The threat of appearing lazy, inefficient, and slow? Stereotype threat in the public sector

ABSTRACT

Public employees are stereotyped as lazy, inefficient, and slow. When made aware of such stereotypes, they may experience stereotype threat that impairs their task-performance. Across two pre-registered, large-scale between-subjects experiments (n₁ = 1,543; n₂ = 1,147), we found that performance in terms of task correctness, processing time, and effort was unaffected by information of negative public employee stereotypes. Our results do not indicate stereotype threat effects for public employees in terms of task-performance. This finding offers valuable theoretical and practical implications for the understanding of public sector stereotypes and public sector reputation.

KEYWORDS:

• stereotype threat
• public sector stereotypes
• job-related task-performance
• experimental research

Introduction

Public employees are stereotyped as lazy, slow, and inefficient (e.g. Bertram, Bouwman, and Tummers 2022; De Boer 2020; Willems 2020; Döring and Willems 2023). Such negative stereotypes can reduce the attractiveness of working for the public sector (e.g. Fowler and Birdsall 2020; Sievert et al. 2020) and harm encounters between public employees and citizens (Davidovitz and Cohen 2021; Del Pino, Calzada, and Díaz-Pulido 2016). They may also have detrimental consequences for the performance of public employees, because of so-called stereotype threat (Steele and Aronson 1995). Stereotype threat is defined as a situational dilemma, in which performance is impaired by concerns of being judged based on a negative stereotype. Because people devote cognitive resources to attempt to disconfirm the stereotype, performance decreases (Leyens et al. 2000; Spencer, Logel, and Davies 2016).

While stereotype threat theory has been used to explain academic performance-gaps and domain-specific underrepresentation of social minorities (e.g. Appel, Weber, and Kronberger 2015; Hoyt and Murphy 2016; Weber, Appel, and Kronberger 2015), its occurrence is also evident in the workplace, negatively affecting employees’ work engagement, trust, and negotiation behaviour (e.g. Purdie-Vaughns et al. 2008; Tellhed and Björklund 2011). Stereotype threat effects have mostly been examined in terms of personal characteristics such as race or gender, with little attention to occupational stereotypes.

Despite the prevalence of occupational stereotypes, including public employee stereotypes (Friehs, Aparicio Lukassowitz, and Wagner 2022; De Boer 2020; Willems 2020), no extensive literature exists on stereotype threat effects of occupational stereotypes, particularly within the public employment context (Dumont and Rainville 2006). Stereotype threat effects in the public sector could have important consequences for the quality of public service provision, yet our knowledge about such intra-psychological processes in public employees is still underdeveloped (Bartels 2013; Döring et al. 2021). This article addresses this gap by examining whether the experience of stereotype threat negatively affects the task-performance of public employees.

We conducted two pre-registered between-subject experiments (n₁ = 1,543; n₂ = 1,147). The unit of analysis of both studies is the individual. Study 1 tests stereotype threat effects among public employees according to the stigmatized identity account (Steele and Aronson 1995). It is assumed that public servants’ stigmatized identity, together with negative stereotype information, triggers stereotype threat. Study 2 examines these effects according to the situationist account (Aronson et al. 1999), assuming that stereotype threat effects arise regardless of a stigmatized identity, but as a result of situational pressures. In both studies, stereotype threat was manipulated by making respondents aware of the negative stereotypes of public employees (Alter et al. 2010; Good, Aronson, and Ann Harder 2008; Spencer, Steele, and Quinn 1999). Afterwards, performance was measured on work skills-related tasks and operationalized as sustained effort (study 1), and correctness and processing time (study 2).

This research makes several theoretical contributions. Other than existing research on stereotype threat, we focus on occupational stereotype threat in the context of public employment. Certain conditions that are typical of employment within the public sector (e.g. public service motivation, emotional labour) may impact the occurrence of stereotype threat effects, thereby challenging a clear-cut application of stereotype threat theory from former researched contexts to the public sector (Döring and Madsen 2022; Hsieh, Yang, and Kai-Jo 2012; Pedersen, Stritch, and Thuesen 2018). This research broadens the discourse of stereotype threat theory by examining the context of occupational stereotypes, and provides meaningful insight into stereotype threat and its consequences for public service performance.

Current findings on public sector stereotypes are mostly descriptive (e.g. De Boer 2020; Willems 2020), and consequences of stereotyping remain largely unclear. Drawing on stereotype threat theory, we examine how stereotyping affects performance. This will contribute to broadening the discourse on public employees’ social identities and how these play a role in public service delivery. From a practical perspective, we offer valuable implications for stereotype management regarding employees’ performance, their well-being, as well as their retention. A detailed discussion of the implications can be found in the conclusion section.

Conceptual framework and hypothesis

Social categorization is the process by which individuals categorize themselves and others into differentiated social groups based on a variety of characteristics, including gender, ethnicity, and occupation (Fiske 2018; Imhoff, Koch, and Flade 2018). These groups are further associated with specific attributes, from which stereotypes evolve (e.g. Friehs, Aparicio Lukassowitz, and Wagner 2022). Occupational stereotypes, in that sense, refer to preconceived beliefs about individuals who hold a specific occupation (Friehs, Aparicio Lukassowitz, and Wagner 2022; He et al. 2019). To illustrate, during a typical public service encounter, one individual holds the occupation of a public employee; the other one holds the role of the client. Both ascribe competencies and traits towards the other (i.e. stereotypes) and have corresponding expectations of the service interaction. Due to uncertainty in these encounters, stereotypes about the counterpart serve to fill in missing social information to enable socially effective and less demanding interactions (Frey and Tropp 2006; Raaphorst 2018; Raaphorst and Van de Walle 2018).

Public employees are a specific occupational group known to be stereotyped by citizens (Bertram, Bouwman, and Tummers 2022; De Boer 2020; Döring and Willems 2023; Van de Walle 2004; Willems 2020). There are also occupational stereotypes variations within the public sector. Teachers are often perceived as caring and warm (Friehs, Aparicio Lukassowitz, and Wagner 2022, Imhoff et al. 2013; Willems 2020), whereas police officers are often seen as helpful and caring but also as racist (Cochran and Warren 2012; Willems 2020).

However, the different occupations within the public sector are often grouped together as one overarching, conflated social category, namely public employees (e.g. Willems 2020). This overarching occupational category is – just like the specific occupations within it – subject to stereotypes. Public employees are commonly stereotyped as being lazy, inefficient, unmotivated, overly bureaucratic, and slow (Willems 2020; De Boer 2020). At the same time, they are also perceived as ethical, prosocial, dedicated, and helpful (e.g. Szydlowski, De Boer, and Tummers 2022; Willems 2020).

The effect of stereotype threat on performance

The theory of stereotype threat presumes that individuals hold multiple, simultaneous memberships in social categories, including their occupation (Steele and Aronson 1995). Depending on the social context, categories can become relatively more psychologically salient to individuals than other categories. Various stimuli, such as a performance task in a certain work- or academic context or an unpleasant interaction, can affect the degree to which a social category is perceived as salient (e.g. Emerson and Murphy 2014; Hewstone and Greenland 2000). As the salience of the social category increases, so does the awareness of the stereotypes associated with it.

When individuals realize the possibility of being negatively evaluated based on a stereotype they became aware of, they believe that their competence or moral goodness is questioned (Steele, Spencer, and Aronson 2002). This experience is called stereotype threat (Steele and Aronson 1995). In order to cope with the unpleasant consequences of stereotype threat, coping processes such as excessive performance monitoring or negative thought suppression are applied. These processes deplete cognitive resources that are needed for other tasks, and ultimately lead to poor performance (e.g. Emerson and Murphy 2014; Inzlicht and Schmader 2012; Schmader, Johns, and Forbes 2008). Literature provides two perspectives on the underlying mechanisms at play when experiencing stereotype threat: the stigmatized-identity account and the situationist account (Aronson et al. 1999; Steele and Aronson 1995). According to the stigmatized-identity account, negative stereotype information activates an individual’s stigmatized identity, which then triggers the experience of stereotype threat (Steele and Aronson 1995). Negative stereotypes convey to individuals associations that devalue their social identity. For example, negative stereotypes accuse women of being emotional, bad at maths, and lacking leadership aptitude.

However, a stigmatized identity is not necessarily a predicament for stereotype threat to arise (Aronson et al. 1999). Instead, stereotype threat has also been defined as a situational threat, which arises as a result of situational cues that signal diagnosticity (i.e. self-knowledge, self-evaluation) of one’s own skills, and potentially threaten a positive self-concept. Consistent with the situationist perspective (Aronson et al. 1999), the experience of ‘stereotype threat requires neither a history of stigmatization nor internalized feelings of performance-inferiority but can arise and become disruptive as a result of situational pressures alone’ (Aronson et al. 1999, p. 1). This is because stereotype threat is assumed to derive its power from the general motive to sustain a positive or competent self-image, and the ability to secure important outcomes (Aronson et al. 1999; Steele, Spencer, and Aronson 2002). These needs are inherently human and appear to emerge based on self-threatening situational cues, regardless of a stigmatized identity (Aronson et al. 1999). Situational cues, for example, referring to a test in which stereotype-relevant characteristics are assessed, have been shown to influence the degree to which certain stereotypes become relevant. Hence, merely situational cues that activate the concern to underperform can serve as a sufficient prerequisite for stereotype threat consequences (e.g. McKay et al. 2003; Shapiro and Aronson 2013; Murphy and Taylor 2012).

Generally, stereotype threat theory has been used to explain performance gaps, or domain-specific underrepresentation of social minority groups. While most research has been conducted in the context of academic achievements of women (e.g. performance on maths or spatial ability tests; Martens et al. 2006; Walton and Spencer 2009; Van Loo et al. 2013) and racially marginalized groups (e.g. performance on intellectual tests; Davis, Claytie, and Salinas 2006; Steele and Aronson 1995; Walton and Spencer 2009), stereotype threat effects are also evident in the workplace. Accordingly, stereotype-concerned employees show performance decrements regarding speed and accuracy in cognitive tasks (e.g. verbal or memory abilities; Schmader, Johns, and Forbes 2008), or managerial decision-making tasks (Bergeron et al. 2006). They also report less work engagement, negotiate worse outcomes, show less trust, seek less constructive feedback towards employers, and are less likely to choose leadership roles (e.g. Kray and Shirako 2011; Purdie-Vaughns et al. 2008; Tellhed and Björklund 2011; Kulik et al. 2016).

Despite work affiliation being a powerful social category on which others stereotype (e.g. Friehs, Aparicio Lukassowitz, and Wagner 2022), stereotype threat theory has not been readily applied to public employees. To illustrate, public employees who are aware of the negative stereotypes of their group might feel concerned about confirming the negative stereotype while performing a job-related task (Kray and Shirako 2011). To protect themselves from the consequences of this stereotype threat, coping processes such as excessive performance monitoring or negative thought suppression are applied. These processes deplete cognitive resources that are needed for work-related tasks, and ultimately lead to poor performance (e.g. Emerson and Murphy 2014; Inzlicht and Schmader 2012; Schmader, Johns, and Forbes 2008).

Investigating stereotype threat in public employment therefore contributes to the growing research on public sector stereotypes and is particularly relevant for the better understanding of public service performance and citizen-state interactions (Bartels 2013). We investigate whether stereotype threat affects public employees’ performance in terms of sustained effort, task correctness, and processing time. In line with previous literature, our main hypothesis postulates that:

Hypothesis:

Stereotype threat reduces public employees’ task-performance when being confronted with negative stereotype information about public employees, compared to no negative stereotype information.

Methods

We conducted two pre-registered between-subject survey experiments in Anglo-Saxon countries (n₁ = 1,543) and in Austria (n₂ = 1,147). Although Austria and the Anglo-Saxon nations have differing governance structures and public administration systems (Peters 2021), they are comparable in terms of citizens’ perceptions regarding the public sector. Public employees are seen as lazy, inefficient, or corrupt. This is observable in the media, in public opinions, employee polls, as well as in published research in both contexts (Federal Ministry of Arts, Culture, Civil Service, and Sport 2022; Carroll and Siegel 1999; IMAS International 2022; Willems 2020).

Study 1 follows the ‘stigmatized-identity’ account for testing stereotype threat effects among public employees (Steele and Aronson 1995). Here, the stigmatized identity of the public employee is activated by writing about a negative public employee stereotype, which should trigger stereotype threat. Study 2 builds on the situationist account (Aronson et al. 1999) and tests stereotype threat effects for a broader sample with employees of different sectors. Here, stereotype threat effects are assumed to arise regardless of a stigmatized identity (i.e. being stigmatized because of the social membership to public employees), but as a result of situational pressures. Situational pressures are experimentally manipulated by exposing respondents to a performance-task, for which the negative stereotype is relevant for. Respondents could experience pressure to achieve a positive outcome and/or self-image and subsequently underperform.

Task-performance covers a person’s contribution to organizational performance, including the completion of tasks specified for an individual’s work role (e.g.; Sonnentag, Volmer, and Spychala 2008; Rich et al. 2010). We operationalize task-performance in terms of respondents’ sustained effort (study 1), and correctness and processing time (study 2), on a series of adapted tasks that mimick the requirements of everyday clerical work and typical assessment centre tasks (e.g. analytical thinking, sense reading, understanding of interconnections). Such clerical work and simple assessment tasks are relevant and common in a wide range of public sector occupations, including jobs as police officers, teachers, or municipal employees.

Both studies were conducted independently and the study designs were uninformed by each other. The point of collaboration occurred by chance after data was shared at a conference. Both experiments, however, share the main objective of documenting stereotype threat effects on task-performance. Thus, even though both studies pre-registered two separate hypotheses (visible in pre-registrations), one main hypothesis that combines the goal of both studies is stated.

Study 1 and study 2 were both pre-registered (Study 1: https://osf.io/7jvxy/?view_only=f58cfba2353c4688b2e4068fc6c660e4; Study 2: https://osf.io/cuyn6/?view_only=1c69573029b14d388acdfdf1efa4e814) and ethically approved. Approvals, data and codes can be found on the Open Science Framework (Study 1: https://osf.io/e2nc5/?view_only=32f17c41b38441dd94bc9301049023e9;

Study 2: https://osf.io/cuyn6/files/?view_only=9d7c681553004ef598c434279f017524).

Study 1: Effects of negative stereotype information on sustained effort in a real-effort task

Study 1 tests whether activating the stereotype of the lazy public employee affects employees’ task-performance in terms of sustained effort. Sustained effort requires continuous attention and vigilance, crucial for various performance settings in the public sector including surveillance, inspection and monitoring work (Matthews et al. 2010). By using a sample of public employees, study 1 follows the ‘stigmatized identity’ approach of stereotype threat. Following our hypothesis, we expect respondents who are confronted with negative stereotypes to perform worse.

Experimental design and procedure

We carried out an online between-subjects experiment. Respondents were asked whether they are currently employed in the public sector. Respondents who confirmed that they are public employees are assumed to categorize themselves to the social group of public employees (e.g. Trepte and Loy 2017) as self-categorization predicts the sense of identification with the group (e.g. Van Dick and Wagner 2002). Thus, only respondents who self-identify as public employees took part in this experiment.

Respondents were randomly assigned to a negative stereotype condition or an active control condition. The original pre-registered study was designed to include four other conditions investigating whether coping strategies can be used to counter the effects of negative stereotyping. However, consistent with the main hypothesis, we focus on the two conditions relevant for the shared main hypothesis: the negative stereotype and active control conditions. Complete information and results pertaining to all conditions can be found in the online supplementary materials of study 1.

In the negative stereotype condition, we activated negative stereotypes through an active writing task (Dijksterhuis and van Knippenberg 1998). Respondents were asked to describe an example at work where the behaviours of their colleagues could be seen as lazy. We asked respondents about others in their occupational group instead of about themselves, because self-focus can moderate the effects of stereotypes (Macrae et al. 1998). Personal standards about work performance may become salient and cause participants to deny the stereotype completely. Since the goal was to activate a negative stereotype that could apply to them as a public employee, describing the behaviours of others who share the same group identity should activate the same stereotype.

Respondents in the control condition wrote about their favourite television shows. By using an active control condition, we ensure that effects (e.g. fatigue or attention) are constant across all groups involved in the study (Haaland, Roth, and Wohlfart 2023).

After being exposed to one manipulation, respondents completed a performance task based on the real-effort task used in DellaVigna and Pope (2018). Respondents had to alternately press the ‘b’ and ‘c’ keys on the keyboard, while an average task-score of 1,142 (known because of a pilot-test) is presented to induce more pressure. Additionally a 5-minute countdown is depicted and the points accumulated. The outcome measure, referred to as the effort score, is the total score respondents receive for correctly pressing the keys. While the task may not be meaningful per se, it has features that parallel clerical jobs (repetition that is monotonous and tiring). Still, concentration and precision are required.

Importantly, respondents in the negative stereotype condition were reminded of the stereotype right before engaging in the performance task: ‘People expect public employees to be lazy and to receive a low score on this task’. Figure 1 (in the online supplementary materials of study 1) shows an overview of the study procedure. All materials can be found in the online supplementary materials of study 1.

Sample

Based on a power analysis for a small effect size of f = 0.10, a significance level of 0.05, and .80 statistical power, we aimed for a minimum sample size of 1,289 respondents. The specification of the effect size is based on extant literature: The documented effect sizes for stereotype threat are rather small (Flore and Wicherts (2015) report a small effect size (d = − 0.22) for gender-related stereotype threats). Since there is a lack of prior research examining occupational stereotype threat, we adhere to the suggested f = 0.10.

We recruited 2,219 public employees in Anglo-Saxon countries through survey companies such as Prolific (88.25%), Dynata (9.93%), and Amazon Mechanical Turk (1.82%; average payout rate of £11.66 per hour per respondent). Respondents had to pass two out of three attention checks to be included in the study. We also excluded respondents for bad data quality in terms of bad verbatim of the qualitative responses to the study’s writing tasks and those with no scores on the performance task. Attention checks can be found in the online supplementary materials of study 1. After exclusion, 1,543 respondents were included in our analyses (M_age = 38 (SD_age = 11.13); 59.62% females). Table 1 in the online supplementary materials (study 1) shows the descriptive of the sample.

Table 1. Results of the one-way between subjects ANOVA and ANCOVA.

	Sum of Squares	Df	Mean Square	F	p-value
ANOVA
Conditions	481610	5	96322	1.19	0.312
ANCOVA
Conditions	484465	5	96893	1.37	0.23
Age	12071125	1	12071125	170.26	<2e-16
Gender	3829157	3	1276386	18.00	1.58e-11
Years of employment in public sector	353152	4	88288	1.25	0.29
Occupation sector	792815	5	158563	2.246	0.05

Occupation sectors are coded as education (1), law, enforcement, and security (2), health and social care (3), utilities and resources (4), tax, revenue and customs (5) and others (6).

Results

Table 2 in the online supplementary materials (study 1) shows descriptive statistics of participants’ task-performance scores in the negative stereotype (M = 1,169.22) and control conditions (M = 1,115.60).

Table 2. Overview of the logistic regression and the 2-way ANOVA (type II) explaining likeliness of correct responses and task processing times.

	Correct Responses Text Task			Correct Responses Diagram Task
Predictors	Odds Ratios	CI	p	Odds Ratios	CI	p
(Intercept)	3.24	2.22–4.86	0	2.78	1.93–4.10	0
Negative Stereotype Information [1]	0.77	0.45–1.29	0.32	0.98	0.59–1.65	0.95
Task Relevance [1]	0.72	0.46–1.11	0.15	0.64	0.42–0.97	0.04
Stereotype Threat	1.3	0.72–2.38	0.39	1.12	0.62–2.02	0.7
Observations	1,147			1,147
R^2 Tjur	0			0.01
	Processing Time Diagram Task			Processing Time Text Task
Predictors	Sum Sq	Df, F	p	Sum Sq	Df, F	p
Negative Stereotype Information	530588.5	1, 0.94	0.33	5.18	1, 1.31	0.25
Task Relevance	115553.36	1, 0.21	0.65	4.5	1, 0.001	0.97
Stereotype Threat	91416.27	1, 0.16	0.69	2.34	1, 0.59	0.44
Residuals	645057990	1143		4.53	1143

Consistent with our pre-registered analysis plan, we ran a one-way, between-subjects ANOVA to test whether respondents’ scores on the effort task differ statistically between those for whom negative stereotypes were activated and the control conditions (including the other experimental conditions). Results revealed no significant differences across conditions (F(5, 1537) = 1.19, p = 0.31).

Post hoc pairwise t-tests between conditions with Benjamini-Hochberg False Discovery Rate corrections also revealed no significant differences between mean scores on the effort task.¹ Being confronted with negative stereotype does not lead to poor task-performance among public employees (M = 1169.22) in comparison to those who are not confronted with the stereotype (M = 1115.60, t(516)= −2.09, p.adjust = 0.37, 95% CI=[−103.97, −3.28]).

As a check for robustness, an ANCOVA was conducted to control for the effects of age, gender, years of employment in the public sector and occupation sector. Results revealed no significant differences despite controlling for these variables. Table 1 summarizes the ANOVA and ANCOVA results. Given that selective attrition can also threaten internal validity (Vazire, Schiavone, and Bottesini 2022), all analyses were also repeated with respondents that were excluded (results do not differ significantly). Only pre-registered analyses are reported in the manuscript. Results with the whole sample can be found in the online supplementary materials of study 1.

In summary, we found no differences in performance between respondents who have been confronted with negative stereotype information and those who have not.

Study 2: Effects of negative stereotype information on task correctness and processing time

Study 2 assesses whether negative stereotype information, for tasks the stereotype is relevant for, affects task-performance in terms of correctness and processing time. Task-performance specifically relates to the completion of tasks in accordance with a respective work role (e.g. Rich et al. 2010). Thus, task-performance is operationalized as whether the task has been carried out factually error free, and whether it has been solved in a reasonable amount of time. A short processing time with a correct solution of the task represents high task-performance.

It follows the situationist perspective of stereotype threat, adapted from the study of Aronson et al. (1999). Accordingly, we conceptualized the design of study 2 independent of whether respondents identify with a stigmatized identity (i.e. public employee identity; Steele and Aronson 1995). We argue that stereotype threat effects can arise and depress performance as a result of situational cues (such as the design of a performance task in which stereotype-relevant behaviour is assessed).

Study 2 tests stereotype threat effects in a broad sample, where not all respondents share a stigmatized identity (i.e. public employees) in the real world. Instead, this study experimentally manipulates the provision of negative stereotype information, and additionally introduces a performance task, for which respondents would feel targeted by the stereotype relevant to the task. Stereotype threat is thus created by the combination (interaction) of both conditions: negative stereotype information and whether the task is relevant for the respective negative stereotypes that respondents were exposed to. To illustrate, for respondents who are working on a task for which ‘skills typically needed in the public sector’ are required, the negative stereotypes about employees who work on such tasks (for skills relevant in the public sector) become salient. By targeting respondents with a stereotype relevant to their task-performance, respondents will experience pressure to sustain a competent self-image and/or to achieve a positive outcome and performance will be impaired.

Accordingly, we test whether respondents who are confronted with negative stereotype information, while performing a task relevant for skills needed in the public sector, will show decreased performance with respect to (1) task correctness and (2) task processing time (compared to respondents who received no negative stereotype information and/or who work on tasks that are not relevant for skills needed in the public sector).

Experimental design and procedure

We carried out a 2 (Negative stereotype information: neutral stereotype information vs. negative stereotype information) × 2 (Task relevance: relevance vs. no relevance regarding skills needed for a task when working in the public sector) between-subject experiment. Respondents were randomly assigned to neutral or negative stereotype information before working on a text- and diagram-task. Next to stereotype information, we manipulated the level of relevance of the task for skills typically needed in the public sector. All respondents received the same task, but relevance was varied by explaining respondents beforehand that the task would require skills that are relevant or not for the public sector.

Stereotype information and task relevance thus represent the independent variables. The combination (interactions) of both variables shall elicit the occurrence or absence of stereotype threat.

As main dependent variables, we measured (1) correctness of the responses to the tasks and (2) processing time to complete the task. Additionally, in accordance to our pre-registered analysis plan, we measured work conscientiousness, work-motivation, and self-efficacy as dependent variables. Since these do not contribute to answering our main hypothesis, their analysis can be found in the online supplementary materials for study 2.

After giving consent, respondents first read an article on an Austrian newspaper website on what society and CEOs from private organizations think about public employees. In the negative stereotype condition, respondents read that public employees are associated with words such as ‘lazy’, ‘slow’, and ‘inefficient’, and that private sectors CEOs are not inclined to hire public employees. In the control condition (no negative stereotype information), it was stated that public employees are associated with mainly neutral words such as ‘uniform’, ‘public’, and ‘work’, and that private sector CEOs see public employees similar to the associated words. The newspaper articles can be found in the online supplementary materials of study 1.

Afterwards, all respondents were assigned two real assessment centre tasks in random order. Before respondents could work on the tasks, they received the information that the tasks concerned testing skills that are important and needed particularly in either the public sector (task relevant for skills in the public sector) or the private sector (task not relevant for skills in the public sector). The diagram task provided information about households’ energy consumption, while the text task depicted information about calendar systems. Afterwards, respondents saw one statement about the content of each task-information and had to choose whether this statement is correct, wrong, or unanswerable. Before starting each task, respondents were informed that they had 80 seconds to finish each task. After 80 seconds, tasks closed automatically. Performance was measured in terms of task correctness and processing time. Correctness was conceptualized for both tasks separately and was operationalized as correct ( = 1) versus incorrect ( = 0). Processing time was operationalized as the time in seconds it took to complete the task. Figure 1 (in the online supplementary materials of study 2) provides an overview of the study procedure. All stimuli materials can be found in the online supplementary materials of study 2.

Attention checks and manipulation check

Two attention and one manipulation check were included in the survey. The first attention check was placed randomly among scale questions and stated ‘To ensure questions are read accurately, please check “strongly agree (7)” for this question’). The second asked about basic information provided earlier about whether the skills were relevant for a task in the public or private sector. The manipulation check was asked after the completion of the tasks and measured whether respondents were aware of the stereotype information (negative stereotype information vs. neutral stereotype information). It was asked ‘According to a study in the previous article, what words/characteristics were most associated with public employees?’ and could answer ‘Lazy and hair-splitting, inefficient, and slow’ or ‘Uniform, public, and work’. This measure is commonly used as a manipulation check for stereotype threat manipulations (e.g. Flore, Mulder, and Wicherts 2018; Pennington et al. 2016).²

Results concerning the manipulation check indicate that the majority answered this measure correctly (81.6%). A logistic regression reveals a significant effect of ‘stereotype information’ (1 = negative stereotype information; 0 = no negative stereotype information (0 is used as reference category)) on respondents’ likelihood to remember and choose the correct response option (1 = negative stereotypical associations in article; 0 = neutral stereotypical associations in article): OR = 3.21, p < 0.001.

Sample

A power analysis for a 2 × 2 full factorial ANOVA using the package ‘pwr’ in R (Champely et al. 2020) indicated a minimum of 274 respondents per treatment group, leading to a total minimum sample size of 1,096. This sample size allows the detection of a small effect size (f = 0.10, power = 0.80, significance level = 0.05).

Data were collected in Austria. Relying on Qualtrics panels, 4,009 respondents participated. The average payout rate was €13.5 per hour for each respondent. We excluded respondents who failed attention and manipulation checks, leading to a final sample of 1,147 respondents for analysis (samples do not differ in respect to measured demographics; nor do the results differ between the reduced and the original sample). Results of the full sample analysis can be found in the online supplementary materials for study 2.

The elimination of respondents led to unequal sample sizes in the four conditions (sample sizes per condition are reported in Table 1 in the supplementary materials of study 2). Potentially, attention checks were remembered differently by the treatment groups, in turn causing different drop-out rates. However, as these sample selection criteria for data quality have been pre-registered and loosening the strict selection criteria would be in contrast to our initial aim of only including respondents that paid sufficient attention, we decided to continue our analysis with the most restrictive sample. We chose an analysis approach that counters potential undesired side-effects of unequal sample sizes and calculated an unbalanced type II ANOVA for the analysis of processing time.

The sample for analysis is gender-balanced (49.61% female, 0.26% diverse, M_age = 41.10 years (SD_age = 15.56)). With respect to education, 4.0% finished primary school, 34.4% finished vocational school, 29.1% finished high school, and 24.6% attended university. Most respondents were employees (75.7%), with a majority employed in the private (48.9%), public (19.9%) or non-profit sector (6.8%).

Results

Descriptive statistics regarding task-performance measures are presented in Table 1 in the online supplementary materials of study 2.

Response correctness and processing-times

Logistic regressions were conducted to test the effects of each independent variable ‘negative stereotype information’ and ‘task relevance for skills in the public sector’ and their interaction (‘stereotype threat’) on response correctness for the diagram and text task (0 = incorrect; 1 = correct). For logistic regression analysis, ‘no negative stereotype information’ (0), and ‘task relevance for skills in the public sector’ (0) were chosen as reference categories. Coefficients indicate the extent to which the likelihood of response correctness differs if negative stereotype information is present, for a task that is relevant for skills in the public sector. Table 2 reports the odds ratios, with their 95%-confidence intervals, for the likeliness to have a correct response based on the main and interaction effects.

Findings for the text task indicate no significant interaction effect (‘stereotype threat’) between ‘negative stereotype information’ and ‘task relevance for skills in the public sector’ on the likelihood for a correct response (OR_text = 1.30, p = 0.39).

For the diagram task, results indicate that ‘task relevance for skills in the public sector’ has a significant effect on respondents’ likelihood for a correct response. That means that working on a task that is relevant for skills needed in the public sector leads to a decreased likelihood for a correct response (OR_diagr = 0.64, p = 0.04). However, findings do not indicate a significant effect of ‘negative stereotype information’ (OR_diagr = 0.98, p = 0.95), nor do they point to a significant interaction effect (‘stereotype threat’) on the likelihood for a correct response (OR_diagr = 1.12, p = 0.70).

Regarding processing times, two-way ANOVAs (type II) were conducted to test the effects of ‘negative stereotype information’ and ‘task relevance for skills in the public sector’, and their interaction ‘stereotype threat’ on processing time on the diagram and text task.

As depicted in Table 2, findings for the diagram task indicate no significant interaction effect (‘stereotype threat’) between ‘negative stereotype information’ and ‘task relevance for skills in the public sector’ on the likelihood for a correct response (F(1, 1143) = 0.16, p = 0.69).

For the text task, findings do not indicate a significant interaction effect (‘stereotype threat’) on task processing time (F(1, 1143) = 0.59, p = 0.44; see Table 2.

The findings indicate that respondents who received negative stereotype information and worked on a task relevant for skills needed in the public sector, do not show a decreased likelihood of a correct task response, nor exhibit increased task processing time. Findings do not support the hypothesized negative effect of stereotype threat on task-performance.

Finally, to account for job identification, we conducted complementary analyses that account for ‘identification with a work activity in a public sector job’ as a control variable. Controlling for identification still indicates no significant interaction effects of ‘negative stereotype information’, ‘task relevance for skills in the public sector’ and ‘identification’ for the likelihood of a correct response for both tasks, and for processing times for the diagram task. Find the detailed results in tables 3 and 4 in the online supplementary materials of study 2.

Summary of results

Table 5 in the online supplementary (of study 2) summarizes the findings for study 1 and 2. Across the two studies, no stereotype threat effects were found in terms of sustained effort, task correctness, or processing times.

General discussion

No evidence of performance decrements is found due to stereotype threat in terms of task-performance. Neither the stigmatized identity (study 1), nor the situationist approach (study 2), yielded significant stereotype threat effects. We reach these conclusions after testing threat effects in two high-powered, independently and carefully designed experiments with different performance measures.

These findings may be explained in various ways. First, despite the extensive literature, the existence of stereotype threat effects are questioned. Critical voices refer to very small effect sizes, the existence of publication bias and the questionable relevance for real-life contexts (e.g. Flore and Wicherts 2015; Ganley et al. 2013). Considering this background, our findings contribute to inconsistent results.

Second, it is possible that the overall public employee identity provides too little basis for identification or self-relevance to feel threatened, or that there is a lot of variation between public employees in terms of their identification. Relatedly, the ‘publicness’ of public sector identity may be less pronounced than assumed by many scholars (Baarspul and Wilderom 2011). Public services are often a mixed delivery system, where sector involvement is ambiguous. Meier, Johnson, and Seung‐Ho (2019) suggested that stereotype applications are often less relevant in such situations. Van de Walle (2004) also argues that such stereotypes matter less at the specific micro-level. Indeed, Döring and Willems (2023) showed that only small differences exist for characteristics attributed to public servants in comparison to private sector employees. They suggest that there may be more variance between different subsectors/industries and job types, than between the general public and private sector. Indeed, the overarching category of public employees that we have used in our studies did not differentiate between the various job profiles it comprises. Different groups of public employees in our samples (e.g. police officers, teachers, or employees of an administrative authority) may not be threatened by stereotypes about the general public employee identity.

Third, there may be specific conditions to working in the public sector that mitigate stereotype threat effects. Employment in the public sector is strongly associated with public service motivation (PSM). This motivation arises from internalized public service values among those who work in the public sector, emphasizing serving the public interest (Perry and Wise 1990; Vandenabeele 2007). Employees with high PSM have a greater sense of mission, responsibility, and commitment, and higher perceptions of meaningfulness (Brewer and Coleman Selden 1998; Pattakos 2004; Van Witteloostuijn, Esteve, and Boyne 2017). PSM may thus serve as a buffer for employees’ self-concept in the face of negative stereotypes (Brewer, Coleman Selden, and Facer 2000), thereby mitigating stereotype threat effects.

Finally, it may be that public employees have developed coping strategies to mitigate the effects of negative stereotyping on performance. Distancing oneself from the social category one is stereotyped for, for example, can be used as a coping strategy (Van Veelen et al. 2020). Furthermore, since positive public employee stereotypes exist as well (e.g. dedicated, helpful), the knowledge of these positive attributions in general may mitigate the effects of negative stereotypes, or even serve as a boost for employees’ performance when explicitly made aware of them (i.e. stereotype boost; Shih, Pittinsky, and Ho 2011).

Limitations and future research

This study is not without limitations. First, the usage of artificial tasks and scenarios does not fairly reflect real-world behaviour of employees. The same accounts for the practice of priming or framing, commonly used in experiments (Doyen et al. 2012).

This introduces the challenge of how to study stereotypes meaningfully and ethically as stereotypes in general are in fact certain framings. One particular concern for our studies is whether the framings were sufficiently strong to affect performance. Based on the manipulation check, we are certain that our framings worked mostly as intended. On the other hand, tradeoffs between realism and high internal validity must be taken when designing experiments. Because we aimed to investigate causal stereotype threat effects in a highly controlled setting, we chose randomized experiments with high internal validity. To the extent that it was possible, an effort to use realistic designs was made. This was realized by showing a realistically manipulated newspaper article, presenting real information, and the use of objective performance measures for tasks partly from organizational assessment centres. The circumstances in which the studies were conducted, however, did not fully mimic the severity of everyday work. Besides a call for replication studies, future research could use qualitative methods (e.g. observations) to realistically investigate stereotype threat in work routines of employees, or within public service encounters with stereotyping clients.

Second, while our findings were replicated across the Austrian and Anglo-Saxon governing context, we cannot ensure full generalizability to other regional, cultural, or administrative contexts. Different governance paradigms have been shown to affect public employee’s performance, specifically through their different pressures and incentives structures (Davidovitz et al. 2022). In administrative contexts similar to a consumer-driven market, public organizations adopt an entrepreneurial orientation, pushing public employees to be competitive. This can influence how the public work environment is perceived by public employees and clients. We therefore suggest investigating stereotype threat research in different public sector settings, with the prime objective of investigating the impact of governing structures.

Relatedly, we recommend exploring stereotype threat effects among different public sector occupations. Different occupational contexts may present distinctive opportunities for identifying with the public employee identity and coping with stereotype threat. In that sense, given that stereotype threat effects appear among members of minority groups (women, racial minorities), examining the intersection of public sector identity and other social identities of the employee should not be neglected in future research. It may be likely these employees are double burdened by stereotype threat, which could impact their performance as well as their retention. Yet from a representative bureaucracy perspective, these employees play a key role for the representation of marginalized groups in society (Bishu and Kennedy 2020). Given that employees of minority- and marginalized groups are already more difficult to attract to government jobs (Linos 2018), stereotype threat effects for these employees deserve our full attention in future research.

Finally, it should be noted that negative public image, including that perpetuated by negative stereotypes, can also be seen as a source of reputational threat to an organization (Eisenegger and Schranz 2011; Gilad, Bloom, and Assouline 2018). As our study investigated a self-threat on the individual level, we do not know how employees’ stereotype threat concerns are related to a shared identity at an organizational level. Döring et al. (2021) found that public managers care about how different audiences perceive their organization, and engage in performance assessments for reputation management. Gilad, Bloom, and Assouline (2018) found that employees who feel central and influential within the organization, organizational attachment is unwavering – or even reinforced – when confronted with negative reputation signals. Conversely, employees who feel marginal and powerless are more vulnerable to reputation signals. In this regard, public employees’ organizational attachment or self-esteem could interact with stereotype related threats to determine their performance. Scholars have argued that reputational threats are central to an understanding of organizations’ behaviour, as they shape the way organizations respond to external pressures and the need for change (Carpenter 2010; Döring et al. 2021; Maor 2015). Against this background, future research should assess the potential interplay between other psychological mechanisms, such as those related to reputational threat, and stereotype threat.

Implications and conclusion

Although this study does not confirm stereotype threat effects on performance, it has important theoretical and practical implications for public administration. We tested the stereotype threat theory by moving beyond gender, age, and race-related stereotype threat, and beyond academic performance domains (e.g. Appel, Weber, and Kronberger 2015, Hyot and Murphy 2016; Nguyen and Ryan 2008; Weber, Appel, and Kronberger 2015). This research focuses on occupational stereotype threat in the context of public employment. Public sector employment entails specific conditions (e.g. public service motivation, emotional labour and high psychological costs, discretion) that might challenge a meaningful translation of stereotype threat effects from former researched contexts to the public sector (Döring and Madsen 2022; Hsieh, Yang, and Kai-Jo 2012; Pedersen, Stritch, and Thuesen 2018). Thus, the employment sector can function as a boundary condition for occupational stereotype threat effects to appear. Thereby the scope of stereotype threat theory is broadened to the meaningful context of public employment.

Given the growing research on public sector stereotypes, investigating stereotype threat is particularly relevant for expanding and deepening our understanding of citizen-state interactions from a micro-level perspective (Bartels 2013). Since stereotypes and their downstream consequences may affect employees’ identification with public service work, their motivation (Vandenabeele 2007), and public organizations’ reputation (Döring et al. 2021; Gilad, Bloom, and Assouline 2018), it is important that future debates on these topics continually develop a more holistic perspective on public identity- and organizational threats.

Moreover, literature points to stereotype application during citizen-state interactions (e.g. Bertram, Bouwman, and Tummers 2022; De Boer 2020; Willems 2020). While our findings do not find evidence of stereotype threat effects on task-performance, it is likely that stereotype threat effects may manifest in other forms, especially those embedded in public service encounters between clients and employees such as quality of citizen-official interaction and citizen satisfaction. In public encounters, when individuals perceive uncertainty due to a lack of information, they tend to rely on stereotypes to fill in missing social information (Frey and Tropp 2006; Raaphorst 2018; Raaphorst and Van de Walle 2018). This reliance may elicit metacognition of how one is likely to be viewed by the respective other – which potentially could result in stereotype threat when negative evaluations are expected.

From a practical perspective, since negative public sector stereotypes are ubiquitous, it is important to understand the different ways in which this impacts the public sector and its employees. In addition to stereotype threat effects, there can be serious consequences for employee retention and recruitment; for well-being and sick-leave, motivation, and service quality (Casad and Bryant 2016; Mikolon, Kreiner, and Wieseke 2016), and for the attractiveness of public sector occupations and the sector as a whole (e.g. Fowler and Birdsall 2020; Sievert et al. 2020).

Finally, even if performance is not affected, stereotype threat may still affect public employees in different ways. For example, public employees might experience stress or frustration in the face of stereotypes, or may monitor whether clients are stereotyping. This might affect their willingness to provide service, their discretionary decisions, or affecting the interaction, rendering it unproductive (Wieseke et al. 2012; Habel et al. 2016; Davidovitz and Cohen 2021).

Understanding the effects of stereotype threat on public employees thus has implications for interventions, policies, and public sector management. Organizations are in the position to reduce the impact that stereotype threat has on its employees (Roberson et al. 2003). Public managers should acknowledge existing stereotypes and openly address them (e.g. Kray et al. 2001) to create strategies to disempower negative stereotypes. This can be achieved by highlighting certain environmental cues (humorous artefacts, positive affirmations), or social networks to discuss stereotypes and their impact (Kulik et al. 2016). Lastly, it may inform strategies for recruitment campaigns, transparent communication in service provision, or social media communication.

Concluding, the studies are important contributions about the effects of negative stereotypes in the public sector. Considering the role of public employees, it is precisely in this population where research on intra-psychological threats is likely to be of great importance. Only continuing research can estimate the severity of stereotype threat to develop evidence-based interventions.

Acknowledgements

We would like to thank the editor and the anonymous reviewers for their constructive suggestions on previous versions of this article.

Disclosure statement

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

Data availability statement

The authors confirm that the data supporting the findings of the studies are available within the article. Data and codes can be found on the Open Science Framework (Study 1: https://osf.io/e2nc5/?view_only=32f17c41b38441dd94bc9301049023e9 Study 2:https://osf.io/cuyn6/files/?view_only=9d7c681553004ef598c434279f017524).

Supplemental data

Supplemental data for this article can be accessed at https://doi.org/10.1080/14719037.2023.2229326

Additional information

Funding

Study 1 was supported by the NWO under Grant 016.VIDI.185.017. Study 2 was supported by the Austrian Science fund under Grant P36098-G; and by the Small Project Grant by WU Competence Center for Experiemental Research.

Notes

1. Full results of the post-hoc pairwise t-tests can be found in table 6 in the online supplementary materials of study 1.

2. This item was pre-registered as an attention check. But given that it is typically used as a manipulation check in the literature, we will similarly treat it as so. The self-designed measure of performance insecurities was originally pre-registered as manipulation check. Since this is an uncommon and in itself biased measure of verifying stereotype threat, we refrain from using it, but report the measure in the online supplementary materials of study 2.