straightforward and not particularly time-consuming questions regarding the school’s profile, registration procedure, and open slots.

The coding scheme used in the article consists of 4 variables (see Table 1 below and the coding examples in Appendix B, Figure 2). Two of these measure formal aspects, that is, whether any reply was received within 2 weeks (0–1) and how many of the three questions were answered (0–3). The third variable is an informal friendliness index (0–5).⁹ More specifically, it measures whether the replier invited future contact (1p), welcomed the sender to the municipality/school (1p), invited the sender to visit the school (1p), provided additional information not directly related to the three questions (1p) and/or used the name of the sender (1p). The fourth variable captures whether the email informed parents that there were open slots at the school: information that could be used to directly encourage parents to choose that particular school.

Table 1.

Summary statistics for email-dependent variables (N = 3394).

Dimension	Variable	Variable description	Mean	Std. dev.
Formal correctness	Reply (1,0)	Did the recipient respond to the email within 2 weeks? Replies were registered from anyone working at the school or the municipality of the school, except for autoreplies and noninformative emails from principals who had left their position.	0.749	0.434
Formal correctness	Questions answered (0-3)	How many of the three questions are answered?	1.653	1.226
Friendliness	Friendliness index (0-5)	The replier invites future contact (1p), welcomes the sender to the municipality/school (1p), invites the sender to visit the school (1p), provides additional information not directly related to the three questions (1p) and/or use the name of the sender (1p)	1.261	1.282
Promotion	Open school slots (1,0)	The replier mentions that there are clearly/likely open slots at the school	0.249	0.433

Nonresponses were coded as zero for all variables to avoid inducing selection bias and conditioning on a posttreatment variable (cf. Coppock, 2019). Moreover, the emails were independently coded by two research assistants. After coding all responses independently, the assistants were instructed to pay special attention to emails where their codings differed to reconcile the final coding. Regarding the simplest variables (e.g., reply), differing cases were reviewed by a third assistant. The names of the fictitious emailers were removed before coding started.

Of the original 3430 emails sent, 3394 units of analysis were ultimately included in the field experiment dataset. Thirty-one emails bounced back and were therefore excluded from the analysis. Five emails were also excluded because the principal replied that she or he had left the position and the email was not forwarded further. No signs of spillover or disclosure of the experiments were found in the email responses. Research ethics are discussed in Appendix C.

Implicit attitude test (IAT)

Emails inviting the principals to conduct a survey and an IAT were sent out to all 3430 email addresses used in the field experiment on 3 February 2020 (see Appendix I for the invitations). The IAT is described in greater depth in Appendix H and Tables 11 and 12. Of the 3394 unique emails to principals who were finally included in the study, 447 (13%) completed the IAT correctly. Only two of the principals started the IAT without fully completing it or did not complete it correctly. Since most of the principals did not take the IAT, it is possible that participation was selective, which could bias the estimated correlation between the implicit measure and discrimination and decrease generalizability. The data show a very small but statistically significant correlation coefficient (0.0624***) between responding to emails in the field experiment and participating in the IAT. Hence, principals who participated in the IAT were slightly more responsive than the average principal in the experiment, decreasing generalizability from the IAT sample. However, the IAT sample appears to be relatively representative of the full experimental sample based on a comparison of municipality- and school-level characteristics (Table 9 in Appendix G) and the general discrimination effects (Table 2 below). Most importantly, we see very similar results in regard to the statistically significant ethnic discrimination effect found on friendliness (comparing Models 5 and 6 in Table 2 ¹⁰), showing no clear selection bias in regard to the dependent variable that is most likely to be affected by the IAT scores. The ethnic treatments also appear to be fairly balanced in the IAT dataset relative to IAT scores and school and municipality characteristics (Table 8 in Appendix G).

Table 2.

Responsiveness to different client groups among Swedish school principals.

	Reply		Questions answered		Friendliness index		Open slots
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Arab	−0.008 (0.017)	−0.044 (0.034)	−0.022 (0.045)	−0.010 (0.102)	−0.224^*** (0.045)	−0.343^** (0.125)	−0.031^* (0.015)	−0.088^* (0.041)
Low-SES	−0.013 (0.016)	−0.044 (0.037)	−0.034 (0.041)	−0.141 (0.112)	−0.178^*** (0.042)	−0.160 (0.122)	−0.044^** (0.014)	−0.106^* (0.041)
Woman	−0.017 (0.013)	0.052 (0.041)	−0.032 (0.040)	0.056 (0.128)	0.026 (0.043)	0.013 (0.147)	0.002 (0.015)	0.034 (0.040)
Constant	0.828^*** (0.087)	0.945^*** (0.174)	1.990^*** (0.265)	2.201^*** (0.533)	2.850^*** (0.294)	3.452^*** (0.748)	0.623^*** (0.098)	0.989^*** (0.196)
Data	Full	IAT	Full	IAT	Full	IAT	Full	Full
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	3394	447	3394	447	3394	447	3394	3394
Adj. R²	0.001	0.026	0.002	0.006	0.044	0.064	0.027	0.055

Notes: ⁺ p <0.10, * p <0.05, ** p <0.01, *** p <0.001. Standard errors (in parentheses) are clustered at the municipal level. Control models include controls for private schools, municipal population, proportion of foreign-born residents, proportion of the municipality that voted for the far right (Sweden Democrats), and average income in the municipality.

To be safe, the results are also presented with controls for the most important municipality characteristics.

Results

If implicit biases are a significant driver of discrimination among public officials, we expect to find evidence of implicit biases according to the IAT as well as a correlation between IAT scores and the differential treatment of clients among the school principals. This is explored in the following sections.

Presence of implicit biases

As seen in Figure 1, the results show, on average, a moderate bias (d score = 0.453, st dev = 0.379; see Figure 1) whereby Arabic names are associated with unpleasant/negative words. A total of 148 principals (33.1%) had a moderate bias (0.35–0.65), and 132 principals (29.5%) had a strong bias (>=0.65) (cf. Greenwald et al., 2003). Hence, the results provide evidence of implicit bias, with most principals in the sample finding it significantly easier to associate Arab Muslim males with unpleasant words than with pleasant words. The IAT scores do not correlate with explicit measures of prejudice (see Appendix A, Table 4).

Figure 1.

The distribution of nonstandardized IAT scores for implicit attitudes against Arab minority men relative to nonmigrant Swedish men.

Differential treatment of clients

Table 2 presents the results regarding the differential treatment of clients among Swedish principals. While most of the coefficients are negative, suggesting that Arab and low-SES parents are treated worse, no negative statistically significant effects are found for the response rate or the number of questions answered (models 1–4). However, on the friendliness index, evidence of possible discrimination against parents with Arabic aliases—namely, negative effects that are statistically significant below the 1% level—is found in both the full sample and the IAT sample (models 5–6).¹¹ The Arab parents are less also less likely to be provided information about open slots (models 7–8, statistically significant below the 5% level).

Effects of IAT scores on the differential treatment of clients

Table 3 tests whether principals with stronger automatic negative associations for Arabs are more responsive to emails signed with Swedish names versus Arabic names using an interaction variable between the IAT score and the Arab treatment. None of the IAT interaction effects with the Arab treatment are statistically significant below the 10% level, regardless of whether controls are included (see also Table 10 in Appendix G). The largest coefficients are found on the response rate (models 1–2), where principals with strong negative associations for Arabic names (d score = 1) are 8 to 9 percentage points less likely to respond to Arabic parents versus Swedish parents than principals with no biases (d score = 0, baseline 81.7–94.9 response rate). The interaction effect is illustrated in Appendix J, Figure 5. For the other outcomes, the coefficients are very small and mainly positive.

Table 3.

Effects of the interaction between the IAT and the Arab treatment on the response rate, questions answered, friendliness index, and information about open slots.

	Reply		Questions answered		Friendliness index		Open slots
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
IAT*Arab	−0.083 (0.090)	−0.094 (0.090)	0.005 (0.307)	0.001 (0.310)	0.078 (0.314)	−0.043 (0.317)	0.069 (0.107)	0.020 (0.106)
IAT	0.041 (0.057)	0.042 (0.057)	−0.114 (0.191)	−0.146 (0.200)	0.127 (0.221)	0.174 (0.226)	−0.048 (0.076)	−0.018 (0.078)
Arab	0.003 (0.052)	−0.001 (0.050)	−0.009 (0.177)	−0.009 (0.179)	−0.343⁺ (0.179)	−0.324⁺ (0.180)	−0.110⁺ (0.064)	−0.097 (0.063)
Constant	0.817^*** (0.046)	0.949^*** (0.176)	1.975^*** (0.141)	2.264^*** (0.536)	1.707^*** (0.172)	3.387^*** (0.755)	0.428^*** (0.054)	0.992^*** (0.192)
Controls	No	Yes	No	Yes	No	Yes	No	Yes
N	447	447	447	447	447	447	447	447
Adj. R²	0.002	0.024	−0.005	0.003	0.007	0.062	0.010	0.050

Notes: ⁺ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. Standard errors in parentheses are clustered at the municipal level.

Discussion and conclusion

In this article, the presence of implicit biases and their possible effects on ethnic discrimination among Swedish school principals were studied. Two interesting results with important implications for the wider discrimination literature as well as for political science and public administration research were reported.

First, in line with what is often argued in the academic debate, the results revealed the presence of implicit biases among public officials (assuming that the IAT captures such attitudes). More specifically, the results showed a negative moderate bias whereby Arabic names were associated with unpleasant/negative words (d score = 0.45). Interestingly, the level of bias found was almost identical to that estimated in a similar unpublished study on implicit bias toward immigrants among Italian teachers (Alesina et al., 2018, d score = 0.47) but lower than that found in a published study on implicit bias toward Arabs among Swedish recruiters (Rooth, 2010, d score = 0.65). The samples in the three studies were, however, too small and unrepresentative to allow valid conclusions to be drawn regarding whether school principals or teachers have lower biases than other groups in society, such as private employees.

Second, principals who showed stronger negative implicit biases according to the IAT were not significantly more responsive to parents with Swedish names versus Arabic names in terms of response rate, questions answered, friendliness, or information about open slots. The statistically nonsignificant results contrasted with those found for Swedish recruiters (Rooth, 2010) and Italian teachers (Alesina et al., 2018), for whom small but statistically significant relationships between IAT scores and discrimination were estimated.

The lack of large and statistically significant effects of the IAT can be interpreted in several ways. One interpretation is that as a result of the low response rate (14%, n = 447), the study was insufficiently powered to identify important effects of the IAT on discriminatory behavior. One consideration that speaks against this, however, is that the study is one of the largest of its kind. Meta-studies have been able to identify the effects of IAT-scores on behavior even though most of the reviewed studies have samples of only approximately 40 observations (40 in Kurdi et al., 2019, and 41 in Oswald et al., 2013). Similarly, the study by Rooth covers 158 recruiters in each IAT but is still able to identify effects on callback rates. Furthermore, according to my power analyses, the sample is at least large enough to rule out the presence of major discrimination effects.¹²

Another possibility is that the lack of statistically significant negative discrimination effects on reply and questions answered made it more difficult to identify interaction effects with IAT. While this is true for these dependent variables, clear discrimination effects were found on both the friendliness index and information regarding open school slots, increasing the likelihood of identifying IAT effects in these cases. The discrimination effects found are substantially important, as the lower degree of friendliness and the lack of information regarding open slots in the responses to Arabic parents could deter them from asking further questions or visiting the school, which could influence their school choice decision.

A potentially more accurate interpretation, then, is that the effects of IAT scores on ethnic discrimination are limited in the sample of Swedish public officials and the type of client interactions studied (email responses). The test–retest reliability of IAT measures is generally modest, suggesting that much of the variance in implicit biases is situationally dependent (Greenwald et al., 2020; Payne et al., 2017). It is also well established among social psychologists, even among proponents of the IAT, that the ability to predict discriminatory behavior of IAT measures is, on average, low (Greenwald et al., 2020; Oswald et al., 2013: 565). These conclusions are primarily based on laboratory studies with students. It is possible that the effects are even smaller in regard to real-life client discrimination among public officials, as studied here. Furthermore, previous IAT-based studies may have overestimated implicit ethnic discrimination by not using majority (white) names that are associated with a similar degree as minority names with low SES (cf. Appendix D). Overall, the lack of large effects in my study and in the meta-studies suggests that implicit discriminatory attitudes (at least those captured by the most established test available) may not be the most significant drivers of ethnic discrimination today among public officials.

Given the limited number of studies on the topic and their limitations (including those of my own study), it is clear that we need more studies to draw reliable conclusions regarding the relationship between IAT scores, implicit biases and real-life discriminatory behavior, especially in relation to public officials. This is important considering the attention that implicit biases have received both in the discrimination literature and in society at large.

Supplemental Material

Supplemental Material - The effects of implicit biases on real-life client discrimination among public officials

Supplemental Material for The effects of implicit biases on real-life client discrimination among public officials by Jonas Larsson Taghizadeh in Research & Politics

Footnotes

Acknowledgments

I want to thank Per Adman for helping me design the field experiment.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Swedish Research Council for Health, Working Life and Welfare (2019-00504) as well as the Swedish Research Council (2019-03305)

Correction (April 2025):

The article has been updated with correct dataverse link in the supplemental material section. For more details, please see the correction notice: .

ORCID iD

Jonas Larsson Taghizadeh

Supplemental material

The supplementary files are available at .

The replication files are available at: .

Notes

References

Alesina

Carlana

La Ferrara

, et al. (2018) Revealing stereotypes: Evidence from immigrants in schools. NBER Research Working paper 25333. Available at: https://www.nber.org/system/files/working_papers/w25333/w25333.pdf (accessed 25 May 2021).

Bertrand

Chugh

Mullainathan

(2005) Implicit discrimination. American Economic Review 95 (2): 94-98.

Blommaert

Van Tubergen

Coenders

(2012) Implicit and explicit interethnic attitudes and ethnic discrimination in hiring. Social Science Research 41 (1): 61-73.

Collins

Dziak

Runze

(2009) Design of experiments with multiple independent variables: A resource management perspective on complete and reduced factorial designs. Psychological Methods 14 (3): 202.

Coppock

(2019) Avoiding post-treatment bias in audit experiments. Journal of Experimental Political Science 6 (1): 1-4.

Costa

(2017) How responsive are political elites? A meta-analysis of experiments on public officials. Journal of Experimental Political Science 4 (3): 241-254.

Crowley

(2001) The political participation of ethnic minorities. International Political Science Review 22 (1): 99-121.

De Houwer

Teige-Mocigemba

Spruyt

, et al. (2009) Implicit measures: A normative analysis and review. Psychological Bulletin 135 (3): 347-368.

Greenwald

Nosek

Banaji

(2003) Understanding and using the implicit association test: I. An improved scoring algorithm. Journal of Personality and Social Psychology 85 (2): 197-216

10.

Greenwald

Brendl

Cai

, et al. (2020) “The Implicit Association Test at age 20: What is known and what is not known about implicit bias”. Unpublished manuscript. Available at: https://faculty.washington.edu/agg/pdf/IAT%20at%20age%2020.with%20title%20page.26Mar2019.pdf (accessed 25 May 2021).

11.

Hagiwara

Penner

Gonzalez

, et al. (2013) Racial attitudes, physician–patient talk time ratio, and adherence in racially discordant medical interactions. Social Science & Medicine 87: 123-131.

12.

Hemker

Rink

(2017) Multiple dimensions of bureaucratic discrimination. American Journal of Political Science 61 (4): 786-803.

13.

Jilke

Dooren

Rys

(2018) Discrimination and administrative burden in public service markets: Does a public–private difference exist? Journal of Public Administration Research and Theory 28 (3): 423-439.

14.

Kurdi

Seitchik

Axt

, et al. (2019) Relationship between the implicit association test and intergroup behavior: A meta-analysis. American Psychologist 74 (5): 569-586.

15.

Läraren (2020) Var fjärde rektor riskerar att bli utbränd. Available at: https://www.lararen.se/chefochledarskap/utbrandhet/var-fjarde-rektor-riskerar-att-bli-utbrand (accessed 25 May 2021).

16.

Olsen

Kyhse‐Andersen

Moynihan

(2020) The unequal distribution of opportunity: A National audit study of bureaucratic discrimination in primary school access. American Journal of Political Science. doi: 10.1111/ajps.12584.

17.

Oswald

Mitchell

Blanton

, et al. (2013) Predicting ethnic and racial discrimination: A meta-analysis of IAT criterion studies. Journal of Personality and Social Psychology 105 (2): 171-192.

18.

Payne

Heidi

Lundberg

(2017) The bias of crowds: How implicit bias bridges personal and systemic prejudice. Psychological Inquiry 28 (4): 233-248.

19.

Penner

Dovidio

West

, et al. (2010) Aversive racism and medical interactions with Black patients: A field study. Journal of Experimental Social Psychology 46 (2): 436-440.

20.

Perry

Hondeghem

Recascino

(2010) Revisiting the motivational bases of public service: Twenty years of research and an agenda for the future. Public Administration Review 70 (5): 681-690.

21.

Quillian

(2006) New approaches to understanding racial prejudice and discrimination. Annual Review of Sociology 32: 299-328.

22.

Rooth

(2010) Automatic associations and discrimination in hiring: Real world evidence. Labour Economics 17 (3): 523-534.

23.

Statistics Sweden (2019) Statistikdatabasen. Available at: https://www.statistikdatabasen.scb.se/pxweb/sv/ssd/ (accessed 25 May 2021).

24.

Statistics Sweden (2021) Svenska namnregistret. Available at: https://www.scb.se/hitta-statistik/sverige-i-siffror/namnsok/ (accessed 25 May 2021).

25.

Taghizadeh

(2022) Are more affluent parents treated more favourably by elementary school principals? Socioeconomic discrimination among local Swedish public officials. Local Government Studies. Available at: https://www.tandfonline.com/doi/full/10.1080/03003930.2021.1986391.

26.

World Value Survey (2014) World values survey wave 6: 2010-2014. Available at: http://www.worldvaluessurvey.org/WVSOnline.jspon (accessed 25 May 2021).

27.

Zigerell

(2018) Black and White discrimination in the United States: Evidence from an archive of survey experiment studies. Research & Politics 5(1): 1–7.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.20 MB