Working 9 to 5? A cross-national analysis of public sector worker stereotypes

ABSTRACT

We present an inductive, citizen-driven approach to identify stereotypes of public sector workers across the United States, Canada, the Netherlands and South Korea (Study 1: n=918; Study 2: n=3,042). Contrary to common negative portrayals, we idetify two positive stereotypes across countries — having job security and serving society; and one neutral/negative stereotype — going home on time. Notably, Americans and Canadians have a more favorable view of public sector workers than the Dutch and South Koreans. This study opens avenues for exploring positive public sector stereotypes and the impact of context on these perceptions.

    

KEYWORDS:

• Public sector stereotype
• public sector bias
• comparative public administration
• inductive research

1. Introduction

Public sector workers are often portrayed as lazy, incompetent, and even evil (Goodsell 2004; Hubbell 1991; Wilson 2019). For instance, former U.S. president Ronald Reagan depicted federal bureaucrats as loafers, incompetent buffoons, and tyrants (Hubbell 1991). The trend continues with Donald Trump using delegitimizing rhetoric and bureaucrat bashing, invoking conspiratorial theories of ‘deep state’ plots or calls to ‘drain the swamp’ (Moynihan 2021). Examples of negative stereotypes of public sector workers are ubiquitous, and the public management literature has seen rapid development on this topic (e.g. De Boer 2020; Döring and Willems 2021; Van de Walle 2004; Willems 2020).

This nascent body of work has already yielded important insights. For instance, Willems (2020) found that stereotypes of public sector workers are dependent on type of profession (i.e. police officer, nurse, teacher, public servant) and that, generally, they are quite positive. Additionally, De Boer (2020) examined perceived warmth and competence of various types of street-level bureaucrats among Dutch citizens and found that regulation-oriented civil servants – such as tax officials – are seen as less warm and less competent than those who are service-oriented, such as nurses.

Despite these important insights, there are still two important gaps limiting our understanding of public sector workers stereotypes. First, most studies are limited to the United States context (e.g. Caillier 2018; Goodsell 2004; Marvel 2016; Willems 2020) or single countries (De Boer 2020; Gilad, Ben-Nun Bloom, and Assouline 2018; Willems 2020). Nevertheless, Fiske (2017) shows that stereotypes can differ strongly across cultures – for example, stereotypes on ethnicity or religion differ based on intergroup relations shaped by the cultural and historical context of a region. Gender and age stereotypes are more universal (Fiske 2017). Since the public sector context can differ greatly across countries, the study of public sector worker stereotypes warrants a cross-cultural investigation. Scholars have noted that cross-country research in public administration is highly valuable to build a more comprehensive body of knowledge, guided by the contextual realities of the systems and practices of different regions (Haque 2013; Van der Wal, van den Berg, and Shamsul Haque 2021).

A second gap lies in the deductive nature of most studies (e.g. Frank and Lewis 2004; Goodsell 2004). Deductive approaches can be helpful to identify relationships and test hypotheses but are less suitable to uncover unexpected stereotypes. In other words, while we can deductively confirm common ideas about bureaucrats – such as laziness or rigidity, we may miss stereotypes that scholars may not think of – for instance that public sector workers are arrogant. Additionally, a deductive approach can reinforce stereotypes that people are presumed to harbour. Therefore, uncovering stereotypes inductively opens new avenues for theoretical progress.

Hence, we present a cross-national, inductive, citizen-driven approach to study citizens’ stereotypes of public sector workers. Using a two-step approach of two studies, we use citizens’ self-generated stereotypes to systematically assess people’s ideas of public sector workers across four different countries: the United States, Canada, the Netherlands, and South-Korea. We specifically study (a) which public sector workers stereotypes citizens have (i.e. the content of the stereotypes) and (b) the extent to which these stereotypes are seen as positive or negative (i.e. the valence of the stereotypes). This enables us to answer two research questions. First, what are the positive and negative stereotypes that citizens associate most frequently with public servants in their country? And second, to what extent do the stereotypes expressed in different national contexts overlap?

2. Public sector worker stereotypes

Stereotypes are ‘beliefs about the characteristics, attributes, and behaviours of members of certain groups’ (Hilton and Von Hippel 1996, 240). Public sector worker stereotypes are beliefs about the characteristics, attributes, and behaviours of people working in the public sector. Such stereotypes can overlap with stereotypes about the public sector in general (Goodsell 2004; Rainey and Bozeman 2000). Stereotypes have been studied from different perspectives (Bordalo et al. 2016). From a cognitive psychology perspective, stereotypes are mental schemas that people use to infer and interpret information about other people (Hilton and Von Hippel 1996). Alternatively, from a socio-cultural perspective, stereotypes are seen as cultural phenomenon, as ideas that exist in a society and are purported in media (Augoustinos and Walker 1998). Both perspectives are relevant for public sector worker stereotypes as stereotypes may influence individual decision-making such as when individuals are considering a public sector job, but they are also the topic of jokes and negativity in popular media (Lichter, Lichter, and Amundson 2000; Pautz and Warnement 2013; Van de Walle 2004).

Public management scholars have investigated public sector worker stereotypes for several decades, albeit not always using the same terminology. Hubbell (1991) wrote about bureaucrat bashing, where politicians paint public sector workers and their performance in a negative light to attract voters (see also Caillier 2018; Garrett et al. 2006). In his classic The Case for Bureaucracy, Goodsell (1983, 2000) argued that negative stereotypes about the public sector workers are undeserved and inaccurate. Similarly, Baldwin (1990) tested whether negative public sector worker stereotypes are valid and concluded that they are not (see also Frank and Lewis 2004). However, the crux of stereotypes is not the degree of their accuracy. Some stereotypes may have aspects that accurately reflect certain elements of the empirical reality – also referred to as the ‘kernel of truth’ hypothesis. But with stereotyping, assumed trait prevalence within a group and differences to other groups are overblown (Bordalo et al. 2016, 1771). Stereotypes typically represent social groups in an unfavourable light (Hilton and Von Hippel 1996, 245) and as inherently homogenous (ibid, p.240).

Relatedly, public services and public sector workers are often referred to as a whole, even though specific public sector occupations can differ greatly. In research on the public sector, citizen perceptions are often measured in terms of public-private dichotomies: The public sector is seen to perform worse than the private sector (Hvidman 2019; Hvidman and Andersen 2016; Marvel 2016; also referred to as public sector bias). Public sector workers are perceived as lazier, less hardworking, more boring, and less creative (Goodsell 2004; Marvel 2016) than private sector workers. Chen and Bozeman (2014) show that public managers, too, view public sector workers as less creative, less talented, and less autonomous than private sector employees. While some of these works study organizations rather than the individuals that work there, general images of the public sector may transfer to those of its workers (Hvidman and Andersen 2016). Döring and Willems (2021) show that the public takes cues from the professional context into account when processing stereotypical information. Thus, the broad category of public sector workers may be particularly relevant. Indeed, research has shown that citizens perceive them as ethical, dedicated, and helpful, but also as unmotivated, lazy, inefficient, bureaucratic, and slow (Baldwin 1990; Willems 2020), reflecting a conflated aggregation of public sector workers.

Since negative stereotypes may have severe effects, they deserve scholarly attention. Negative stereotypes can result in recruitment problems for the government, with highly skilled workers being less attracted to start working as civil servants. To illustrate, graduates from elite public policy schools are increasingly becoming consultants and bankers instead of civil servants (Piereson and Schaefer Riley 2013). Additionally, negative public sector worker stereotypes may affect work performance and employability of public sector workers. Negative stereotyping can reduce performance on various cognitive and social tasks across domains (Inzlicht and Schmader 2011). For example, when women are reminded that performance in negotiations is predicted by stereotypically male characteristics such as assertiveness and rationality, they set lower goals for negotiations and perform worse (Kray, Thompson, and Galinsky 2001). In public management, cross-sectional work shows that negative stereotyping is related to lower confidence, motivation, and morale of public sector workers (Chen and Bozeman 2014).

Negative stereotypes can also have detrimental effects on a societal level, as they can erode perceived legitimacy of and trust in the public sector (Bouckaert and van de Walle 2003). Consequently, citizens may be less willing to participate or cooperate with policy activities (Lee and Lauer Schachter 2019; Uslaner and Brown 2005). This, too, could ultimately lead to poorer performance of the public sector, further reinforcing negative stereotypes. In this way, continuous and sustained negative stereotypes can lead to a self-fulfilling prophecy.

3. A socio-cultural perspective on public sector worker stereotypes

As noted, stereotype research shows that stereotypes can differ strongly across countries and cultures (Fiske 2017). Yet, most of the work on public sector worker stereotypes focuses on the U.S. (e.g. Caillier 2018; Goodsell 2004; Marvel 2016; Willems 2020), and the few studies that look at other countries take a single-country approach (De Boer 2020; Gilad, Ben-Nun Bloom, and Assouline 2018). Over-reliance on Americentric and single-country research overlooks the role of culture and administrative traditions in shaping stereotypes.

A socio-cultural perspective assumes that stereotypes can exist ‘outside’ of human cognition, as a cultural phenomenon or social norm (Augoustinos and Walker 1998, 222). Public sector workers stereotypes are often openly discussed and laughed about (Van de Walle 2004). Contrary to gender, ethnicity, or sexuality stereotypes, they are perceived as innocuous. A review of top-10 box office grossing movies from 2000 to 2009 revealed that 91% of movies featured at least one government worker character depicted negatively (Pautz and Warnement 2013; see also Lichter, Lichter, and Amundson 2000). This suggests that, at least in Western countries, stereotyping public sector workers may act as a social norm (Van de Walle 2004).

Yet social norms can vary greatly across contexts. Across North America, the Netherlands, and South Korea, different administrative traditions influence citizens’ expectations of public sector workers (Neo, Grimmelikhuijsen, and Tummers 2022). For example, the philosophy of governance in East Asian countries, such as South Korea, Singapore, and Japan, is influenced by Confucianism (Van der Wal, van den Berg, and Shamsul Haque 2021). Governing rules are prescribed in moral codes rather than formal law and regulations (Lijing and Rutgers 2017). ‘Good government’ relies on ethical persons to maintain a ‘natural order’ – a harmonious social hierarchy – and public sector workers are seen as one of the most elite groups in society (Cho and Lee 2001). This may contribute to different perceptions of public sector workers than in countries with other administrative traditions. For example, administrations rooted in Weberian traditions, such as the Netherlands, view the role of public sector workers as neutral, impersonal, and rational executioners of political rulings based on legality (Peters 2021) – whereas those rooted in Anglo-American traditions see public sector workers as societal managers whose first and foremost goals are to be efficient and effective (Biesbroek et al. 2018; Peters 2021). These different perceptions of the role of public sector workers could result in different stereotypes of public sector workers.

With the scant literature that exists, we cannot confidently predict potential differences in public sector worker stereotypes across countries, nor the country-level factors that would contribute to stereotypes. Yet, it is plausible that factors such as cultural context and administrative traditions play a role in public sector worker stereotypes (Marriott, Skingle, and Tyers 1975, Peters 2021). This lack of literature and the serious consequences of stereotyping highlight the need for reliable, systematic study of the universality of public sector worker stereotypes.

4. A multi-country, inductive approach

In this study, we take an inductive, cross-country approach to studying stereotypes, by basing ourselves on participants’ own input rather than asking about specific characteristics such as skill or performance. We study stereotypes across four countries. Choosing countries involves trade-offs. We included Canada, the Netherlands, and South Korea – a selection which was, in part, based on the availability of research funds and collaboration partners, but with which we also aimed to create a diverse set of country contexts – in terms of continents, cultures (Hofstede, Hofstede, and Michael 2010), and administrative traditions (Peters 2021). We further included the U.S. as a benchmark, to validate previous findings on stereotypes of public sector workers based mostly in the U.S.

In addition to culture, location, and administrative tradition, the countries differ in the level of confidence that their citizens have in the civil service and in government, as measured in the World Values Survey (Wave 7; Haerpfer et al. 2022). South Korea scores highest in confidence, with a little over half of the population stating that they have quite a lot or a great deal of confidence (51% in government, 56% in civil services; data from 2018), followed by Canada (46% in government, 56% in civil services; data from 2020). The U.S. score markedly lower (33% in government, 41% in civil services; data from 2017), as do the Netherlands (38% in government, 34% in civil services; data from 2022). Stereotypes are often reflective of peoples’ general attitudes (Blair 2002; Devine 1989; Greenwald and Banaji 1995) and research suggests that confidence in government is closely tied to stereotypes (Lerman 2019). As such, peoples’ general confidence in civil service may provide an indication of potential differences in stereotypes across different countries.

We do not claim that the country selection is representative of all different cultures and traditions across the world. Yet, the U.S., Canada, South Korea, and the Netherlands vary enough on the dimensions of geography, culture, administrative tradition, and general attitude towards the public sector to allow for meaningful comparisons. As the current body of literature on public sector worker stereotypes and their antecedents is limited, we refrain from hypothesizing about specific country differences. Rather, we take an exploratory approach. As such, this study is a first step in assessing the influence of country context, enabling us to find commonalities and differences across countries using a common method.

Just as country selection involves trade-offs, so does choosing a methodological approach. While qualitative inductive research allows for exploration to a degree that quantitative research cannot offer, a criticism of this method is that findings cannot be extended to wider populations with the same degree of certainty as quantitative research (Atieno 2009). However, as Atieno (2009) argues, The line between qualitative and quantitative is less distinct. All qualitative data can be quantitatively coded in an almost infinite variety of ways. This does not detract from the qualitative information. Recognizing the similarities between qualitative and quantitative information opens new possibilities for interpretation that might otherwise go unutilized. (p. 5). To benefit from the qualitative inductive approach while also circumventing its pitfalls, we substantiate qualitative findings with quantitative information using a two-step approach. The section below describes the approach in further detail.

5. Empirical studies

To identify stereotypes, we use a method proposed by Katz and Braly in their now-classic study from 1933. Their two-step approach combines qualitative induction in the first step with quantitative methods in the second. This method is still being used in recent research (Madon et al. 2001; Schneider and Bos 2014). The first step of this method inductively creates a list of potential stereotypes; the second step quantitatively assesses which of these stereotypes are most common. In the first step, participants are asked to list traits and characteristics they find typical of a specific social group of interest – in this case, public sector workers. These answers are coded, and the most frequently mentioned traits are used to compile a ‘master list’. In the second step, this master list is used to identify the most common stereotypes among a new and larger sample, by asking participants to select from this list the traits they find most typical of that social group.

More recent studies often forego the first step of the method and use the list of traits generated in the original study by Katz and Braly (1933) to study racial stereotypes. However, researchers have highlighted that the list may be outdated, and that it should not be assumed to be applicable to all social groups (Devine and Elliot 1995; Gilbert 1951; Madon et al. 2001). To fit the public sector context and to avoid these limitations, we want to generate a master list catered to public sector workers specifically. Therefore, we performed both step 1 and step 2 as two separate studies in the four countries under study.

Two further additions were made to the original Katz and Braly method: First, the cross-cultural nature of this study calls for sensitivity to differences in normative interpretation and meaning. Certain words have different meanings in different cultures and could be interpreted differently across countries (Lee 2012). We therefore added a question in Step 2, asking participants to rate how desirable they found the traits that they selected as most typical of public sector workers. This allows us to study (a) which public sector stereotypes participants mention across countries (the content of stereotypes) and (b) to what extent these stereotypes are seen as positive or negative across countries (the valence of the stereotypes). Second, stereotypes are now a more sensitive topic than in 1933. To ensure that participants felt comfortable and secure enough to give their honest opinions, we emphasized participant anonymity in the surveys, underscoring that participants’ answers would not be judged for accuracy and that it was acceptable to generalize.

Figure 1 summarizes the procedures of the 2-step approach. In Study 1, we asked participants in the four countries to list the traits they associated with public sector workers (n = 918). Findings were used to construct a master list of traits commonly associated with public sector workers. In Study 2 (n = 3,042), we presented participants with the master list, from which they select up to five traits they found most typical of public sector workers (i.e. stereotype contents), and asked them to rate each selected trait in terms of desirability (i.e. whether they found these traits to be desirable for public sector workers; stereotype valence). To improve validity and generalizability of our study, we mirrored our samples to the population margins of all four countries in terms of age, gender, and level of education.

Figure 1. Overview of the procedures of the 2-step approach.

The project was pre-registered via the Open Science Framework (link url: https://osf.io/uxk76). Since our paper takes an exploratory approach, we did not pre-register hypotheses, but we included our research questions, methods, analysis plan, and materials. Ethical approval was granted by the ethical committee of the Faculty of Law, Economics, and Governance of Utrecht University.

We compiled the surveys in English, and translated these into Dutch, Korean, and French (for French-speaking Canada) using the TRAPD framework for translation (Harkness, van de Vijver, and Mohler 2003). Bilingual translators for each of the languages produced a first draft of the translations, which was then discussed and refined with members of the research team. These translated surveys were tested using cognitive interviewing (Willis 2008), leading to a few final edits. The final surveys used for Study 1 and 2 are shown in the online Supplementary Materials, Appendices A and B respectively.

Participants were recruited through panel platform Lucid – a survey firm based in New Orleans, Louisiana. Lucid employs numerous national and regional survey panel providers, and recruits survey respondents via these different panels. Samples recruited via the Lucid platform have been shown to be suitable for social science inquiries (Alexander and McClellan 2019). For a fuller characterization of the value of samples acquired through Lucid, we suggest Hisler and Twenge (2021, 2).

Both surveys took approximately 15 minutes to complete. All participants provided informed consent, and were compensated by the panel providers through various means, including financial compensation and online game spending points, depending on the specific panel through which they partook. Compensation for study participation was dependent on completing the questionnaire. We aimed to obtain representative samples of the population in terms of age, gender, and education for each of the four countries. Therefore, we used quotas to match the population margins on these variables, based on population data from OECD (David and Amey 2020). Quotas refer to a predetermined portion of the sample that needs to fill certain criteria, in this case: age, gender, and education levels. To fill these quotas, participants first filled in a profiling questionnaire covering basic socio-demographic information.

It should be noted here that in the presentation of our data and results, we deviate from our pre-registration, as we initially intended to compare the stereotypes of the general category of public sector workers to four other occupational groups: police officers, tax officials, judges, and private sector workers. In explaining the study procedures for both studies below, we therefore refer to randomizing participants to specific occupational groups. However, the data proved to be too rich for our aim of investigating public sector worker stereotypes. Therefore, in the presentation of results of Study 2, we focus only on the findings regarding public sector worker stereotypes. Data for the other occupational groups can be found in the online Supplementary Materials, Appendix C.

6. Study 1

The purpose of Study 1 was to construct a master list of public sector stereotypes to be used in Study 2. The goal was to create a citizen-driven, inductive list of traits most associated with public sector workers by asking participants to list the traits they find typical of public sector workers (Katz and Braly 1933; Schneider and Bos 2014).

6.1. Sample

A total of 1,217 people participated in Study 1. Of these, 313 participants were excluded based on the quality of their answers (133 in the U.S.; 57 in Canada; 97 in the Netherlands, and 26 in South Korea): These participants provided gibberish or random answers (e.g. ‘tyyyyh’ or ‘brara’); illogical or inconsistent responding (e.g. ‘I like birds’ or ‘looking good bruh’), or repetitive responses (e.g. ‘idk’ to all questions; a list of exclusions can be found at https://osf.io/qsr7x/?view_only=83ffc7368c714c5c8d6b5ec3c1412a81). This led to a final sample of 918 participants: 215 in the U.S. (mean age = 49.92, SD age = 18.29; 119 females), 282 in Canada (mean age = 46.60, SD age = 18.21; 147 females), 216 in the Netherlands (mean age = 45.17, SD age = 16.51; 98 females), and 205 in South Korea (mean age = 40.87, SD age = 13.05; 91 females).

Table 1 shows sample descriptives, including how our sample compares to population margins in terms of age, sex, and level of education.

Table 1. Sample descriptives of study 1.

		U.S. (N=215)	Canada (N=282)	Netherlands (N=216)	South Korea (N=205)
% of sample (% of population, OECD 2020)
Sex	Female	55.35 (50.75)	52.13 (50.31)	45.37 (50.37)	44.39 (49.89)
	Male	44.65 (49.25)	47.52 (49.69)	54.63 (49.63)	54.15 (50.11)
	Prefer not to say	0.00	0.35	0.00	1.46
Age	16–24	12.56 (16.21)	16.67 (14.54)	15.74 (14.68)	15.61 (12.81)
	25–34	15.81 (17.03)	17.73 (16.56)	18.06 (14.97)	19.02 (15.30)
	35–44	12.56 (15.58)	14.18 (15.84)	13.89 (14.31)	22.44 (17.09)
	45–54	15.35 (15.62)	15.25 (15.52)	20.83 (17.68)	27.80 (19.23)
	55+	43.72 (35.56)	36.17 (37.54)	31.48 (38.36)	15.12 (35.57)
Level of Education	Low (No formal education, grade school)	7.44 (9.20)	7.80 (8.40)	19.44 (21.00)	2.44 (11.80)
	Mid (High school, vocational school, no college degree)	58.14 (43.40)	32.98 (33.70)	41.20 (40.70)	45.85 (39.20)
	High (With college degree)	34.42 (47.40)	59.22 (57.90)	38.89 (38.30)	50.73 (49.00)

Note. Sampling quotas were used to ensure the sample was comparable to population margins, in terms of age, sex, and level of education. Numbers between parentheses indicate population margins.

6.2. Survey procedure and analyses

Participants first gave their informed consent, followed by questions about their age, sex, and education level. Next, we asked participants: Please list as many specific characteristics or traits as you think are typical of the following occupational group (max. 5): public sector workers. Here, participants saw five form fields to fill in with a maximum of five traits. The survey ended with some questions about participants’ own occupational history.

As per our pre-registered analysis plan, we compiled the master list as follows. First, all French, Dutch, and Korean answers were translated to English. We assessed sentences given as answers and where possible, replaced them with single words. For instance, when a participant wrote that a characteristic of public sector workers is that they have a lot of knowledge, we replaced this with knowledgeable. Next, we standardized entries by removing capitals and spaces. We then ran a frequency analysis and selected the top ten most frequently mentioned traits per country. In selecting the top ten words, we discarded words that do not reflect traits or attributes (for example, ‘lawyer’ and ‘government’). We did include the characteristics of ‘well paid’, ‘going home on time’, and ‘job security’ – although these may not strictly be seen as personal traits, they do describe characteristics of the work and its workers.

When aggregating the top ten of the four countries, many traits were repeated because of overlap between the countries. To meet the targeted number of traits as specified in our pre-registered plan, we therefore selected the top fifteen for a more comprehensive master list. Lastly, we grouped together close synonyms like smart and intelligent (summarized as intelligent), brave and courageous (courageous), and compassionate and empathetic (empathetic), leading to a total of thirty-six traits.

6.3. Results

Table 2 shows the master list of thirty-six traits. Participants from all countries generated a mixture of positive traits – such as hardworking, intelligent, impartial, and honest – and negative traits – such as boring, corrupt, inflexible, and lazy. Some of the traits are consistent with the stereotypical image assumed by the literature and popular media, for example lazy, corrupt, and inflexible. However, some traits that frequently came up are unexpected and contrary to existing literature and popular media depictions, such as caring, empathetic, and hardworking.

Table 2. Master list of thirty-six traits.

Arrogant	Go home on time	Knowledgeable
Authoritative	Friendly	Lazy
Boring	Good	Loyal
Calm	Hardworking	Patient
Caring	Helpful	Responsible
Conservative	Honest	Serious
Corrupt	Impartial	Serving
Courageous	Independent	Stable
Difficult	Inflexible	Strict
Educated	Integrity	Strong
Empathetic	Intelligent	Trustworthy
Fair	Job security	Well paid

7. Study 2

In Study 2, we presented participants with the master list compiled in Study 1. The goal was to identify which traits from the list are most associated with public sector workers, among a larger sample of citizens from each country (i.e. stereotype content). Additionally, we wanted to know how desirable participants found these traits for public sector workers, (i.e. how positive or negative they found the traits, stereotype valence).

7.1. Sample

A total of 4,588 participants were recruited across the four countries. As specified in our pre-registration, we used three attention checks to ensure data quality. Following survey platform policy and institutional review board requirements, participants had to successfully pass two out of three checks to be included in the study. 446 participants were excluded for failing to meet these criteria (U.S. = 139, Canada = 88, the Netherlands = 130, and South Korea = 89). Like Study 1, we excluded 89 participants for bad data quality (i.e. gibberish or nonsensical answers; 42 in the U.S.; 47 in South Korea). A list of exclusions can be found at https://osf.io/uxk76/?view_only=f2fd057c79bf4870b3077b6df2c26961.

Participants were further randomized to rate three (out of five) occupations. The total sample rating public sector workers was 3,042 (U.S.: n = 610; Canada: n = 632; the Netherlands: n = 1,176; South Korea: n = 633). We achieved broadly representative samples in terms of age, gender, and education, with a maximum difference of 8% between the observed and targeted proportions in all countries. An overview of sample descriptives and population margins is shown in Table 3.

Table 3. Sample descriptives of study 2.

		U.S. (N=610)	Canada (N=632)	Netherlands (N=1176)	South Korea (N=633)
% of sample (% of population, OECD 2020)
Sex	Female	50.98 (50.75)	49.44 (50.31)	50.17 (50.37)	47.24 (49.89)
	Male	48.69 (49.25)	50.40 (49.69)	49.15 (49.63)	50.71 (50.11)
	Prefer not to say	0.33	0.16	0.68	2.05
Age	16–24	10.82 (16.21)	11.88 (14.54)	15.48 (14.68)	9.48 (12.81)
	25–34	15.90 (17.03)	17.34 (16.56)	17.09 (14.97)	16.27 (15.30)
	35–44	16.89 (15.58)	16.37 (15.84)	16.41 (14.31)	16.90 (17.09)
	45–54	17.70 (15.62)	15.73 (15.52)	12.5 (17.68)	19.27 (19.23)
	55+	38.69 (35.56)	38.68 (37.54)	38.52 (38.36)	38.07 (35.57)
Level of Education	Low (No formal education, grade school)	7.70 (9.20)	5.78 (8.40)	23.89 (21.00)	4.58 (11.80)
	Mid (High school, vocational school, no college degree)	41.48 (43.40)	33.39 (33.70)	36.56 (40.70)	45.97 (39.20)
	High (With college degree)	50.82 (47.40)	60.83 (57.90)	39.54 (38.30)	48.03 (49.00)

Note. Sampling quotas were used to ensure the sample was comparable to population margins, in terms of age, sex, and level of education. Numbers between parentheses indicate population margins.

7.2. Survey procedure and analyses

After giving their informed consent and answering questions about their age, sex, and level of education, participants answered questions based on the master list of traits about three occupational groups (as noted above, we diverged from the pre-registered analysis plan by focusing on one occupation, namely public sector workers, as the main group of interest). Throughout the questionnaire, they were presented with three attention check questions. They were also asked several questions for other studies that were included in this data collection – these studies were pre-registered separately (see https://osf.io/mv9fp/?view_only=a834ee0cce6045299d459036f0a29e64 and https://osf.io/fqn9a/?view_only=a47c628afc624c539d6986f2e779bea6). The survey ended with some additional socio-demographic questions.

To identify stereotype contents, participants were presented with the master list of thirty-six traits and read the following instructions: Read through the following list of words and select those which you find typical of public sector workers. Choose as many words as you think are necessary to characterize this group adequately. In case participants selected more than five traits, we then asked them to select their top five: Now from the words which you have chosen above, select from the drop-down boxes below the top five words which you find the most typical of public sector workers.

To assess stereotype valence, we then asked participants to rate the traits they selected as most typical in terms of desirability: To what extent do you find these traits desirable for public sector workers? Participants answered on a 5-point scale from 1 (very undesirable) to 5 (very desirable; Lee 2012).

7.2.1. Stereotype profiles

We identified stereotypes by examining the traits that people associated most with public sector workers, leading to a so-called stereotype profile. This profile consists of (1) the stereotype content based on the traits that were most frequently selected as one of the top five most typical, and (2) stereotype valence, based on mean desirability scores for those traits.

We compiled stereotypes profiles for each country, based on the top ten most frequently selected traits within each country, as well as one cross-country profile for the four countries combined. We selected ten traits because in comparing between different occupations, post-hoc pairwise odds-ratio analyses of the trait valence indicated that the largest differences between occupations lie in the first ten traits (odds-ratios: ≥ 1.5; see online Supplementary Material, Appendix D). We then calculated stereotype valence as the mean desirability score for each of the top ten traits within a country. An overall, weighted mean valence score for the full profile was also added, such that the mean valence scores for each of the top ten traits were weighted based on the frequency with which they were selected. A full report of frequencies and valence scores for all thirty-six traits, across all countries and per country, can be found in Appendix A (Tables S1 - S5).

7.3 Results

7.3.1. Overall stereotype profile

Figure 2 shows an overall stereotype profile, combining the data from all four countries (n = 3,042). The top three most frequently associated traits are serving, going home on time, and job security, and these are widely endorsed: More than 20% of all participants associated at least one of these three traits with public sector workers. These traits were chosen consistently across age, sex, and education levels (see online Supplementary Materials, Appendix E).

Figure 2. Stereotype profile of public sector workers across countries.

Note. The words show the top ten most frequently selected traits as typical of public sector workers. The grey bars in the left panel show the percentage of participants that selected the trait as one of their top five most typical of public sector workers. The right panel shows the mean valence scores for each trait.

Most of the traits in the overall stereotype profile have a positive mean valence score, with going home on time (M = 2.91, SD = 1.13) and inflexible (M = 1.91, SD = 1.09) being the only negative traits. The weighted mean valence score for the full profile is 3.71 (SD = 0.85), indicating that overall, public sector workers are regarded positively by participants. Taken together, these findings do not support the stereotypically negative image of public sector workers purported by some previous literature and popular media.¹

7.3.2. Cross-country comparisons

Figure 3 shows the stereotype profiles of public sector workers per country. Again, the same three traits are universally shared in the stereotype profiles of the four countries: serving (U.S.: 3^rd most frequently mentioned, selected by 23.8%; Canada: 3^rd, 21.2%; Netherlands: 5^th, 22.5% South Korea: 4^th, 18.3%), going home on time (U.S.: 7^th, 15.4%; Canada: 1^st, 23.1%; Netherlands: 1^st, 40.3%; South Korea: 5^th, 15.8%), and job security (U.S.: 5^th, 17.7%; Canada: 6^th, 17.8%; Netherlands: 2^nd, 34.5%; South Korea: 1^st, 53.4%).

Figure 3. Stereotype profiles of public sector workers in the United States, Canada, the Netherlands, and South Korea.

Note. The grey bars in the left panel show the percentage of participants that selected the trait as one of their top five most typical of public sector workers. The right panel shows the mean valence scores for each trait. Error bars represent 95% confidence intervals of the mean valence scores.

Despite this overlap in traits, Figure 3 also shows some striking differences. First, the U.S. and Canada have highly similar profiles. They share eight out of the ten traits in their stereotype profiles, whereas they share no other traits with the Dutch and South Korean profiles beyond serving, going home on time and job security. Additionally, there are no negative traits in the profiles of the U.S. and Canada (with no valence scores below the neutral score of 3). Instead, participants from the U.S. and Canada selected mostly traits, such as helpful, responsible, hardworking, knowledgeable, and educated. In contrast, Dutch and South Korean participants associated more negative traits with public sector workers, with half of the valence scores in these countries falling below a neutral score of three. While these profiles do include positive traits, such as serving, stable, and responsible, they also include clearly negative traits such as boring and lazy (the Netherlands), corrupt (South Korea), authoritative and inflexible (both).

Furthermore, we see that participants valued some characteristics differently. For example, although going home on time appears in all profiles, it was rated differently in terms of valence: In the Netherlands, going home on time was rated more negatively (M = 2.60), while it was seen as neutral in the U.S. and Canada (U.S.: M = 3.10, Canada: M = 3.27) and quite positive in South-Korea (M = 3.69).

To assess whether the four countries differed statistically in terms of the valence of their stereotypes, we conducted a one-way, between-subjects ANOVA to compare participants’ mean valence scores of their selected traits. Results revealed significant differences between the four countries, F(3, 3038) = 33.36, p < 2e⁻¹⁶, η² = 0.03, 95% CI = [3.50, 3.57]. Table 4 shows the results of post-hoc pairwise t-tests with Bonferroni corrections between the four countries. Results show no significant differences between mean valence scores in the U.S. and Canada (U.S. M = 3.80, SD = 1.72, Canada M = 3.79, SD = 1.74, t(1231) = −0.08, p =1.0, Cohen’s d = 0.01, 95% CI = [−0.14, 0.13]). In contrast, comparing the United States to the Netherlands (Netherlands M = 3.31, SD = 2.07, t(1784) = −8.06, p = 7.99e⁻¹⁵, d = 0.40, 95% CI = [−0.60, −0.37]) and South Korea (South Korea M = 3.46, SD = 1.75, t(1241) = −4.73, p = 1.48e⁻⁵, d = 0.27, 95% CI = [−0.47, −0.20]), differences between means are significant. This indicates that the stereotypes are significantly more negative in the Netherlands and South Korea than in the U.S. The same applies to Canada: comparing Canada to the Netherlands: t(1797) = 8.26, p = 1.69e⁻¹⁵, d = 0.41, 95% CI = [0.37, 0.59], and Canada versus South Korea: t(1254) = 4.86, p = 7.78e⁻⁶, d = 0.27, 95% CI = [0.20, 0.46].

Table 4. Post-hoc pairwise country comparisons.

		Mean Difference	Std. Error	Significance		95% Confidence Interval for Difference
				p-value	Adjusted p-value	Lower Bound	Upper Bound
United States	Canada	0.00	0.07	0.93	1.00	−0.14	0.13
	The Netherlands	0.49	0.06	1.33e^−15	7.99e^−15*	−0.60	−0.37
	South Korea	0.33	0.07	2.47e^−6	1.48e^−5*	−0.47	−0.20
Canada	The Netherlands	0.48	0.06	2.82e^−16	1.69e^−15*	0.37	0.59
	South Korea	0.33	0.07	1.30e^−6	7.78e^−6*	0.20	0.46
The Netherlands	South Korea	0.15	0.06	0.01	0.06	−0.27	−0.04

Note. Asterisks indicate a significant difference, after Bonferroni p-value adjustments, at α = .05.

*= p < .05.

Finally, although the valence scores are similar in the Netherlands (M = 3.33, SD = 1.08) and South Korea (M = 3.41, SD = 0.77; t(1807) = −2.57, p = 0.06, d = 0.13, 95% CI = [−0.27, −0.04]), the Dutch and South Korean stereotype profiles differ in terms of contents, with participants associating different traits with public sector workers. For example, while Dutch participants associated public sector workers with being well paid, boring, and lazy, South Korean participants did not – conversely, South Koreans associated them with, amongst others, being corrupt, arrogant, and stable, while these did not appear in the top ten of the Netherlands.

8. Discussion

Using two pre-registered, large scale citizen surveys with representative samples (total n = 3,960), we mapped stereotypes of public sector workers in the U.S., Canada, the Netherlands, and South Korea. We find three stereotypes that are universal across these countries: having job security, going home on time, and serving. These are largely universal across age, gender, and education levels. However, we find stark differences when comparing the stereotypes across countries. Public sector worker stereotypes in the U.S. and Canada are similar and remarkably positive: In both countries, there are no negative traits in the top ten stereotype profile, and the most frequently selected traits beyond the universal traits were hardworking, responsible, and helpful. In comparison, stereotypes in South Korea and the Netherlands are more negative, with associated traits like inflexible (in both countries), boring (in the Netherlands), and corrupt (in South Korea). This indicates that, although ideas about public sector workers may be universal to some extent, country differences should not be overlooked.

Our findings should be considered in light of some limitations. First, the translation of our survey may have suffered from a trade-off between literal accuracy and ease in comprehension of meaning. Even though we worked closely with bilingual translators to achieve a thorough contextual understanding of ambiguous words, some words may bear different connotations in different languages. For example, the Dutch word we chose to use for public sector worker (‘ambtenaar’) is most used, but it also carries a negative connotation, as the stereotype and the word itself seem to have become linguistically entwined.

Second, we assessed stereotype valence by asking participants how desirable they found traits for public sector workers, but this may be interpreted in different ways. While some might answer based on what is desirable from the viewpoint of the public, others may interpret the question as what is desirable from a public sector worker’s point of view. For instance, going home on time may be desirable for workers, but not necessarily for their clients. Future studies could further explore perceived stereotype valence, and whether this depends on who the stereotype pertains to.

Lastly, we are limited in our interpretation of whether stereotypes are mentioned with reference to public sector jobs or public sector workers. For instance, stereotypes such as job security, well paid, and going home on time can characterize both the worker (i.e. motivated by pay, preference for job security) or the occupation (i.e. well paid job, having job security). Stereotypes are multidimensional and represent myriad beliefs about the characteristics, traits, and behaviours of members of certain groups. Without in-depth qualitative data, it is difficult to disentangle what stereotypes mean. For example, while going home on time can be taken to mean that public sector workers are lazy and do not work more than required, this interpretation would be in direct opposition to the trait hardworking – which is also associated with public sector workers. The complexity in meaning of stereotypes becomes more evident when comparing across countries when they regard the same trait of going home on time differently – positive (South Korea), neutral (U.S. and Canada) and negative (the Netherlands). Our current methods do not allow us to make accurate inferences about what the stereotypes mean to respondents. Future studies should refine their study designs, using alternative qualitative research methods to disentangle the stereotypes associated with the job versus its workers, and the meanings behind the stereotypes.

Despite these limitations, we believe that our approach allowed for the identification of stereotypes that may have been overlooked by using deductive survey methodologies.

8.1. Theoretical implications

Our findings have important implications for the understanding of public perceptions of public sector workers. A first contribution of our findings is that they help us understand country differences by giving insight into plausible explanations and antecedents of stereotypes. More importantly, they show where existing theories fall short in explaining stereotypes.

While researchers have suggested a strong relationship between stereotypes and confidence in government and public services (Lerman 2019), the patterns in our findings suggest otherwise. Data from the World Values Survey (Wave 7; Haerpfer et al. 2022) shows that in terms of confidence in the civil service, South Korea scores the highest, followed by Canada, the U.S., and the Netherlands. In contrast, our results show overwhelmingly positive stereotypes for the U.S. and Canada and more negative stereotypes for South Korea and the Netherlands.

Administrative tradition and prevailing sentiments towards the public sector in each country are also inadequate in explaining our findings: The United States’ Jeffersonian vision of being fearful of technocracy and being suspicious of big government (Hubbell 1991) does not match the positive stereotypes that we find. Administrative culture in the Netherlands has been characterized by New Public Management reforms since the 1980s, which have led to criticism of slow bureaucracy (Hood 1991; Pollitt and Bouckaert 2017). While this may explain the stereotypes found in the Netherlands such as lazy and inflexible, similar NPM reforms were taking place in the U.S., Canada, and South Korea (Brandsen and Kim 2010). Yet, stereotypes there are greatly different from those in the Netherlands – both in terms of content and valence. Confucian ideals in South Korea also fall short in explaining the predominantly negative stereotypes found there. According to Confucian ideas, public sector workers are seen as the most elite group in society (Cho and Lee 2001), yet the stereotypes we find in South Korea do not reflect this perspective and are much more negative than those in North America.

Another factor that may have affected perceptions of public sector workers in our study is the types of occupations that people associate with the term ‘public sector workers’. De Boer (2020) used the Stereotype Content Model to study public sector worker stereotypes and found that workers in service-oriented jobs are seen as warmer and more competent than those in regulation-oriented jobs. Willems (2020) also found that public sector workers in the U.S., when seen as similar to bankers, are regarded less positively than firefighters, police, teachers and nurses. Since we asked about public sector workers in general, respondents in different countries may have thought of different types of public sector occupations. This is partly substantiated by our presurvey data. In the presurvey, we asked participants which occupations came to mind when listing traits associated with public sector workers. A post-hoc coding of the occupations revealed that participants in the U.S. and Canada thought of interpersonal, service-oriented occupations such as nurses, social workers, and firefighters more readily than Dutch or South Korean participants. Dutch participants mentioned more organizational or technical occupations such as tax administrators and municipal workers.² Therefore, thinking of more interpersonal service-oriented occupations may have led to more positive stereotypes among U.S. and Canadian respondents in comparison to Dutch and South Korean respondents. However, participants also mentioned many occupations that do not fall readily into the service- versus regulation-oriented dichotomy, indicating that this dichotomy is too simplified to fully explain our results.

A promising alternative explanation for the differences in stereotypes may lie in the interplay between expectations and actual experience of public service. Research suggests that the difference between how public sectors are and should be is an important determinant of citizen satisfaction (Van Ryzin 2015). More specifically, the expectancy disconfirmation model (EDM) proposes that citizen satisfaction can be determined by a process in which actual performance is compared to prior expectations (James 2009; Van Ryzin 2004). One way in which prior expectations are formed is through establishing an ideal. Social psychologists have found that this is done by constructing a mental picture of what the ideal should look like, and comparing the qualities of a given target to the qualities believed to be embodied by the ideal benchmark (Miller, Wattenberg, and Malanchuk 1986). The closer the perceptions match the expectations, the more satisfied a person should be with the target (Bonito 2004; Hall 2012). In the context of public services, citizens could use such a strategy in the evaluation of public sector workers by comparing them to an ideal benchmark. In short, a larger discrepancy between what citizens typically observe in depictions of, and encounters with public sector workers versus their expectations of how they should be, could affect stereotypical perceptions.

Additionally, research shows that the ideal traits and values that citizens believe public employees should embody differ across cultures and are influenced by administrative traditions (Neo, Grimmelikhuijsen, and Tummers 2022). For example, Dutch citizens value responsiveness, serviceability, effectiveness, and efficiency while South Korean citizens valued honesty, incorruptibility, and accountability. Indeed, the values that one holds depend on the values that are prescribed and reinforced by the traditions of one’s social environment (Leung and Cohen 2011). As such, the influence of the administrative tradition of a country may be such that it affects the traits considered important for an ideal public sector worker, which in turn influences the stereotypes that are formed. Stereotypes such as lazy and inflexible may reflect public discontent in the Netherlands with regards to these ideal normative values; likewise, stereotypes of corruption may result from the discrepancy between ideal values and what is observed in recent high-profile political scandals in South Korea. While we believe this to be a promising hypothesis, it requires more in-depth research, and future research is needed to assess the extent to which the explanation holds.

Another important contribution of our work lies in recognizing the different components of a stereotype: content and valence. These two components can be differentially affected by the sociocultural context. For example, serving, going home on time, and job security are traits endorsed by participants across all four countries and demographic groups. Although these stereotypes are shared across countries, they are not perceived equally in terms of how positive they are. Therefore, focusing only on stereotype content in research may lead to blind spots and researcher bias, with researchers being bound by their own cultural context in assuming the valence associated with a specific stereotype. These findings allude to the importance of context sensitivity in the theoretical understanding of public sector worker stereotypes.

Our results also illustrate challenges in the framing of language when studying public opinion regarding public sector workers. Our findings imply that word choice in survey tools may influence opinions and attitudes towards public sector workers, such as in the case of the use of ‘ambtenaren’ in the Dutch survey. A word with similar connotation in English is ‘bureaucrat’. Given its negative connotation, a survey asking participants to select traits associated with ‘bureaucrats’ may yield different sets of stereotypes in comparison to one asking for traits associated with ‘public sector workers’. Research has shown that word choice in surveys can lead to false conclusions as they bias respondents (Ashford, Brown, and Curtis 2018). This has important implications especially for research seeking to understand citizen attitudes and satisfaction towards public service providers.

Lastly, our data suggest that public perceptions of government institutions and services are different from perceptions of the people that provide these services. For example, while confidence in government in the United States is markedly low (WVS Wave 7), our data show that stereotypes about public sector workers are definitively positive there. Van de Walle (2004) argued that citizens’ attitude towards the administration is dependent on the target of evaluation. For example, people have different attitudes towards local municipal administration than towards the general administration. Ideas that citizens have of the public sector are, apparently, not homogenous across the different components that make up the public sector. People could therefore have contrasting perceptions between government and its public sector workers. This calls for caution in research practices that assume attitudes towards governments can be generalizable to the different components of the public sector.

At the same time, our findings show that it is also meaningful to study perceptions of the general category of ‘public sector workers’, even if specific jobs within that sector differ greatly from each other. In an earlier study of public sector worker stereotypes, Willems (2020) found that respondents were able to associate more traits with specific occupations (such as nurses) than with public sector workers in general. They concluded that no single epithet of the public sector worker exists. Yet, our additional data (see Appendix C in the online Supplementary Materials) shows that the stereotypes of public sector workers in general were no more dispersed than those of the specific public sector occupations we asked about, namely police, judges, and tax officials. In the four countries, the top ten stereotypes of public sector workers were shared by 12% to as much as 30% of participants. This suggests that the public sector worker, as a general cognitive entity, is just as clear and prone to stereotypes in citizens’ minds as, for example, police officers, judges, and tax officials.

In sum, the findings call for future research into the factors that determine public sector worker stereotypes. The unexpected findings in our study, which run contrary to common negative depictions of public sector workers, warrant further research. The discrepancy between cultural stereotypes and individual ideas means that scholars should not assume cultural stereotypes – such as portrayals by newspapers, movies, or politicians – to be representative of individual citizen beliefs. Doing so would disregard the positive stereotypes of public sector workers that exist. Taking an inductive, rather than a deductive, approach allows scholars to see what people’s beliefs are, beyond taken-for-granted assumptions about public sector stereotypes. Furthermore, scholars should not assume that stereotypes are homogeneous across the globe. Studying one phenomenon across multiple countries can lead to surprising findings, providing inspiration for follow-up questions and further research.

We hope this study helps to further develop and refine public sector worker stereotyping research. Here, we offer two concrete suggestions. First, future efforts should be targeted at conducting larger scale replications involving more countries. This allows for a more comprehensive understanding of public sector worker stereotypes and serves as a basis for theory building. Specifically, it allows us to identify and systematically study potential antecedents and causes of stereotyping public sector workers, such as administrative culture, corruption levels, or individual-level trust in government. Second, evidence from psychological research has pointed to the damaging consequences of negative stereotyping across a wide span of domains including work performance, motivation, and wellbeing (Kray and Shirako 2012). It is, therefore, imperative to invest efforts in understanding the consequences of stereotyping public sector workers.

8.2. Practical implications

Finally, our findings have important practical implications, too. First, the positive stereotypes we find can be strategically leveraged to attract and retain talent in the public sector. For example, Linos (2018) found that advertisements that emphasize the personal or career benefits of joining the police force are three times more effective than neutral advertisements at getting qualified individuals to apply. More importantly, these messages are particularly effective for people of colour and women and can therefore support policy goals in different countries to increase diversity within the public sector workforce. The stereotype of job security can be strategically used for talent recruitment and employer branding of public organizations; the stereotype of going home on time can be used to signal that careers in the public sector allow for a healthy work-life balance.

Second, positive stereotypes can also be used to improve the work of public sector workers. Research has shown that reminding individuals of positive stereotypes of their social group can be used to improve task performance, especially among individuals who identify strongly with their group. Given that one’s work affiliation can be a powerful and meaningful social category that individuals identify with (Darja and Day 2016), positive stereotypes of public sector workers such as hardworking, responsible, and helpful can be used to boost workers’ confidence and performance, ultimately improving public service quality (Shih et al. 2002). Overall, positive stereotypes can be leveraged to attract, hire, develop, and empower talented individuals within the public sector. This directly supports key policy goals of various governments such as those illustrated in the Biden-Harris President’s Management Agenda (OMB 2021) and can inspire strategies for governments struggling to attract new employees.

9. Ethical statement

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee. Ethical approval for both studies was granted by the ethical committee of the Faculty of Law, Economics, and Governance of Utrecht University, the Netherlands.

Data and materials availability

The data that support the findings of this study are publicly available at https://osf.io/qsr7x/?view_only=83ffc7368c714c5c8d6b5ec3c1412a81.

The codes that for the findings of this study are publicly available at https://osf.io/qsr7x/?view_only=83ffc7368c714c5c8d6b5ec3c1412a81.

Acknowledgement

The authors would like to thank Freyia Han, Itizez Slama and Nina Breedveld for their outstanding assistance in the data collection and analysis process. We would like to thank the anonymous reviewers and the editor who provided us with constructive and insightful comments, significantly improving the article.

Disclosure statement

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

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The study was supported by the NWO Grant 016.VIDI.185.017. This work was also supported by the National Research Foundation of Korea Grant, funded by the Korean Government (NRF-2017S1A3A2067636).

Notes

1. Following our pre-registered procedure, we also asked participants to select attributes they associated with three specific public sector occupations: police officers (n = 2,449), tax officials (n = 2,508), and judges (n = 2,490). The results of the analyses of this data can be found in the online Supplementary Materials, Appendix C.

2. A complete list of codes can be found at: https://osf.io/qsr7x/?view_only=83ffc7368c714c5c8d6b5ec3c1412a81.

Appendix A Full stereotype profiles

Tables S1– S5 below show which traits the participants in Study 2 selected as typical of public sector workers. We show how often they were selected as typical of public sector workers, the corresponding percentage of participants that selected the trait, the mean valence of the trait, and 95% confidence intervals. Table S1 shows the public sector worker stereotypes across countries, Tables S2– S5 show the stereotypes in the United States, Canada, the Netherlands, and South Korea, respectively.

Table S1. Stereotypes for public sector workers in all countries combined (n = 3,042).

	Trait	Selection frequency	% of participants	Mean stereotype valence	Lower 95% CI	Upper 95% CI
1	Job security	963	31.66	3.66	3.59	3.73
2	Go home on time	812	26.69	2.91	2.83	2.99
3	Serving	658	21.63	4.47	4.41	4.54
4	Inflexible	600	19.72	1.91	1.82	2.00
5	Responsible	530	17.42	4.55	4.47	4.63
6	Well paid	529	17.39	3.30	3.21	3.39
7	Helpful	421	13.84	4.56	4.47	4.64
8	Hardworking	419	13.77	4.42	4.33	4.51
9	Knowledgeable	356	11.70	4.42	4.32	4.52
10	Serious	327	10.75	4.13	4.02	4.24
	Weighted mean valence of stereotype profile (top 10 traits only)					3.71 (SD=0.85)
	Unweighted mean valence of stereotype profile (top 10 traits only)					3.83 (SD=0.84)
11	Boring	325	10.68	2.20	2.09	2.32
12	Authoritative	322	10.59	2.50	2.36	2.65
13	Stable	321	10.55	4.19	4.09	4.28
14	Arrogant	320	10.52	1.78	1.64	1.91
15	Friendly	301	9.89	4.39	4.28	4.49
16	Corrupt	295	9.70	1.69	1.55	1.83
17	Lazy	292	9.60	1.61	1.49	1.73
18	Conservative	280	9.20	2.87	2.72	3.02
19	Educated	280	9.20	4.23	4.12	4.35
20	Trustworthy	261	8.58	4.63	4.53	4.73
21	Difficult	223	7.33	2.11	1.94	2.28
22	Intelligent	213	7.00	4.32	4.19	4.45
23	Patient	204	6.71	4.14	4.00	4.28
24	Good	203	6.67	4.12	3.98	4.27
25	Loyal	184	6.05	4.24	4.10	4.39
26	Fair	183	6.02	4.20	4.04	4.36
27	Caring	178	5.85	4.17	4.00	4.35
28	Honest	178	5.85	4.34	4.18	4.49
29	Integrity	178	5.85	4.46	4.31	4.60
30	Calm	163	5.36	4.17	4.01	4.32
31	Strong	136	4.47	3.83	3.63	4.04
32	Strict	135	4.44	2.73	2.53	2.94
33	Impartial	117	3.85	4.16	3.96	4.36
34	Independent	106	3.48	3.85	3.64	4.06
35	Empathetic	103	3.39	4.04	3.80	4.28
36	Courageous	96	3.16	4.07	3.88	4.27
	Weighted mean valence of stereotype profile (all 36 traits)					3.56 (SD=0.95)
	Unweighted mean valence of stereotype profile (all 36 traits)					3.65 (SD=0.95)

Table S2. Stereotypes for public sector workers in the United States (n = 610).

	Trait	Selection frequency	% of participants	Mean stereotype valence	Lower 95% CI	Upper 95% CI
1	Hardworking	187	30.66	4.45	4.31	4.60
2	Responsible	146	23.93	4.48	4.31	4.64
3	Serving	145	23.77	4.46	4.31	4.61
4	Helpful	123	20.16	4.46	4.27	4.65
5	Job security	108	17.70	3.61	3.35	3.87
6	Knowledgeable	106	17.38	4.46	4.28	4.64
7	Go home on time	94	15.41	3.10	2.84	3.36
8	Friendly	90	14.75	4.38	4.18	4.58
9	Educated	81	13.28	4.36	4.13	4.58
10	Trustworthy	73	11.97	4.45	4.22	4.68
	Weighted mean valence of stereotype profile (top 10 traits only)					4.26 (SD=0.45)
	Unweighted mean valence of stereotype profile (top 10 traits only)					4.22 (SD=0.45)
11	Caring	72	11.80	4.28	4.02	4.54
12	Good	72	11.80	4.06	3.76	4.36
13	Serious	70	11.48	3.81	3.52	4.11
14	Stable	62	10.16	4.34	4.10	4.58
15	Honest	61	10.00	4.26	3.97	4.56
16	Well paid	61	10.00	3.48	3.14	3.81
17	Intelligent	58	9.51	4.33	4.05	4.61
18	Strong	58	9.51	4.03	3.73	4.34
19	Lazy	56	9.18	1.77	1.44	2.10
20	Inflexible	55	9.02	2.05	1.74	2.37
21	Loyal	54	8.85	4.07	3.76	4.39
22	Authoritative	53	8.69	3.02	2.65	3.39
23	Arrogant	48	7.87	1.60	1.31	1.90
24	Patient	48	7.87	3.98	3.67	4.29
25	Fair	44	7.21	3.77	3.40	4.14
26	Integrity	44	7.21	4.43	4.12	4.74
27	Difficult	40	6.56	2.05	1.63	2.47
28	Independent	40	6.56	3.83	3.47	4.18
29	Corrupt	39	6.39	1.49	1.17	1.80
30	Conservative	38	6.23	3.76	3.36	4.16
31	Strict	38	6.23	3.03	2.60	3.45
32	Boring	34	5.57	2.15	1.74	2.55
33	Calm	32	5.25	4.25	3.91	4.59
34	Courageous	29	4.75	4.34	4.05	4.64
35	Empathetic	26	4.26	3.85	3.25	4.44
36	Impartial	19	3.11	3.74	3.12	4.35
	Weighted mean valence of stereotype profile (all 36 traits)					3.86 (SD=0.93)
	Unweighted mean valence of stereotype profile (all 36 traits)					3.67 (SD=0.91)

Table S3. Stereotypes for public sector workers in Canada (n = 623).

	Trait	Selection frequency	% of participants	Mean stereotype valence	Lower 95% CI	Upper 95% CI
1	Go home on time	144	23.11	3.27	3.08	3.47
2	Helpful	136	21.83	4.66	4.53	4.79
3	Serving	132	21.19	4.36	4.20	4.52
4	Responsible	124	19.90	4.56	4.39	4.73
5	Hardworking	112	17.98	4.45	4.26	4.63
6	Job security	111	17.82	3.84	3.64	4.04
7	Knowledgeable	102	16.37	4.43	4.22	4.64
8	Stable	96	15.41	4.20	4.00	4.39
9	Well paid	94	15.09	3.49	3.27	3.71
10	Educated	89	14.29	4.28	4.09	4.47
	Weighted mean valence of stereotype profile (top 10 traits only)					4.15 (SD=0.44)
	Unweighted mean valence of stereotype profile (top 10 traits only)					4.15 (SD=0.44)
11	Friendly	82	13.16	4.51	4.33	4.70
12	Patient	71	11.40	4.42	4.20	4.64
13	Inflexible	68	10.91	1.99	1.72	2.25
14	Caring	63	10.11	4.27	3.98	4.56
15	Lazy	61	9.79	1.43	1.22	1.63
16	Serious	59	9.47	4.02	3.76	4.28
17	Good	57	9.15	4.23	3.98	4.48
18	Intelligent	55	8.83	4.27	3.99	4.56
19	Authoritative	46	7.38	2.89	2.52	3.26
20	Calm	46	7.38	4.20	3.90	4.49
21	Trustworthy	46	7.38	4.78	4.63	4.93
22	Integrity	44	7.06	4.45	4.13	4.78
23	Arrogant	42	6.74	1.90	1.52	2.29
24	Loyal	42	6.74	4.07	3.73	4.41
25	Fair	41	6.58	4.00	3.60	4.40
26	Conservative	38	6.10	3.26	2.85	3.67
27	Empathetic	37	5.94	4.27	3.94	4.60
28	Difficult	34	5.46	1.56	1.25	1.87
29	Boring	33	5.30	1.88	1.54	2.22
30	Corrupt	31	4.98	1.58	1.20	1.96
31	Strong	30	4.82	3.90	3.42	4.38
32	Courageous	29	4.65	4.21	3.81	4.60
33	Honest	28	4.49	4.50	4.09	4.91
34	Impartial	24	3.85	3.88	3.32	4.43
35	Independent	23	3.69	3.52	2.99	4.06
36	Strict	20	3.21	2.65	2.27	3.03
	Weighted mean valence of stereotype profile (all 36 traits)					3.85 (SD=1.01)
	Unweighted mean valence of stereotype profile (all 36 traits)					3.67 (SD=0.99)

Table S4. Stereotypes for public sector workers in the Netherlands (n = 1,176).

	Trait	Selection frequency	% of participants	Mean stereotype valence	Lower 95% CI	Upper 95% CI
1	Go home on time	474	40.31	2.60	2.51	2.69
2	Job security	406	34.52	3.25	3.15	3.36
3	Well paid	347	29.51	3.16	3.06	3.27
4	Inflexible	301	25.60	1.68	1.57	1.79
5	Serving	265	22.53	4.55	4.47	4.64
6	Boring	235	19.98	2.21	2.09	2.34
7	Responsible	230	19.56	4.67	4.58	4.75
8	Serious	192	16.33	4.30	4.17	4.43
9	Authoritative	170	14.46	2.30	2.11	2.49
10	Lazy	154	13.10	1.55	1.40	1.71
	Weighted mean valence of stereotype profile (top 10 traits only)					3.03 (SD=1.10)
	Unweighted mean valence of stereotype profile (top 10 traits only)					3.03 (SD=1.10)
11	Conservative	147	12.50	2.50	2.30	2.69
12	Arrogant	144	12.24	1.55	1.39	1.70
13	Helpful	134	11.39	4.60	4.47	4.73
14	Trustworthy	131	11.14	4.71	4.58	4.84
15	Knowledgeable	124	10.54	4.45	4.30	4.60
16	Difficult	113	9.61	1.88	1.67	2.09
17	Friendly	107	9.10	4.31	4.13	4.49
18	Hardworking	97	8.25	4.35	4.18	4.52
19	Educated	95	8.08	4.05	3.86	4.25
20	Corrupt	82	6.97	1.55	1.31	1.78
21	Honest	82	6.97	4.35	4.14	4.57
22	Stable	80	6.80	4.15	3.98	4.32
23	Intelligent	79	6.72	4.37	4.19	4.54
24	Fair	77	6.55	4.60	4.42	4.77
25	Loyal	75	6.38	4.47	4.27	4.66
26	Calm	73	6.21	4.14	3.90	4.37
27	Integrity	73	6.21	4.58	4.41	4.74
28	Patient	71	6.04	4.00	3.77	4.23
29	Impartial	68	5.78	4.38	4.17	4.60
30	Strict	64	5.44	2.48	2.21	2.76
31	Good	59	5.02	4.12	3.91	4.33
32	Independent	39	3.32	4.05	3.75	4.35
33	Caring	38	3.23	3.82	3.37	4.26
34	Strong	37	3.15	3.59	3.24	3.95
35	Empathetic	35	2.98	4.06	3.66	4.46
36	Courageous	32	2.72	3.88	3.56	4.19
	Weighted mean valence of stereotype profile (all 36 traits)					3.33 (SD=1.08)
	Unweighted mean valence of stereotype profile (all 36 traits)					3.59 (SD=1.05)

Table S5. Stereotypes for public sector workers in South Korea (n = 633).

	Trait	Selection frequency	% of participants	Mean stereotype valence	Lower 95% CI	Upper 95% CI
1	Job security	338	53.40	4.10	4.00	4.20
2	Inflexible	176	27.80	2.23	2.06	2.41
3	Corrupt	143	22.59	1.85	1.63	2.08
4	Serving	116	18.33	4.41	4.24	4.59
5	Go home on time	100	15.80	3.69	3.49	3.89
6	Arrogant	86	13.59	2.19	1.88	2.50
7	Stable	83	13.11	4.10	3.93	4.26
8	Conservative	57	9.00	2.98	2.71	3.26
9	Authoritative	53	8.37	2.30	1.92	2.68
10	Difficult	36	5.69	3.42	3.04	3.79
	Weighted mean valence of stereotype profile (top 10 traits only)					3.26 (SD=0.90)
	Unweighted mean valence of stereotype profile (top 10 traits only)					3.13 (SD=0.89)
11	Responsible	30	4.74	3.97	3.57	4.36
12	Helpful	28	4.42	4.29	3.91	4.66
13	Well paid	27	4.27	3.96	3.64	4.29
14	Knowledgeable	24	3.79	4.00	3.59	4.41
15	Boring	23	3.63	2.65	2.14	3.17
16	Hardworking	23	3.63	4.30	4.08	4.53
17	Friendly	22	3.48	4.36	4.01	4.72
18	Fair	21	3.32	4.05	3.59	4.51
19	Intelligent	21	3.32	4.24	3.88	4.59
20	Lazy	21	3.32	2.19	1.61	2.77
21	Integrity	17	2.69	4.00	3.39	4.61
22	Educated	15	2.37	4.40	4.08	4.72
23	Good	15	2.37	4.07	3.45	4.69
24	Patient	14	2.21	4.00	3.35	4.65
25	Loyal	13	2.05	4.23	3.84	4.62
26	Strict	13	2.05	3.23	2.46	4.00
27	Calm	12	1.90	4.00	3.46	4.54
28	Strong	11	1.74	3.36	2.52	4.21
29	Trustworthy	11	1.74	4.18	3.67	4.70
30	Honest	7	1.11	4.14	3.63	4.65
31	Courageous	6	0.95	3.17	2.23	4.10
32	Impartial	6	0.95	4.17	3.56	4.77
33	Serious	6	0.95	3.50	2.66	4.34
34	Caring	5	0.79	4.20	3.47	4.93
35	Empathetic	5	0.79	3.20	1.90	4.50
36	Independent	4	0.63	4.00	2.87	5.13
	Weighted mean valence of stereotype profile (all 36 traits)					3.41 (SD=0.77)
	Unweighted mean valence of stereotype profile (all 36 traits)					3.64 (SD=0.73)