Abstract
When frontline workers interact with citizens, they often need to categorize individuals and decide who receives what and how. Previous research highlights that some frontline workers rely on negative stereotypes when evaluating certain social groups, thereby perpetuating inequalities. However, studies also suggest that not all frontline workers rely on stereotypes; some are able to suppress these biases during their interactions with citizens. The reasons behind this variation in the use or suppression of stereotypes remain underexplored in the literature. This paper aims to address this gap by proposing a framework to better understand the differences in how frontline workers use and suppress stereotypes. We argue that categorization operates along a continuum with distinct logics. On one end is stereotypical categorization, where individuals are classified as members of a specific group based on a cultural-cognitive framework that amplifies perceived differences between members of that group and others. On the other end is investigative categorization, a dynamic approach that involves forming hypotheses about individuals, testing them, and revising or reformulating these hypotheses based on new information. To illustrate these two categorization logics, we analyze the reactions of 46 Brazilian kindergarten professionals to a vignette, highlighting the theoretical and practical implications of these approaches.
Introduction
Frontline workers, also known as street-level bureaucrats, are public employees who interact directly with citizens as part of policy implementation. Their primary role is to transform policies into concrete actions and outcomes (Lipsky, 2019; Lotta, 2015). These workers regularly engage with citizens and exercise discretionary power to address individual cases (Lipsky, 2019). Because guidelines and laws cannot predict or detail all the potential situations frontline workers may encounter (Lipsky, 2019; Lotta, 2015), they must interpret these guidelines and address real-life scenarios. This involves analyzing and simplifying complex individual cases, assigning them to political and administrative categories, and determining who qualifies for benefits or is subject to sanctions (Lipsky, 2019; Lotta, 2015; Lotta and Costa, 2020; Møller, 2009).
Categorization is a daily political practice that shapes decisions about who receives what and how (Lotta and Kirschbaum, 2022; Møller, 2009) and, therefore, affect policy outcomes, eligibility, equity and equality (Raaphorst et al., 2025a). While this process is partly based on formal laws, guidelines, and eligibility standards, assessing whether a citizen meets the criteria is often ambiguous. Frontline workers frequently need to evaluate non-objective aspects of citizens to make their decisions (Lotta and Santiago, 2017; Møller, 2009). These judgments can consider a wide array of factors, including perceptions of deservingness and entitlement (Møller, 2009; Jilke and Tummers, 2018; Keulemans and Van de Walle, 2018). These perceptions may guide the prioritization of some citizens and the discrimination towards others (Dietrich et al., 2023).
Alongside political and administrative categories, some frontline workers also rely on stereotypes, such as those based on social class, gender, or race (Lotta and Kirschbaum, 2022; Harrits, 2019a; Raaphorst and Groeneveld, 2018). For instance, Oliveira and Abramowicz (2010) observed how early childhood education teachers stereotype black children and avoid physical contact with them. Since categorization determines access to rights or imposition of sanctions based on how citizens are classified (Lotta and Costa, 2020; Møller, 2009), the use of stereotypes can exacerbate existing inequalities (Millet and Croizet, 2016).
Research on stereotypes highlights variation in how frontline workers categorize citizens (Harrits, 2019a; Lotta and Kirschbaum, 2022; Maynard-Moody and Musheno, 2003; Møller, 2009; Oberfield and Incantalupo, 2021). Some workers rely on (positive or negative) stereotypes when dealing with specific social groups, while others manage to suppress them. Factors such as years of experience, professional qualifications, social group affiliation, and workplace context are linked to differences in categorization practices (Harrits, 2019b; Lotta and Kirschbaum, 2022; Maynard-Moody and Musheno, 2003). Additionally, Oberfield and Incantalupo (2021) found that the race or ethnicity of organizational leaders can influence the use of stereotypes by frontline workers. There remains a theoretical gap in understanding the reasons behind variations in the use of stereotypes and how these variations may be linked to specific characteristics of frontline workers. More empirical research is needed to address this gap. It is crucial to deepen our understanding of how frontline workers develop and apply categories to classify citizens during policy implementation (Lotta and Kirschbaum, 2022) and to investigate the underlying influences shaping these categorization processes (Møller, 2016), given that the categorization process affects policy outcomes and access to public services, occasionally leading to discrimination and unequal treatment (Amaria et al., 2025; Raaphorst et al., 2025b; Heteren et al., 2025).
The goal of this paper is to propose and illustrate a framework to better understand the variation in how frontline workers categorize citizens. We argue there is the need to better understand what constitutes a non-stereotyped categorization. Thus, this framework aims to focus not only on workers who rely on stereotypes but also on those who are able to avoid using them. Our research question is: how do different categorization logics - stereotypical and non-stereotypical - manifest among frontline workers? By more clearly characterizing the differences between these categorization logics, this study seeks to inform the development of strategies to reduce the reliance on stereotypes by frontline workers. We developed the framework based on previous literature, and then use the case of Brazilian kindergarten professionals to illustrate how it can be applied empirically. This framework contributes to the public policy and public administration literature by increasing research capacity to identify different forms of categorization logics that can affect differently how policies are delivered.
This paper is structured as follows. The theoretical section outlines the concepts of categorization and stereotypes. The next section introduces a framework for analyzing the categorization process, contrasting stereotyped categorization with investigative categorization. To demonstrate the framework’s utility, the paper examines the responses of 46 Brazilian kindergarten professionals to a qualitative vignette: 12 reflecting stereotyped categorization, 11 identified as investigative categorization, and 23 in-the-middle categorization. Finally, the conclusion discusses the theoretical contributions of the framework and suggests pathways for future research.
Literature on categorization and stereotypes
Categorization process
Categorization is a relational, social, and cognitive process through which human beings classify people or objects perceived as sharing one or more common characteristics into groups. This process helps simplify understanding and interactions with the world (Harrits and Møller, 2011; Møller, 2009). We categorize more frequently than we realize, as categorization streamlines our daily tasks. By organizing the world into categories, we make it more manageable (Harrits and Møller, 2011; Møller, 2009). However, categories do not pre-exist in nature, and categorization is not an objective process that merely reflects reality. The criteria used to group or separate entities into categories are neither predetermined nor created by a single individual. Instead, they are socially constructed, reflecting the specific characteristics of societies and varying by place and time (Yanow, 2003). These criteria highlight certain attributes of people or objects as the basis for comparison and differentiation, deciding whether they belong to the same or different categories. At the same time, other characteristics that fall outside the chosen criteria are disregarded (Yanow, 2003). Categorization thus involves ideas of comparison, belonging, and boundaries (Harrits and Møller, 2011).
Studying categorization involves examining the criteria used by classifiers to group or exclude individuals or objects (Møller, 2009). For instance, one might investigate the criteria frontline workers use to decide whether a citizen qualifies for an employment policy. Yet, categorization extends beyond practical decision-making. Categories carry implicit attributes and values that are often automatically transferred to those classified under them. These values can be positive or negative, influencing whether someone is stigmatized or admired, granted rights, or punished (Møller, 2009). Categories also construct symbolic and identity-based boundaries between individuals, facilitating processes of inclusion and exclusion (Lamont et al., 2014; Møller, 2009; Amaria et al., 2025; Raaphorst et al., 2025b). As a relational process, categorization inherently involves differentiation: inclusion in one category implies exclusion from another (Harrits and Møller, 2011).
There are various ways to examine categorization by frontline workers. One approach is to analyze it as a process, while another is to study it as a practice (Harrits and Møller, 2011; Møller, 2009). When viewed as a process, categorization encompasses the entire sequence: constructing categories, evaluating individual cases, and reasoning about where to assign each case. In contrast, categorization as a practice focuses on the outcome—the actual allocation of an individual or unit into a specific category.
Another distinction lies in the types of categories being used, which can be divided into different domains, such as political and social categories (Harrits and Møller, 2011). Political categories are formally established by policies, while social categories are informal constructs shaped by societal processes. However, these systems of categorization are not independent; they interact and influence one another. First, political categories—such as defining the beneficiaries of a policy—are often informed by pre-existing social categories. Second, frontline workers may draw on social categories to decide which individuals fit into a specific political category. Conversely, political categories can also impact social categories, either reinforcing or challenging their use. For instance, the design and implementation of political categories may validate existing social categories or work to disrupt them (Harrits and Møller, 2011).
Stereotyping as a type of categorization
This paper analyzes a specific type of categorization: stereotyped categorization. In the psychology literature, stereotypes are defined as internalized cognitive schemas that incorporate the associations and beliefs we hold about members of a particular group (Dovidio et al., 2010; Hilton and Von Hippel, 1996). Cognitive schemas are composed of sets of elements and patterns of association. These schemas function as relational structures that connect beliefs, attitudes, and behaviors, enabling individuals to make assumptions based on established relationships (Heteren et al., 2025; Lotta and Kirschbaum, 2021, 2022). For example, if a cognitive schema links A with B, but not with C, one is likely to assume B is present upon observing A, while simultaneously assuming C is absent.
When individuals make judgments based on stereotypes, they draw on a set of characteristics associated with the group they perceive the other person to belong to (Dovidio et al., 2010). Through stereotypes, people attribute specific traits to groups, supported by a network of beliefs about why these traits are linked (Hilton and Von Hippel, 1996). The concept of stereotypes is related to the concepts of heuristics and unconscious bias. Heuristics is a mental short-cut that simplifies the decision-making process. On some occasions, heuristics can lead to systematic biases, which happens when there are “systematic deviations from the optimal or rational course of action in the judgements and decision-making processes of frontline workers.” (Moseley and Thomann, 2020: 2). Some types of heuristics and biases – but not all of them – happen when there is the use of a stereotype. For instance, the representativeness heuristic happens when one interacts with the other based on relevant stereotypes about their social group (Moseley and Thomann, 2020).
The sociological concept of stereotypes incorporates a social function: distinguishing between “us” (the in-group) and “others” (the out-group) (Harrits, 2019a). In interactions, individuals often associate the “other” with a social identity, expecting them to exhibit the collective attributes believed to characterize members of that group (Møller, 2009). In other words, categorization establishes normative expectations about the “other’s” behavior, anticipating they will act as members of their identified social category are presumed to act (Goffman, 1990). One origin of stereotypes, according to Dovidio et al. (2010), is intergroup conflict. Stereotypes can serve as a tool for social groups to limit access to resources for members of other groups in situations of competition. This function is particularly advantageous for dominant groups seeking to maintain their power (Dovidio et al., 2010).
While categorization can occur in a more nuanced and individualized manner, stereotyped categorization represents a simplified application of social categories. This approach often exaggerates differences between the in-group and the stereotyped social category (Harrits, 2019a). Stereotyped categorization emphasizes a “small number of putative similarities” (Jenkins, 2014: 156), reducing the complexity of individual identity. In contrast, non-stereotyped interactions approach the “other” without assuming they possess all the stereotypical characteristics attributed to their social group. Instead, such interactions foster openness, allowing for a more individualized understanding of the person. Through direct experiences, the individual is seen as unique rather than as a representative of a broader category.
When individuals interact with members of their own in-group, they generally do so in a less stereotypical manner. Conversely, interactions with out-group members are often characterized by a perception of homogeneity, retention of less detailed and positive personal information, and a stronger belief that stereotypes about the out-group apply uniformly to all its members (Dovidio et al., 2010; Hilton and Von Hippel, 1996). This phenomenon contributes to intergroup bias, which is defined as “the systematic tendency to evaluate one’s own membership group (the in-group) or its members more favorably than a non-membership group (the out-group) or its members” (Dovidio et al., 2010: 3). Within this framework, frontline workers may exhibit a tendency to apply more negative stereotypes when dealing with out-group citizens (Webeck and Lee, 2022).
In studying social workers’ categorization practices, Møller (2009) identifies three key aspects associated with stereotypical categorization: (i) clarifications reduced to confirming tools: Clarification processes, such as tools designed to evaluate individual cases, are used not to gather new information but to reinforce pre-existing impressions; (ii) use of commonalities: Citizens are analyzed and judged based on generalized traits rather than professional or individualized explanations; (iii) normative perspectives: Judgments are made based on an implicit standard of what a “normal” citizen should be, rather than considering the individual’s unique circumstances. These aspects highlight the tendency of stereotyped categorization to prioritize preconceived notions and generalized criteria over deeper, case-specific evaluation, reinforcing biases and limiting nuanced understanding.
In contrast to stereotyped categorization, Møller (2009) proposes the concept of individualized categorization, where social workers consider the unique circumstances and external factors influencing each case. In this approach, clarification processes aim to gather more detailed information about these individual circumstances. Interestingly, Møller (2009) observed that the same worker might simultaneously apply both stereotyped and individualized logics when handling a case. Similarly, Møller (2016) differentiates categorizations based on social stereotypes from more comprehensive discretion processes, which involve a “responsive client approach” (Møller, 2016: 10).
Raaphorst and Groeneveld (2018) explored how stereotypes influence tax officials when inspecting entrepreneurs’ tax returns. They suggest that stereotypes function either as shortcuts or lenses, with the latter being more pervasive. When used as shortcuts, officials rely on observable attributes to infer unobservable characteristics, effectively linking a citizen’s appearance to presumed identities. As lenses, stereotypes shape how observable signals are interpreted, leading to the same attribute being evaluated differently depending on the citizen’s perceived social group. Their study revealed that the same signals could be differently evaluated in terms of competence and intention depending on citizen social group status.
Harrits (2019a) investigated the use of stereotypes in child care, education, and health care, finding that teachers and nurses often use stereotypes as tools to interpret situations and individuals, simplifying complex cases. Harrits identifies two distinct stages in stereotype usage by frontline workers: activation and application. Activation occurs when stereotypes influence reasoning about decisions or actions, an uncontrolled process due to the deeply internalized nature of stereotypes. Application, however, involves consciously basing decisions or actions on stereotypes, which can potentially be controlled and avoided. Harrits emphasizes that while activation is challenging to suppress, application can be mitigated through deliberate reasoning processes. Supporting this, Hilton and Von Hippel (1996) argue that proactively reflecting on stereotypes and the theories that sustain them—rather than outright suppression—can effectively reduce stereotyped reactions.
While the existing literature has extensively examined how frontline workers activate or apply stereotypes, it has largely overlooked those who manage to avoid using stereotypes altogether. This gap leaves a critical question unanswered: What constitutes non-stereotyped categorization? The characterization of what is a non-stereotyped categorization can enhance our analysis of the variation in the use of stereotypes, enabling us to better understand why and under what circumstances are some frontline workers able to avoid drawing on stereotypes. This is crucial if we aim to design policies to reduce the use of stereotypes by them. The next section introduces a framework designed to address the variation in the use of stereotypes and a better characterization of what is a non-stereotyped categorization.
Proposing a framework of logics of categorization: Stereotyped and investigative categorizations
Building on the literature discussed above, we developed a framework of logics of categorization. We propose analyzing stereotyped categorization not as a single act but as a dynamic process. Upon first encountering someone, we instinctively form an impression based on readily observable characteristics, such as appearance, voice, and other noticeable traits. This initial impression often leads us to associate the person with a specific social group, along with the attributes commonly tied to that group. The less familiar we are with the social group, the more likely we are to perceive its members as homogeneous and to ascribe to them all the stereotypical characteristics associated with that group. This stage marks the activation of stereotypes, which are based on the perceived social identity of the individual.
To better understand stereotyped categorization, the concepts proposed by Goffman (1990) provide valuable insights. Goffman distinguishes between virtual social identity—the assumptions and expectations formed about an individual based on their perceived social identity—and actual social identity, which consists of the verifiable attributes that an individual possesses. The degree of discrepancy between these two identities varies and highlights the gap between perception and reality. Goffman introduces the notion of personal identity, which refers to the unique attributes and facts that distinguish an individual within their social group. A person is seen as unique when they are recognized as possessing a network of traits and experiences that cannot be fully transferred to anyone else. This uniqueness often becomes apparent in smaller social settings, where personal identification is more feasible. Interestingly, as we build a personal identification of someone, the interpretation of their social identity often shifts. In other words, the process of recognizing a person as unique may alter how their social characteristics are understood (Goffman, 1990).
When socially identifying someone, stereotypes about their social group may be activated. However, the activation of stereotypes does not necessarily dictate the interaction that follows. Many encounters extend beyond an initial moment, involving conversations, meetings, or ongoing professional and social relationships. These longer interactions provide opportunities to refine our understanding of the other person, creating a personal identification that moves beyond first impressions. During these interactions, we may reassess our beliefs about the individual and their group, realizing that the person does not uniformly exhibit all the traits stereotypically associated with their perceived social identity. Conversely, if we remain unwilling to reconsider our initial beliefs, we may persist in interacting with the individual as though they embody all the stereotypical attributes of their group. In such cases, the personal identification formed remains closely aligned with the initial impression, and the categorization process continues to be dominated by stereotypes.
This categorization process is depicted in Figure 1, where we propose a continuum of categorization logics. At one end of the spectrum lies stereotyped categorization, characterized by the classification of a citizen into a social group about which the frontline worker holds a network of cultural-cognitive beliefs. These beliefs tend to exaggerate the differences between the group’s members and non-members. In stereotyped categorization, the worker treats the citizen as if they embody all the stereotypical traits associated with their social group, without attempting to develop a more nuanced personal identification through direct interaction or accumulated experience. Even if a personal identification is formed, it remains heavily influenced by stereotypes, resulting in a biased perspective. Process of categorization.
In contrast, we propose investigative categorization, which relies on experience and operates as a continuous process of hypothesis formulation, verification, and refinement. Investigative categorization aligns with the principles of abductive logic, a reasoning approach that suggests potential explanations for observed phenomena (Simpson, 2018). Unlike deductive or inductive reasoning, abductive logic acknowledges the fluidity and uncertainty of the world, emphasizing the need for ongoing inquiry. Simpson (2018) likens abductive reasoning to the work of detectives, who gather evidence, propose plausible explanations, and then test and refine these explanations. This process combines abduction for generating hypotheses, deduction for testing them, and induction for confirming results.
Frontline workers engaging in investigative categorization adopt a similar approach. While they may start with initial hypotheses about a citizen, they remain open to revising these based on new observations and interactions. This method involves a continuous effort to verify hypotheses and incorporate fresh information, distinguishing it from stereotyped categorization. We use the term investigative categorization instead of individualized categorization, as proposed by Møller (2009), to specifically address frontline workers—such as teachers, pedagogues, and social workers—who engage in long-term relationships with the same individuals. In these cases, categorization naturally involves some degree of personalization due to sustained interactions, but the focus remains on avoiding the uncritical application of stereotypes even in these extended relationships.
Stereotyped categorization tends to view members of outgroups as homogeneous, applying stereotypes broadly across all members (Dovidio et al., 2010; Hilton and Von Hippel, 1996). Investigative categorization, on the other hand, encourages the reconsideration of initial beliefs, enabling the formation of a more accurate personal identification. It is important to note, however, that investigative categorization does not rely solely on personal interactions. It also integrates prior knowledge, past experiences, and the perceived social identity of the citizen. The distinction lies in the reasoning process: while stereotyped categorization relies primarily on internalized and taken-for-granted social category systems, investigative categorization involves a more sophisticated and reflective thought process. A crucial element of investigative categorization is the openness to test and reformulate hypotheses, breaking the path from the activation to the application of stereotypes. It means that what Harrits (2019a) calls the ‘activation of stereotype’ can integrate the investigative categorization, but the classifier reasoning blocks the way to the ‘application of the stereotype’.
We propose a continuum of categorization logics because we assume not all categorization are in one of the extremes. There can be categorizations that have elements of both stereotyped and investigative categorization. To illustrate how this framework can be applied empirically, we present and analyze here the case of Brazilian kindergarten professionals. In the next section, we briefly present the context and the data and later we present the application of the framework.
Context, data and methods
This research was part of the international collaborative project Discos (Different contexts - similar tasks: implications for public policy provision), which aims to investigate how Brazilian and Danish kindergarten professionals handle similar tasks in very different countries (Møller et al., 2023). For this paper, we used only the Brazilian data and have a different analytical focus from Discos project. In this section first we say some words about Brazilian society and stereotypes that have been found to influence the work of frontline workers, especially teachers. Then we present the fieldwork design and methods. Finally, we explain how we analyzed the data.
Context
Colonialism, the genocide of the indigenous population, and three centuries of slavery have generated in Brazil a society marked by inequalities, racism, gender violence, and other forms of discrimination (Almeida, 2020; Ribeiro, 2015). In Brazil, gender, race, and ethnicity still circumscribe social opportunities together with class (Medeiros, 2019). Scholars who analyzed the work of frontline workers in Brazil have found they commonly use negative stereotypes about poor, black, and non-traditional families (Eiró, 2019; Jaccoud and Abreu, 2019; Lotta and Kirschbaum, 2022).
Brazilian scholars have found that some teachers treat and evaluate black, poor, and male children in a worse way. For instance, teachers commonly give fewer compliments and worse grades to black students compared to white students with the same performance in the national standard assessments (Botelho et al., 2013; Cavalleiro, 2000; Oliveira and Abramowicz, 2010). Gender stereotypes are also enacted in the Brazilian educational context, such as the belief that boys are more intelligent, courageous, and aggressive, whereas girls are expected to be more organized, obedient, mature, and responsible (Finco, 2010).
These stereotypes about children and families are commonly activated by teachers when they face a child who has a low educational development (Lotta and Kirschbaum, 2022). Winter (2024) analyzed stereotypes about Brazilian children who attend public school and causal narratives that explain their low educational development. She found that there are commons stereotypes about public school children, such as that they are lazy and do not value school. Usually, students’ failure is seen related to their socioeconomic background or to their race.
Patto (2015) investigated stereotypes that Brazilian teachers have about children who have educational development challenges and fail school. Widespread theories about school failure blamed the individual (and his/her lack of competence) or the (lack of) culture as responsible for children’s failure (Patto, 2015). This blame is usually sustained by stereotypes about poor families and their social-interaction patterns, which are considered “more aggressive, negligent, uninterested in their children, erratic, addicted and immoral” (Patto, 2015: 72). Thus, educational failure is typically seen by education professionals as being caused by family, moral, or psychological issues (Patto, 2015).
Methods
For this paper, qualitative research was carried out through a multi-case study in 8 Brazilian kindergartens and data analysis, which were conducted in an integrated and interpretive approach. Kindergartens were selected to be part of the research, and 46 professionals were interviewed. The design of methods was an effort of 8 months done in collaboration with the Discos team. The data collection methods used were observation, document analysis, interviews, and application of vignettes. However, for this paper, we base ourselves only in the interviewees’ reaction to the baseline vignette, observing how the responses to the vignette can be understood using the analytical framework developed here. Thus, we focus the description on the vignette.
Vignettes are fictional stories about individuals and situations that are presented to research participants, who are asked to comment and answer some questions about these stories, such as what they would do if they were in that situation (Barter and Renold, 2000). The aim of qualitative vignettes is not to predict what would be the participants’ behaviors but to investigate how participants interpret and reason about the presented situation (Jenkins, 2014). Usually, moral dilemmas integrate the vignette stories and questions, being useful to investigate processes of meanings and interpretations that integrate people’s reasoning and how cultural norms integrate this process (Barter and Renold, 2000).
For this research, we designed a vignette with a very common situation experienced by kindergarten professionals in their routine. The vignette presents situations where a boy misbehavior and one kindergarten professional has difficulties supporting his development. When the vignette was introduced, in the middle of the interview, the interviewee was first invited to read it. Then 5 follow-up questions were done to stimulate them to reason about it. The follow-up questions asked whether they recognized the situation, what they thought the cause of the situation was, what they would do if they were the professional dealing with that boy, how they could help a colleague dealing with that situation, and if they thought the boy should be maintained in that group. The aim was to investigate the perceptions of causality and stereotypes that the interviewees enacted when they reasoned about a child who is causing problems. As suggested by Patto (2015) and Winter (2024), students’ failure is commonly interpreted by the lens of class and race stereotypes. Your colleague just had another conflict at the dining room table with a 5-year-old boy. He often fights with the other children. This time, your colleague did not manage to stop an episode in which he hit another child very hard. You know your colleague has difficulty dealing with him in group with the other children and have talked to her before about how you can understand his behavior and reactions in the kindergarten. The boy has two siblings who have previously attended kindergarten. His parents are happy to have him in kindergarten, and they participate in parenting activities and kindergarten events. Although he accepts the kindergarten’s rules and routines such as taking off his shoes when he comes in from the playground and washing his hands before eating, he is very extroverted in his behavior towards both adults and children, but according to the parents never at home. In recent months, your colleague has tried to do something about the boy’s role in the children’s group and has talked to his parents about their understanding of his behavior, but even though they cooperate with the kindergarten, there is not much progress … on the contrary… Your colleague is in doubt about what she should do with the boy…Vignette presented to kindergarten professionals
The reactions to the vignette were transcribed and analyzed with the support of NVivo software. Qualitative data analysis based the coding process. The goal of the analysis was to identify in which position of the categorization logic each interviewee’s reactions to the vignette was. With this goal and based on the proposed framework, we created analytical and theoretical categories. Analytical categories are the ones that move away from the simple description of the data and gain analytical value. Theoretical categories refer to theories.
Codes for classifying the logic of categorization.
Source: authors.
The reactions of the professionals to the vignette were coded to these six codes. As a result, for each professional reaction to the vignette, there was data on which of these elements were present in their reasoning about the vignette’s boy. The next step was to aggregate these elements with the aim of positioning each kindergarten professional reasoning in the scale of categorization logics. To do so, we created a scale of 7 points – from −3 to 3. The 3 represents the investigative categorization and the −3 is the stereotyped categorization. For each code of investigative categorization identified in the professional reasoning about the vignette, it was added 1. For each code of stereotyped categorization, it was subtracted 1.
Thus, we were able to analyze their answers to explore the logic of categorization the professionals based on the same type of situation. There was a large variation in the logic of categorization used by the 46 interviewees when reacting to the vignette. Professionals were placed in all the possible positions, from the most stereotyped to the most investigative, showing a large variation in how they applied or were able to block stereotypes (Figure 2). Interviewees’ position in the Categorization Logic Scales when reacting to the vignette.
In this paper, we picked 3 reactions to the vignette to illustrate one case of stereotyped categorization, one of investigative categorization, and one of in-the-middle categorization. Then we analyze how the six elements (Activation of stereotypes; Formulation of a single hypothesis; Disregard or skepticism toward elements of the story; Process of investigation of the case; Formulation of multiple hypotheses; Reliance on elements from the vignette’s story) are present in the reactions to the vignette of the 46 interviewees.
Illustrating the logics of categorization
The interviewees exhibited a broad spectrum of logics of categorization in their reactions to the vignettes, ranging from highly stereotyped to deeply investigative approaches. This variation highlights differences in how professionals applied or resisted stereotypes in their reasoning. The same fictitious boys presented in the vignettes were categorized in notably different ways. While 12 participants relied heavily on stereotypes to interpret the scenarios (positioned in −3 or −2 in the Categorization Logic Scale), 11 adopted a more investigative approach (positioned in 2 or 3 in the Categorization Logic Scale), demonstrating openness to gathering and analyzing further information. Another group of 23 participants presented elements of both investigative and stereotyped categorization, thus being situated in the middle of the continuum (positioned in −1, 0 or 1 in the Categorization Logic Scale). In this section, we present three contrasting examples: one that exemplifies a stereotyped reaction, one that illustrates an investigative reaction, and one that illustrate an in-the-middle reaction. These cases offer insight into the distinct ways professionals engage with and interpret the challenges presented in the vignettes, shedding light on the dynamics of categorization logic in practice. I think it is a lack of limits. Because the child that is educated at home, they know how to behave in any place. So, lack of conversation at home and limits. The child has to understand how she should behave at home, or in the grocery store, or in the church, or in the park, and everywhere. And the child who knows how to behave, they know. But the child that does not have limits, whose father does not talk with them, that does not have conversations at home, that child does not have clarity of what can be done and what can’t be done. And nowadays we see the parents just have minimal… they do not know how to educate, how to talk. They just give things to the children and do not know how to set rules. [Professional 25]Stereotyped categorization
This speech reveals three key elements characteristic of a stereotyped categorization: activation of stereotypes, formulation of a single hypothesis, and disregard or skepticism toward elements of the story. Activation of stereotypes is evident in Professional 25’s reasoning about families. The professional draws upon a stereotypical cognitive schema, assuming that families fail to educate their children, do not establish rules, and lack meaningful communication. This schema underpins a relational assumption linking family behaviors to children’s misbehavior: families nowadays do not educate properly, thus children are unable to behave appropriately. This stereotypical vision about public school children and their families is common in Brazilian society and within teachers (Patto, 2015; Lotta and Kirschbaum, 2022). This vision supports the belief that children fail because of their parents (inappropriate) behavior, and not because of problems in the school’ system itself.
Professional 25 also formulates only one hypothesis to explain the boy’s behavior, attributing it solely to the family’s supposed lack of discipline. This lack of alternative explanations underscores the dominance of the activated stereotype in shaping the professional’s reasoning. Finally, disregard for elements of the vignette further solidifies the stereotyped categorization. For instance, the professional overlooks the detail that the parents report the boy behaves well at home. By ignoring these contextual elements, the professional focuses exclusively on the social group to which the fictitious boy and his family belong. This approach privileges stereotypical beliefs over the unique, personal attributes presented in the vignette, limiting the scope of understanding and problem-solving.
When analyzing the reactions of the 12 interviewees positioned in −3 or −2 in the Categorization Logic Scale, we find that all of them activated stereotypes and formulated only one hypothesis. Analogously to professional 25, the other 11 also activated stereotypes about the family of the fictitious boy, activating a net of beliefs that connects children with misbehavior with characteristics of families, such as lack of education, non-structured, needy, conflictive, or not careful with the child. For example, professional 2 said: “His mirror is the family. Even if you want to change this child, it is too hard, because, as we know, education comes from home”. The activation of stereotypes combined with the formulation of only one hypothesis means that the only hypothesis elaborated is based in stereotypes. Thus, the non-elaboration of other hypotheses reinforces the power of the activated stereotypes in the reasoning.
11 of the 12 interviewees positioned in −3 or −2 in the Categorization Logic Scale ignored or doubted elements of the story presented in the vignette. They ignored mainly elements related with the family characteristics, for example ignoring that the boy had siblings, that the siblings had studied in the same kindergarten, or that the family was collaborating with the situation. For instance, professional 44 said “sometimes the child is the only child and then the mother is pregnant again”.
The most stereotyped reactions were the ones that gather the three elements presented here, as in the example from professional 25. One element reinforces the other. Interviewees who activated stereotypes, ignored elements of the story and elaborated only one hypothesis expressed exhibited a conviction about their stereotyped beliefs, and this conviction blocked them to elaborated more complex hypotheses. Maybe it is just a matter of time and of trying to seat, talk, and explain to the group that sometimes he becomes nervous. Maybe it also can be an issue with his speech abilities. Because at home the parents understand everything that the child says. (…) usually, when I see a child that is like that, who hits, sometimes it is related to not being able to communicate through speech. And then he is not able to hold it, because he tries to speak and nobody understands him. He tries to talk, but nobody understands; then he goes and takes things using his force. And [I would] also talk with the mother about how is the interaction with the brothers. If they hit him if he is the younger one. (…). So, I would try to investigate what is happening. I think I should not close a diagnosis. (…) Maybe make an anamnesis. How was he when he was younger? I would not close it. I think we cannot close the case because we do not know what is happening in all the contexts that he inhabits. So, I would try to understand and then make a suggestion. [Professional 15]Investigative categorization
Professional 15’s response to vignette A exhibits three key elements of investigative categorization: a proposed process of investigation, the formulation of multiple hypotheses, and reliance on elements from the vignette’s story. First, Professional 15 emphasizes the importance of conducting an in-depth investigation of the case, suggesting an anamnesis of the boy’s situation. She highlights the need for time to gather more information, reinforcing the significance of long-term interactions in forming a comprehensive understanding. However, merely proposing an investigation does not automatically indicate an investigative categorization. For such categorization to be valid, the reaction must include additional elements. Second, Professional 15 formulates multiple hypotheses to explain the boy’s behavior. She proposes several potential explanations: it could be a matter of time, a speech issue, or something related to interactions with siblings. This range of possibilities reflects a recognition of the inherent uncertainty in the case, as signaled by her repeated use of the word “maybe.” This openness to exploring and verifying multiple hypotheses signifies a high degree of flexibility in her reasoning.
Third, her reasoning is grounded in elements of the vignette, demonstrating attentiveness to the specifics of the story. She references details such as the boy’s lack of problems at home and his relationship with siblings, which allows her to create a more personalized identification of the boy’s situation, rather than relying on generalized assumptions or stereotypes. By doing so, she distinguishes him from others, acknowledging his individuality. This approach aligns closely with Simpson’s (2018) description of abduction, which involves gathering observations, formulating possible explanations, and verifying those explanations.
Professional 15’s emphasis on collecting more information about the boy’s individual circumstances and context mirrors Møller’s (2009) argument that non-stereotyped categorization relies on clarification processes to consider case. Thus, Professional 15 exemplifies the qualities of an investigative logic, leveraging both the vignette’s details and a structured, hypothesis-driven approach to understand individual circumstances and external factors in evaluating the boy’s behavior.
When analyzing the reactions of the 11 interviewees positioned in 2 or 3 in the Categorization Logic Scale, we find that 10 of them explicitly proposed a process of investigation, some of them even used the verb “to investigate”: “I think the first thing is that we have to investigate why the child is behaving like that” [professional 8]
Eight interviewees positioned in 2 or 3 in the Categorization Logic Scale formulated multiple hypothesis. These interviewees elaborated hypotheses about the family, the child, the kindergarten context, the professional (of the vignette), or the neighborhood. Usually, they did not stand on only one kind of hypotheses, suggesting they could not know an answer without knowing more about the case. Finally, 7 interviewees positioned in 2 or 3 in the Categorization Logic Scale explicitly grounded their reasoning in elements of the vignette, indicating some information about the boy behavior, the family, or the pedagogue presented in the vignette. It enabled them to create a more personal identification for the boy, distancing him from a generalist picture. It’s quite complicated for us to try to analyze and come to a conclusion. Our break time is 30 minutes. In 30 minutes it’s very difficult for you to observe… you observe, but it’s difficult to come to a conclusion. It’s easier for the teacher to try to come to a conclusion because she spends a lot more time with him than for another teacher in another classroom to come to a conclusion in half an hour. It’s happened to me, but depending on the classmate, we get into conversations, I observe a lot, we talk, sometimes I give some tips […]. There are cases and cases and it’s very difficult to reach a conclusion. So sometimes you spend months, even the classroom teacher, observing, questioning the parents, trying to reach a conclusion. But we can’t close the diagnosis because we’re not doctors after all, right? But over time, like 27 years of experience, we learn a lot, but it’s through observation. […] in this case, it seems, it’s very brief here, but it seems to be a lack of boundaries. […] Like the case I told you about, I don’t know if you remember, the little boy who spent almost the whole year observing and trying every which way. Now I’ve managed to understand that the problem was at home. I don’t know if she’s afraid to tell the reality at home. So it’s easier for her to come and say “look, everything’s fine at home, his behavior is great, there’s no fighting at home, everything’s peaceful”. We find that strange, don’t we? Why is everything peaceful at home and the opposite at school? Whenever there’s this contradiction, we’re left with immense doubt and we keep trying until we arrive at what we thought it was, that the problem really is at home. Or this child could have autism, could be hyperactive, could be something minor, because the more severe autism you notice quickly. There are various degrees. […] At first I see it as a lack of limits. Even here, they say it’s not like that at home. Actually, I doubt it because children who are like that at school can be sure that they’ll be like that at home too. As I said, parents prefer to say my child is beautiful, wonderful, my child is perfect. Like an autistic child. For a mother who brings that child to school, she sees some different things at home, but she never admits that her child has a problem. [Professional 27]In-the-middle categorization
Professional 27 reaction to the vignette has both elements of stereotyped and of investigative categorization. She activated stereotypes and was skeptical toward elements of the story (elements of stereotyped categorization), as well she elaborated multiple hypotheses and proposed a process of investigation (elements of investigative categorization).
Professional 27 activated stereotypes about families, arguing families usually do not say the truth about their children and lack giving limits. It is similar to what professional 25 said. Professional 27 was also skeptical toward elements of the story when she says “they [the parents] say it’s not like that at home. Actually, I doubt it because children who are like that at school can be sure that they’ll be like that at home too.”
By the other side, she formulated multiple hypothesis about the situation. In addition to the family, she said “this child could have autism, could be hyperactive, could be something minor”. She also proposed a process of investigation when she argued there was the need of more observation and saying it is difficult to reach a conclusion.
When analyzing the 23 interviewees positioned in the middle of the Categorization Logic Scale (−1, 0 or 1), we find that 14 of them activated stereotypes, 6 ignored or doubted elements of the story, and 8 elaborated only one hypothesis. By the other hand, they also had elements of investigative categorization: 8 of them proposed a process of investigation, 12 elaborated multiple hypotheses, and 10 based their reasoning on elements of the story. Having elements of both stereotyped and investigative categorization resulted being placed in the middle of the scale.
Discussion and conclusion
In this paper, we examined the process of categorization performed by frontline workers, emphasizing its significance due to its impact on the distribution of both symbolic and material resources. As some scholars have already found, the use of stereotypes by teachers led to a worse treatment and worse grades to certain social groups, such as black and poor (Botelho et al., 2013; Cavalleiro, 2000; Oliveira and Abramowicz, 2010; Patto, 2015). This is aligned with public administration research that shows how the use of stereotypes by frontline workers may lead to discrimination and inequalities (Amaria et al., 2025; Hetern et al., 2025; Raaphorst et al., 2025b). However, in this paper, more than analyzing the use of stereotypes and its consequences, we were interested in understanding the variation in the use of stereotypes.
Building on the existing literature that investigates these processes and concepts, we proposed that categorization operates along a continuum. At one end lies stereotyped categorization, characterized by a lack of openness to new information about the individual and limited attempts to construct a more precise personal identification. At the other end is investigative categorization, which involves an ongoing process of exploring and understanding the individual’s characteristics through deeper inquiry and analysis.
Through the analysis of Brazilian kindergarten professionals’ reactions to a vignette, we identified and characterized the elements of both logics of categorization. Stereotyped categorization is marked by the activation of stereotypes, the development of a single hypothesis, and the omission or questioning of specific elements of the story. In contrast, investigative categorization is defined by a process of inquiry into the case, the formulation of multiple hypotheses, and reasoning grounded in the details provided in the vignettes.
The proposed framework has two main strengths. First, it provides a clear characterization of non-stereotyped categorization, particularly when frontline workers engage in personal identification of citizens. The suppression of stereotypes by frontline workers still overlooked by the literature on public policy and public administration (Harrits, 2019b) and offers an innovative perspective. This framework enables scholars to understand how stereotypes are blocked, leading to a deeper understanding of the mechanisms at play and, therefore, to understand the categorization process in a more nuanced way. Second, it enables the identification of variations in frontline workers’ logics of categorization, paving the way for further research to explore the factors associated with these differing approaches and how they affect policy implementation. Future research should advance in this agenda analyzing what can explain the variation in the logic of categorization used by each professional and under what conditions frontline workers tend to be more investigative.
As others, the paper has limitations. First, it is originated by a contextualized analysis about the categorization logic used by a limited number of interviewees: 46 frontline workers. The context where the interviewees worked matters. They were kindergarten professionals working in Brazilian society, thus all they are part of one profession working in a specific country. Future research could test the framework with other professionals working in different contexts. Second, the proposed framework was elaborated mainly to understand categorization process that are built in continuous interactions. It would be harder to apply it to situations where the frontline worker has to take instant decisions based on their first picture of the situation, as is the case of some of the situations lived by the police. Finally, the paper applied the logic of categorization framework to interviewees reaction to a vignette. Future studies can use it to analyze real-world observations of how frontline workers categorize citizens.
The characterization of the logics of categorization provides practical contributions. First, frontline workers and managers can better understand how categorization and stereotypes work by getting to know the two logics of categorization: stereotyped and investigative. The empirical illustration of the variation in logic of categorization used by professionals highlighted there is an important variation in how frontline workers use stereotypes. If there is variation, with some being more investigative, it means that organizations can implement measures to change the use of stereotyped categorization by the others. This can be enacted by managerial practices of education, reflection, incentives (Harrits, 2019a), among others, that can improve how frontline workers categorize citizens. If the public administration wants frontline workers to use less the stereotyped categorization and more the investigative one, organizations should propose practices and professional qualification to enhance the process of investigation, formulation of multiple hypotheses, and better observation of the situations, understanding that these elements affect access to public services and inequalities.
Footnotes
Acknowledgments
We thank the support of Marie Østergaard Møller and Michael Hill on the development of the research.
Ethical considerations
This study was approved by Comitê de conformidade ética em pesquisas envolvendo seres humanos – CEPH/FGV (Parecer n. 321/2022).
Consent to participate
All the interviewees signed informed consent statements.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was financed, in part, by the São Paulo Research Foundation (FAPESP), Brasil [Process Number 2013/07616-7, 2017/24750–0, 2019/13439–7, 2019/24495–5 2020/15430-4, 2025/06737-2]; by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES) [Finance Code 001]; and CNPQ [305180/2018–5].
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
Data is not available to maintain the confidentiality of the interviewees.