What influences the behavior of street-level bureaucrats (SLBs) in the Brazilian Social Assistance System (SUAS)?

Abstract:

This article analyzes three groups of factors that influence the practices and public categorization schemes of street-level bureaucrats (SLBs) in the Service for the Protection and Care of Families and Individuals: (i) professional and relational factors of SLBs; (ii) organizational factors; and (iii) client characteristics. The research studied 11 SLBs working in Belo Horizonte, employing intensive observation of daily routines and services provided, interviews, and analysis of follow-up records, coded using Atlas.ti. The findings suggest that individual characteristics of SLBs and organizational aspects are less significant than client attributes in shaping street-level work. Client demands influence referrals and assistance, while SLBs’ motivation and capacity to modify behavior impact user categorization. These attributes generate unique cases, leading to personalized actions and a flexible, adaptive service implementation.

Keywords:
policy public implementation; street-level bureaucracy; influence factors; Brazilian social assistance system

Resumo:

O artigo analisa três grupos de fatores que influenciam as práticas e a construção de esquemas de categorização do público por burocratas de nível de rua (BNR) do Serviço de Proteção e Atendimento Integral à Família e Indivíduos: i) características profissionais e relacionais dos BNR; ii) fatores organizacionais; iii) características dos usuários. A pesquisa envolveu 11 BNR atuantes em Belo Horizonte, com observação intensiva da rotina e de atendimentos, entrevistas e análise de Prontuários de Acompanhamento, codificados no Atlas.Ti. Concluiu-se que as características individuais dos BNR e os aspectos organizacionais são menos relevantes que os atributos dos usuários para a atuação no nível de rua. Demandas impactam encaminhamentos e atendimentos, a motivação e a capacidade de modificar condutas influenciam a categorização de tipos de usuários. Tais atributos geram casos únicos, resultando em atuações personalizadas e na implementação flexível e adaptativa do serviço.

Palavras-chave:
implementação de políticas públicas; burocratas de nível de rua; fatores de influência; Sistema Único de Assistência Social

Introduction

This article discusses three groups of factors that influence the actions of street-level bureaucrats (SLBs) in the implementation processes of Serviço de Atendimento Especializado a Famílias e Indivíduos (Paefi) (Specialized Care Service for Families and Individuals), within the Sistema Único de Assistência Social (SUAS) (Brazilian Unified Social Assistance System). The factors analyzed include the professional and relational attributes of the street-level bureaucrats studied, the characteristics of the users served, and the elements of the organizational work context of these bureaucrats.

Since the 1970s, public policy implementation studies have been interested in describing and analyzing how interventions work in practice and what influences their functioning and outcomes. Certain approaches argue that, even when decision-makers clearly establish implementation norms, there are factors that interfere in the process, such as local institutional contexts and the actions and relationships of the actors and organizations involved. In this way, the political negotiation and decision-making process inherent to public policy production does not stop at the design stage but extends into the implementation phase. This literature conceptualizes implementation as an ongoing process of interaction and negotiation between different actors, leading to the constant adaptation of policy content, or a certain “negotiated order” (Barrett, 2004). The original policy is transformed during execution, and this transformation results from the interactions between different actors embedded in institutional arrangements and personal and professional relationship structures, which can either limit or expand the actions of such actors.

With these premises, street-level bureaucrats (Lipsky, 2019), those who directly deliver services in interactions with citizens, distributing benefits and sanctions, inevitably become central variables in the analysis of the political process. The study of these bureaucrats has become a specific field within public policy analysis, aiming to uncover how they act as agents in implementation processes, making decisions and judgments to deal with uncertainties and inadequate working conditions. The behavior of street-level bureaucrats is characterized by discretion, as they make allocative decisions in their daily encounters with users to deliver goods and services, interpreting and classifying their conditions and needs. Discretion can lead to variations in the implementation process and potential effects on users’ living conditions. Therefore, this literature suggests that the decisions made by street-level bureaucrats, their routines, and their coping strategies in response to the dynamic work context concretely shape the public services implemented (Brodkin, 2011; Hupe; Hill, 2007; Lipsky, 2019; Lotta, 2019).

This is the theme of the article - the practice of public policy implementers - specifically in social assistance services defined by the Brazilian Unified Social Assistance System. The analysis presented here utilizes results from empirical research on the Paefi. Aimed at families experiencing rights violations, Paefi provides social assistance support based on practices of building bonds and providing guidance between street-level bureaucrats and the public, which, in theory, requires an approach with some degree of discretion and judgment. The research focused on the actions of SLBs within the service, analyzing two interrelated dimensions: the practices (both expected and unforeseen daily activities) and the categorization schemes of the served public (unofficial classifications based on social stereotypes), identifying the factors that influence the actions of the implementer agents.

The Paefi exists to respond to complex problems, often marked by family and community tensions, which can lead to the weakening or even breaking of bonds. These are multicausal social phenomena that encompass economic, social, cultural, and subjective aspects, positioning the service within the specialized level of social protection in the SUAS. The service is a mandatory provision of the Centro Especializado de Assistência Social (CREAS) (Specialized Social Assistance Center), which includes public or private physical units responsible for delivering the system’s services, including social assistance services (Brasil, 2011, 2012, 2014).

It is implemented as follows: cases involving families with individuals experiencing violations of relational rights (such as neglect, abandonment, violence, child labor, etc.) are referred by other SUAS services, health or education policies, guardianship councils, rights councils, and justice system defense bodies (such as defense attorneys and the Public Prosecutor’s Office). These cases are directed to street-level bureaucrats, social workers, and psychologists, who must assess the eligibility of the case and establish a support plan with the family. This plan includes various specialized care practices, such as counseling sessions, providing general guidance and socio-legal advice, home visits, referrals to other services, and art-education workshops.

In addition to sufficient human resources and time to ensure close and consistent follow-up, the implementation of Paefi depends on an organizational structure based on guidelines for professional practice, workflows, and the availability of resources and services from other policies. However, its primary input is the listening, time, and interaction of the street-level bureaucrat who leads the follow-up with the family.

The street-level agents of Paefi also depend on coordination with the benchmark social assistance³ network of CREAS and with agents from income transfer policies, health, education, housing, labor, employment, and others (Brasil, 2011, 2012, 2014; Belo Horizonte, 2020). As the literature on street-level bureaucrats has revealed (Hupe; Hill, 2007; Lipsky, 2019; Lotta, 2015; Meyers; Vorsanger, 2003), these organizational and relational conditions also have implications for the actions of SLBs in the service, influencing their implementation practices.

The actions of SLBs are also influenced by their own characteristics, despite the expectation of the ideal typical conception of bureaucracy of an impersonal, rule following order (Weber, 2000, 2011). Empirical studies reveal that SLBs also guide their practices by moral judgments, based on social stereotypes (Maynard-Moody; Musheno, 2003), as well as by professional logic, knowledge, and experience (Harrits; Cecchini, 2019).

This is no different for the street-level bureaucrats of Paefi, given, on one hand, their focus on situations of rights violations, which are surrounded by a set of representations and stereotypes with meanings associated with criminalization, punishment, and blame (Carmona, 2019). These stereotypes expand the space for SLBs to make judgments, categorizing the users they assist. On the other hand, the professional practice of psychologists and social workers in SUAS is of well-known prominence, not only because they are mandatory members of the reference teams for social assistance services (Brasil, 2006). Also, because their professional councils have published guidelines for their roles in the policy, encouraging the particularities of each profession’s intervention in the service (Brasil, 2016; CFESS, 2011; CFP; CONPAS, 2016). Consequently, it is assumed that these professional guidelines also affect the implementation practices adopted by the SLBs in Paefi.

Considering these assumptions: i) about the actions of street-level bureaucrats in the implementation process; and ii) about the attributes of Paefi that expand the space for the discretionary exercise of these agents; this article presents a subset of the results from the research. As mentioned, its aim is to analyze three sets of factors that influence the practices carried out and the categorization schemes of the public constructed by the implementer agents of Paefi. The factors are:

1. Individual characteristics of SLBs: professional (profession and specialized knowledge, career trajectory, and job satisfaction) and relational (engagement of agents in social groups and relationships with peers, management, and the service network);

2. Characteristics of the users: personal attributes, demands, behaviors, motivation towards the service, and actions; and

3. Organizational context of service implementation: guidelines and regulations about the service, general working conditions, and the local implementation arrangements.

The question guiding this article is: to what extent and in what ways does the professional knowledge of the street-level bureaucrats in Paefi, the relationships they establish at work, and the profile of the users they assist influence their decisions on which practices to implement (such as conducting a home visit, referring to the health center, or requesting legal advice)? Additionally, how do they classify users (e.g., adherent-resistant or organized-disorganized) in the service implementation process?

The analysis presented is based on an interpretative case study of a team composed of 11 street-level bureaucrats working in the service in Belo Horizonte. Data collection was triangulated using techniques such as intensive direct observation of the implementation routine - work meetings, in-person and remote services, and collective activities - semi-structured interviews with bureaucrats, managers, and users, as well as administrative records from the Case Monitoring Files. All this material was systematized using the coding procedure in Atlas.ti, following the guidelines for thematic qualitative data analysis (Domenico, 2020).

The focus of the analysis of the factors influencing the actions of SLBs in Paefi is not to explain individual variation in the actions of the agents studied, but rather to observe how these factors affect their actions from a more aggregated perspective. This is because the research concerns a single case with a small number of bureaucrats in the reference team, which already presents challenges in identifying significant or representative variations in their behavior.

Even with this caveat, an attempt was made to conduct an individualized analysis, and it was indeed found to be inadequate for evaluating possible variations. This is because the profile of the agents studied revealed that some professional factors (such as functional status, prior experience before Paefi) and relational factors (such as relational engagement and perceptions about the intensity, purpose, and evaluation of interorganizational relationships) are quite homogeneous among them. If these factors do not vary, they cannot be associated with individual differences in performance. It was also already known that organizational factors are common to all the agents studied, given that the research focuses on a single CREAS.

In any case, by addressing the factors that influence the actions of public policy implementers and highlighting the importance of user characteristics in decisions and classifications, the main contribution of the empirical study presented in this article to the literature in the field is to show that services whose implementation depends on the active participation and behavioral changes of users - such as Paefi - are, at the front line, co-produced through the interaction between street-level bureaucrats and citizens.

In addition to this introduction and the concluding remarks, the article is divided into three sections: i) arguments from the literature on the factors influencing the actions of SLBs; ii) methodological note; and iii) analysis of the actions of Paefi SLBs, considering the practices performed, the categorization schemes of the public, and, finally, an examination of the extent to which, and the way in which, the three sets of factors influence their actions, either separately or together.

SLBs behavior and influencing factors

The conceptualization of street-level bureaucracy by Lipsky (2019) refers to work in the public sector limited to a few typical situations: i) constant face-to-face interaction with citizens in the work routine; ii) the existence of discretion to exercise authority; and iii) the potential effects of the work on citizens are quite extensive. For example, street-level bureaucrats include security agents in police stations, military police officers on their rounds, in vehicles, or at community bases, teachers in classrooms, health professionals in basic care units or family health programs, reference technicians in Social Assistance Centers, and judges and public defenders in courts.

These functionaries operate under conflicting working conditions, characterized by scarce resources, ambiguous objectives and targets, insufficient performance measures, and cross-pressures from managers and users. To navigate these conflicts and pressures, they discipline and categorize users, distribute benefits, and structure the context of interaction, using discretion to implement policies. They also create coping mechanisms to manage these pressures, such as breaking rules, using personal resources, bureaucratizing care, prioritizing easier cases, or directing their practices toward user demands in search of job satisfaction (Brodkin, 2011; Hupe; Hill, 2007; Lipsky, 2019; Lotta, 2019; Tummers et al., 2015).

The discretion of street-level bureaucrats has a more structural or delegated dimension in reference to the rules, which characterizes the existing space to interpret them, decide and act, and a more practical dimension, which expresses what street-level bureaucracies actually do (Hupe, 2013; Lotta; Santiago, 2017). It can also be understood as agency, from a sociological perspective, referring to the typically human ability to use language to form judgments and act (Maynard-Moody; Musheno, 2012).

This overview of the work and discretionary exercise of street-level bureaucracies allows their actions to be analyzed in three dimensions: practices, styles of interaction and processes of categorizing the public. Together, these dimensions shape what the agents actually do in implementation and analytically operationalize the core tasks of their work, in other words, the SLB act by constructing practices within interactive contexts and using categories to classify the public. Practices (of a “harder” nature), interaction styles and social categorization schemes of the public (of a “softer” nature) (Hoff, 2013; Nunes; Lotta, 2019) are interconnected dimensions of the SLB action and are influenced by different factors.

This occurs because the analysis of public policy implementation assumes that street-level bureaucracies are “open systems”, permeable to environmental and external pressures in competition with the internal pressures and values of the bureaucracy (Lotta, 2019). In addition to identifying and describing its dimensions, understanding the work of street-level bureaucracies must also take into account the different factors that influence it or the various sources of discretion, since no single factor can sufficiently explain this exercise. Some of these studies focus on the dynamics that explain the actions of the implementing agents (Hupe; Hill, 2007; Gofen; Sella; Gassner, 2019; Lotta, 2019; Meyers; Vorsanger, 2003).

The literature has already made significant progress in understanding the factors that influence discretion and has identified some sources: i) the characteristics of the worker themselves, such as personal attributes, professional trajectory, personality traits or individual values and beliefs; ii) sociocultural and behavioral attributes of users and their social value; iii) organizational aspects, such as rules, normative procedures and training, incentive mechanisms, working conditions; iv) the nature of the policy in terms of ambiguity and conflict of objectives, instruments and goals; v) codes and norms of the SLB profession, among others (Gofen; Sella; Gassner, 2019). It is interesting to review here the factors analyzed for the Paefi case: the characteristics of the bureaucrats, the characteristics of the users and the organizational aspects.

Personal attributes, values, relationships and professions of SLBs

Those studies that analyze the relationship between the personal characteristics of implementing agents and their performance emphasize personal attributes. Keiser (2010) found that, in order to elect disability pension beneficiaries in the United States of America (USA), SLBs are influenced by their personality traits, such as speed of service and adherence to the agency’s goals and mission. In the same vein, Nielsen (2006) found that gender affects the performance of street-level agents in six Danish public professions, showing that women follow more rules than their male colleagues in the same roles.

Meanwhile, individual and political values and beliefs affect the actions of SLB, especially when they define the social status of the public (Lipsky, 2019), the character and identity of the user, or construct what the literature calls public categorization schemes (Møller; Harrits, 2013; Møller, 2016). Street agents use racial, geographical and other grammatical qualifiers that intensify certain characteristics and behaviors or attenuate others, segmenting the public based on social values and stereotypes about what is appropriate, normal or deserved (Maynard-Moody; Musheno, 2003).

In this regard, the work done at the street level is necessarily relational and the multiple relationships that the agents establish on a daily basis influence their own actions (Lipsky, 2019). They act as “co-decision-makers”, based on the networks of relationships in which they are inserted (Hupe; Hill, 2007). On this theme, Maroulis (2017) found that the more extensive the networks and social ties within in an American urban high school, the more innovative the practices of the teachers. In the same country, Siciliano (2017) also observes that teachers invest time in relationships with colleagues to seek advice, access resources and information and improve classroom practices. Gale and others (2017) observed that health professionals negotiate their decisions with other colleagues and with the communities in which they work to adapt the intervention to the local area, acting with what they call “street-level diplomacy”. But in addition to these inter-organizational relationships, the different memberships, affiliations and social transits of street agents in institutions, groups and networks are other important relationships for their discretionary action (Lotta, 2015), but they are still little researched (Lotta; Marques, 2020).

In conclusion, professional factors are another characteristic of SLB considered by studies to influence their behavior. Møller (2016) pointed out that the less-than-pure professionalism of street agents affects their working reasoning, leading them to act either organizationally (with reduced discretion) or occupationally (with comprehensive discretion). Harrits and Cecchini (2019) found that teachers act according to technical knowledge and previous experience to interpret the situation and define their practices. Evans (2013) identified that social workers continually activate professionalism as a commitment to users’ needs and freedom at work, limiting managerial control. Pires and Santos (2021) found that, even though they are part of a multi-professional team in the same health service focused on harm reduction, psychologists and social workers understand and carry out their work with an “expanded scope”, while occupational therapists, nursing technicians and doctors do so under the perception of a “path to abstinence”.

Therefore, a wide range of SLB characteristics can influence the exercise of discretion at street level. Individual aspects, such as gender and personality, which are more in line with organizational objectives, have already been mapped out in the literature, as have the agents’ values and stereotypes of normality and worthiness, their inter-organizational relationships with peers and the service network, as well as their professional knowledge and training.

User attributes

Since the early formulations on the work of SLBs, the relevance of attributes and relationships with users for the performance of these agents has been highlighted (Lipsky, 2019). Bureaucrats’ perceptions of user value are fundamental to implementation efforts (Harrits; Møller, 2014; Maynard-Moody; Musheno, 2003). This influence seems to be in a chain: the actual characteristics of users influence the judgment of the SLB, which related to their performance. Sociocultural characteristics, such as class (Harrits, 2018), race and ethnicity (Bundgaard; Nørgaard, 2018), and behavioral qualities, such as user motivation and performance (Guul; Pedersen; Petersen, 2021; Møller, 2016; Tummers, 2017), are generally considered in studies as sources of frontline discretion (Gofen; Sella; Gassner, 2019).

Harrits (2018) found that teachers modify their concern and attention to the student’s problem when the class condition varies, whether it is low or high. Bundgaard and Nørgaard (2018) observed that health professionals act differently with native users compared to foreign users, evidencing that it is influenced by the ethnicity of the users. Guul, Pedersen and Petersen (2021) found that the attention and efforts of social workers vary depending on the motivation and performance of the public. Møller (2016) confirms that the perception of the motivation of unemployed people applying for benefits is one of the judgments made by social workers in constructing their worthiness.

Despite advances in research into these relationships, the role of users in frontline work has been little developed theoretically and little estimated empirically (Guul; Pedersen; Petersen, 2021), and a consensus has yet to be reached on the social representations involved in these judgments, especially on who is the deserving, easy and difficult user (Gofen; Sella; Gassner, 2019).

Organizational aspects

Implementing agents work in street-level facilities, where the intervention is carried out with a certain amount of autonomy in carrying out the tasks and difficulty in direct management (Gofen; Sella; Gassner, 2019). Characteristics of these organizations regulate or delimit the action repertoires of implementation workers and influence the choice of courses of action within the present conditions (Lotta, 2015).

Organizational factors can be diverse. One of them is the uncertain and conflictive working conditions common on the front line, with ambiguous, contradictory (Harrits; Møller, 2014; Hupe; Hill, 2007; Lipsky, 2019) or excessive organizational rules (Evans, 2013); insufficient resources (Hupe; Buffat, 2014); informational uncertainties (Raaphorst, 2018) and an overload of work and cases to attend to (Brodkin, 2011; Lipsky, 2019; Thomann, 2015). For example, the lack of training and the perception of insufficient resources affect the exercise of discretion by health professionals (Lima; D’Ascenzi, 2017).

Normative and orientation documents that guide implementation often fail to clearly what street agents should do, leading them to look for mechanisms to simplify their work. The ambiguity of these guidelines can influence what is done at street level (Hupe; Hill, 2007; Lipsky, 2019). The level of conflict and ambiguity of the objectives and means of a policy can lead to different types of implementation (Matland, 1995), and the greater or lesser programmability of tasks and the need for interaction with users are important in defining the dynamics of implementation (Nogueira, 1998).

Management elements, such as the level of control and centralization of the manager and his or her more or less flexible way of managing (Evans, 2013) and the adoption of performance incentive mechanisms (Brodkin, 2011), are also organizational factors that influence the action of SLBs. For example, in healthcare facilities with greater control over work, where rules are more applied, teams work together and managers monitor agents closely, there is less room for discretion and less variation in practices (Lotta, 2015; Lotta; Marques, 2020).

As such, the organizational configurations of work at street level limit the possibilities for action by SLBs, leading them to make choices, to carry out their practices in different ways and with different levels of performance. In each implementation context, there will be specific organizational factors that can increase or reduce the exercise and space for discretion. The Box 1 systematizes the various factors that influence the performance of the SLBs reviewed.

Box 1
Factors associated with the performance of SLB mapped in the literature reviewed

Associated factors	Synthesis
SLB characteristics	Personal attributes (gender and personality traits) Personal and social values and beliefs (stereotypes of normality and worthiness) Relational factors (peer relationships, size and structure of inter-organizational networks and personal social networks) Professional factors (knowledge and career paths)
User characteristics	Personal attributes (race, ethnicity, class) Demands Motivation and performance in service
Organizational aspects	Working conditions (availability of resources and training, work overload, informational uncertainties) Implementation arrangements (rules, means and ends) Managerial elements (level of centralization and management control, performance mechanisms)

• Source: Prepared by the author.

Methodological note

The data used in this research is the result of eight months of fieldwork in one of the nine CREAS units in Belo Horizonte between 2021 and 2022. The case selection used the Índice de Desenvolvimento das Unidades CREAS (Idcreas) (CREAS Unit Development Index) from 2017, the last year available at the time of this stage of the research. The Index assesses the quality of the management of these facilities in terms of physical structure, services and human resources, on a scale of one (furthest from ideal) to five (closest to ideal). The average Idcreas score for the nine CREAS in Belo Horizonte in 2017 was 3.26. The one chosen would be the one closest to the average, avoiding biases related to good or poor performance in the indicator, but two facilities had the same score (3.17). A draw was made between them, finally arriving at the selected⁴ CREAS, where 13 technicians were part of the Paefi reference team and 11 agreed to take part in the survey.

The fieldwork involved 11 interviews with SLB,⁵ 3 with service managers and 2 with users, as well as the systematization in field notes of 13 user consultations and 12 directly⁶ observed team meetings. All this material was systematized and coded in Atlas.ti. In addition, we used data from 29 case records available in the CREAS archive up until the cut-off date for data collection.⁷ Administrative records from the Belo Horizonte Sistema de Gestão de Políticas Sociais (SIGPS) (City Hall’s Social Policy Management System) were also used. These reports and documents were read and typed into an Excel file, highlighting the main characteristics of the monitoring carried out, in the light of the research questions.

The data collected was systematized using the coding procedure described by Domenico (2020). Initially, an exhaustive reading of the files was carried out in the software, identifying words, expressions, terms or segments of text related to the factors of influence summarized in the previous literature review (Box 1), but also looking for other aspects based on the case researched. At the second level of coding, the initial codes were grouped into categories by thematic affinity in relation to the type of influencing factor (e.g. individual factors, professional factors, organizational conditions, routines, flows and resources, information on the public served). At the third level, the groups of codes were brought together in the conceptual categories “characteristics of Paefi agents”, “organizational aspects of CREAS” and “description of the users served”. The subsequent analysis focused on the citations in these conceptual categories, identifying and interpreting patterns of associations between the factors and the work of the Paefi technicians. It is worth noting that, due to the type of case study, the analyses are only valid for the context of the case researched.

The following elements and attributes were considered as characteristics of the technicians: i) relational factors: the agents’ engagement in social groups and their inter-organizational relationships, teamwork, relationships with peers, managers and the social assistance network; ii) professional factors: the technicians’ profession, their working codes and logics, career path and motivation. The relational aspects were identified through direct observation of the day-to-day implementation and through interviews with specific questions on the subject. Professional aspects were also identified through questions about the professional background and motivation of the workers and the interpretation of case reports narrated by the technicians.

The users’ characteristics, which refer to their personal attributes, their needs and demands, and their behaviors, such as their motivation towards the service and their ability to overcome the violation, were identified in the case reports in the Follow-up Records and in the interviews with the technicians and users.

The organizational aspects refer to the guidelines and regulations about the service, the general working conditions and the local arrangements for implementing Paefi. They were identified especially in the data from the observations of the weekly meetings of the reference team and in the interviews, in questions about the challenges of implementing the service and in spontaneous case reports.

The influence of the three sets of factors on the practices carried out and the categorization schemes constructed by the techniques was interpreted on the basis of the data collected. By mobilizing the literature and considering that the nature of the service provided by Paefi, based on intense employee/citizen interaction, leads to the need to build specific assistance alternatives for each case, with the aim of interrupting situations of rights violation and changing behavior, this analysis started from the following assumption: the practices implemented and the public categorization schemes built by the SLB in the service are influenced by the three sets of factors, highlighting the complexity involved in the discretionary exercise of these agents on the front line.

Data analysis

In this section, the findings of the research will be presented, first in relation to the performance of the SLB through the practices carried out and the public categorization schemes constructed, and then in relation to the factors that influence the more general performance of the agents in the Paefi implementation process.

How do Paefi technicians carry out their practices?

One of the constituent elements of street-level bureaucracies are the practices, activities that street-level bureaucrats carry out, which can indicate their own routine ways of doing things, or of exercising their discretion, managing the resources at their disposal and interpreting policy rules and guidelines. These habitual patterns of carrying out tasks are, for bureaucrats, the most appropriate - or possible - way of getting the job done within the constraints (Hupe; Hill, 2007; Lipsky, 2019).

The 25 practices foreseen for the implementation of Paefi, mapped out in the regulations and guidelines, are reception with qualified listening and interdisciplinary assessment of the case, individual and collective care (art-education workshops), social assistance follow-up (visits, remote and face-to-face care, socio-legal guidance, referrals), coordination with the service network and discussion of cases (Brasil, 2011, 2016; Belo Horizonte, 2020). When it was observed that these occurred in the actual implementation of the Paefi survey, it was noted that the SLB’s work is in line with the regulations.

Of the 25 sub-practices planned, only eight were not mentioned in interviews, field observations or medical records. Despite this, the normative structure of Paefi does not eliminate the discretion of the technicians, who use the reduced but existing space, choosing what not to do (such as collective activities) or interpreting the situations and characteristics of the case in order to build a way of monitoring.

They combine different planned practices, starting with the welcome, when they listen to problems and needs in order to agree on a follow-up plan with the families. They seem to make up the process with three of the most common practices: i) appointments - a set of questions to check on the user’s condition and compliance with the agreements; ii) home visits - to broaden the view of the family and the home dynamics; and iii) referrals - Paefi’s objective offer which refers to the destination of families to other public services (school enrollment, health care; legal proceedings at the Rights Defense Center; internships and apprenticeship programs, etc.). These practices are also discussed and developed with the network, during coordination activities, and with the local coordination team, during case discussion meetings. Having carried out a set of activities and interpreted “the extent” to which the family has been able to respond to the process, the technicians move on to a final stage, when they suggest closing the case, usually at the case management meeting with the coordinator. The following quotes highlight these findings:

I’m going to take the script there, what I’m going to implement for this family. You did it, didn’t you? And the next step, then I’ll agree with them, “look, we’ve already done this, you’ve already done this, this is [sic] missing, so let’s think about when you can do it”. And then I’ll propose to the family what we can think of for this follow-up (Interview 02).

And then we do the follow-up, which is the visit, the individual appointments, within our follow-up, within Paefi’s reality, the contact with the network is much greater than the contact with the family itself. So, we need to do this very thoroughly (Interview 05).

Follow-up isn’t just about care, it’s articulated with the network, with other public policies, it’s a visit (Interview 06).

Although they are guided by the service’s guidelines, the technicians develop their own “way of doing things”, adjusting the list of activities to the needs observed. This reflects Hupe’s (2013) legal view of discretion, where the interpretation of norms guides the choice of the most relevant actions to implement a public policy.

Two unforeseen practices were also identified: “investigating” and “responding” to cases. This is how the SLB name, respectively, the practices of investigating cases with insufficient information and providing information and testimonies on the cases they monitor to the justice system bodies. Although the SUAS protocol states that investigating complaints is not the responsibility of the technicians and that their participation in hearings for judicial purposes is forbidden, the SLB describe such activities as outside the scope of the social assistance technician and, therefore, denote unplanned practices, in the light of the following evidence:

It’s social assistance, it’s not our role to investigate. Sometimes I think I’m a civil police officer, because of this subpoena, this “it’s not to investigate, but you have to go and see if it’s really happening so you can investigate (Interview 06).

And they call us to answer about reports, and we have almost no support from the city hall, so we go there to answer about it [...] including answers that aren’t ours. We have [the central body for relations with the SGD] to filter this out, to deal with it, but we receive cases with no name so we can deal with them. So, I don’t know to what extent this is working, you know (Interview 05).

Therefore, the SLB of the Paefi research largely adhered to the list of planned activities. The few choices and adaptations and the common “roadmap” for carrying out the accompaniment express some use of the space for discretion, but their practices tend more towards standardization than divergence.

What user categorization schemes do the techniques mobilize?

As mentioned, the categorization of the public by the SLB is one of its central tasks (Lotta, 2019) to control conflicts over working conditions. The official categories establish a categorization within the framework of administrative practices and reflect eligibility criteria. However, the social dimension of this process of ordering, comparing and allocating citizens into administrative categories of access is not automatic and unequivocal, as it leaves open a space for the creation of other categories, based on particular criteria adopted by bureaucrats in the face of flesh and blood users with specific situations. Therefore, the categorization of the target public of public policies is the result of a political or normative decision, but not only that, since the moral judgments of the SLB themselves are reproduced in the interaction with the public and (re)construct the limits of the formal categories (Maynard-Moody; Musheno, 2003).

The work of the Paefi technicians, analyzed using categorization schemes, showed that when they encounter real people and situations that fall into the administrative categories, they trigger two behavioral categorization schemes: that of the “resistant-adherent” user (already identified in related studies); and that of the “organized-disorganized” user, which has not yet been observed in studies on SLB.

The behavioral expectation that the family will actively participate in Paefi and, what’s more, that they will co-produce it with the techniques, agreeing and consenting to each step of the follow-up plan, is considered in order to classify them as more “adherent” or “resistant”. The former are those who create bonds, comply with referrals and connect with the SLB. Those who are resistant are seen as suspicious, aggressive, don’t establish a bond and don’t open up, and are considered “difficult” to work with. Therefore, motivation and commitment to the service are the attributes of users judged for these categorizations.

The second scheme segments the public based on the judgment of their behavior in relation to their own condition of violation. They are idiomatically categorized by the techniques as “organized” or “disorganized”. The former behaves actively towards their conditions, becoming aware of their actions, changing certain attitudes or taking steps to become protective and overcome the violation. Some of the attitudes of the “organized” are more objective: “arranging a special day” for the child to go to school, “taking the elderly out for a walk”, “listening” to the family, “breaking off the abusive relationship”, “having rules” for living together. However, there are less concrete behavioral descriptions, such as “finding ways” to manage care, “making a move”, “being able to understand” and making a “break in this cycle of violation”. The “disorganized” user, on the other hand, acts differently: “doesn’t talk to anyone” at home, “doesn’t do anyone any favors”, “can’t ask a question”. Thus, this classification is used to represent users in terms of their ability to make behavioral changes to overcome the violation of their rights, which depends on each user, but must be encouraged by the follow-up. The following quotes highlight the categorizations:

It’s the families who often don’t understand why they’re being referred. Families who are resistant don’t open up; they don’t want to talk. Now, a family that isn’t willing, we have to look for communication strategies with them all the time (Interview 04).

The family that adheres to the program usually welcomes you at appointments, on visits, they go to CREAS, the referrals you make, they tend to comply with the referrals, so dealing with the family is more relaxed and much easier (Interview 08).

The organization is when the abusive relationship breaks down, or when the family resolves with the sibling how to take in the elderly, when we notice that the child wasn’t going to school and now, they are, they weren’t going to the clinic and now they are, so the family has become more organized (Interview 09).

But it can be a super disorganized family, where nobody talks to anybody, a family where nobody interacts, nobody does each other any favors, they don’t socialize, they don’t talk to each other. Here it’s very common to find a family so disorganized that they can’t formalize a demand, they can’t tell you what they’re feeling, they can’t say “I’m fine” (Interview 07).

Therefore, in interactions with users, the techniques use behavioral categories to classify the public by judging attitudes that differentiate them as “resistant” versus “adherent”, “organized” versus “disorganized”. The classifications reveal the techniques acting as agents, or exercising their sociological discretion (Maynard-Moody; Musheno, 2012), by making judgments about motivational behaviors and the ability to overcome violations, wrapped in stereotypes of normality and a dependent public. Nevertheless, what influences the practices of the techniques and the user categorization schemes they construct? This answer is presented in the next section.

The influence of professional knowledge of techniques and their internal relationships

In general, the group of technicians surveyed have similar, experienced professional careers, with little social or community engagement, showing that part of their professional and relational profile is quite similar. The professional characteristics that vary the most among the SLB are length of time working in the service, profession and job satisfaction. Of the 11 technicians, six have worked at Paefi for up to four years and five for between four and eight years. This is the same proportion between motivated (six) and unmotivated (five) technicians. In terms of profession, six are social workers, three psychologists, one lawyer and one artist. Motivation seems to be higher among technicians with less time working in the service and psychologists are proportionally more unmotivated than social workers are.

Exercises were carried out to see if these professional characteristics, although few in number, could be associated with differences in their performance. However, cross-referencing these individual characteristics with the follow-up practices did not prove to be analytically adequate for this case, because, as mentioned, in terms of practices, the technicians act in a way that adheres to the norms and in a standardized way when following up on cases. Even the crosschecks between the varying characteristics of the professional profile (profession, satisfaction and length of time working) and the two-categorization schemes identified did not show any relevant results.

As no significant findings were found in the analysis of the individual differences in the techniques’ performance according to profession, length of time and professional satisfaction, we tried to identify whether these aspects and the other factors considered (relational, user attributes and organizational aspects) have a general influence on the practices and user categorization schemes.

What was discovered about the influence of professional knowledge on the behavior of the technicians is that they share a socio-assistance orientation for their work that dilutes the specificities of their professional training. Even so, they articulate specific knowledge from their professions to carry out certain practices in the implementation of the service, in the same direction as other studies (Harrits; Cecchini, 2019; Pires; Santos, 2021). The following quote highlights this consideration:

You’re not a social worker and you’re not a psychologist, it’s a technical position, you’re a public policy analyst, of course your professional training affects the way you look at cases, but there’s no social work unit here (Interview 06).

This influence of the social assistance orientation is revealed in two ways. Firstly, in the use of the expression “therapeutic”, typical of the field of knowledge of clinical psychology, applied by two of the three psychologists interviewed to characterize part of the follow-up work carried out at Paefi. Secondly, in the references typical of the field of social service knowledge used several times by one of the social workers, to indicate the limits - associated with the digitalization of public services - and the possibilities for Paefi to contribute to overcoming the situation of violation of the family’s rights, which involves “social inclusion” and “employment”. The excerpts below illustrate these considerations:

It’s like we say in psychology that this is social assistance, but there’s no way, this therapeutic alliance that you create as a psychologist, as a social worker [...], to break the ice, so that later, when the person is feeling well [...], then they can talk a bit about themselves and build the plan (Interview 07).

I had to understand this grandmother’s relationship with this mother, why this mother acted with these girls in this way and then I would try to point out to each one their place, what they were doing to improve this relationship [...]. Sometimes it works, sometimes it doesn’t, but it’s very therapeutic nonetheless (Interview 03).

Outsourcing, precariousness, and there was also this issue of the digital platform that came in, capital has incorporated technology so much, I mean robots, the internet, that there’s no room for human labor. Of course, there will be, because capital profits from the exploitation of labor. Digital exclusion, how many children in the families I work with don’t have the internet, their mother doesn’t have a cell phone (Interview 06).

Meanwhile, the inter-organizational relationships that most affect the implementation of practices by technicians are those established between themselves and with the social assistance network, which reinforces the importance of the “micro-network of multiple relationships” in which frontline agents carry out their work (Hupe; Hill, 2007). In the light of the studies reviewed (Maroulis, 2017; Siciliano, 2015), the relationship with colleagues is considered to be to ask for help to discuss doubts, difficulties and possible ways forward in the cases being monitored, or to resolve specific legal issues with the lawyer or, to a lesser extent, for collective activities with the art educator. For example, in a face-to-face service observed at CREAS, the technician contacted the legal advisor for guidance on a pension case related to a teenage user. The advisor was helpful and explained that it was a simple case, which required a few documents, the parents’ authorization and a subsequent referral to the public defender’s office (Observed service 10).

With the social assistance network, the relationship proved to be fundamental for the technicians to interpret the users’ situation in the preliminary assessment of the case or to decide on referrals or simply to discuss the case. The relationship with the network at Paefi’s street level brings a cognitive dimension to the realization of practices: understanding the family together with other professionals, discussing the case, thinking about strategies, is seen as something very positive and necessary for working in the service, even if resources are limited, as the following quotes show:

We discuss cases, we discuss with the network, it’s good, it’s also a way for us to get to know the family (Interview 05).

We get a lot of support, a lot of feedback from the NIR, which is the Regional Intersectoral Center. We discuss a case, when you need to talk to the network, what the network is doing, even to avoid doing work, rework, and we get very good feedback (Interview 07).

The child is out of school, where’s the place in the nursery? There aren’t any. Where’s the place in the integrated school? It’s a struggle to get it (Interview 06).

To summarize, there were no significant influences of the profession on the work of the technicians in terms of individual variations. They share a common socio-assistance orientation for their work, but seem to articulate specific knowledge from their professional training to characterize the practice of accompanying the implementation of the service. The internal relationships influence a certain logic of collective construction of the SLB’s work, with information, knowledge and strategies being exchanged between them and other bureaucrats in the social assistance network, even though each of them monitors a case individually. This brings to light the political foundation of intersectorality, i.e. its concrete incorporation into the design of public policies (Grau, 2014).

Influence of the organizational context

Research has shown that the organizational context for implementing Paefi is characterized by complex work, an overload of demands, with a waiting list of cases for constant follow-up, insufficient resources, especially services and benefits from other policies, and a lack of any regulation of workflows and referrals to the justice network and system. The local arrangement for implementing Paefi foresees the participation of a set of actors and relational dynamics for the case to enter and leave the service, operating with a chain movement with four moments (insertion of the case, assignment to techniques, systematic monitoring of the case and formal dismissal), with operating dynamics and participants specific to each of them. Given this context, it is assumed that Paefi technicians make interpretations and discretionary decisions to deal with pressures and circumstances perceived as uncertain and inadequate. The context in which the techniques surveyed operate is an element common to all of them. However, the general behaviour in the process of implementing the service seems to be influenced by three sets of organizational factors: i) the normative structure of the policy; ii) working conditions; and iii) the local implementation arrangement itself.

Paefi’s dense standardization is expressed in the extensive list of practices planned for implementation, which, as mentioned, are carried out by the technicians in a way that is very much in line with the guidelines, with little divergence between what is planned and what is carried out on the front line. This normative structure limits the agents’ discretion in carrying out the practices, but does not eliminate it, as they choose what not to do within the options and combine the planned practices with the real possibilities of the cases, constructing a way of following up based on the interpretation of the users’ entry and exit conditions. One possible explanation for this is the unscheduled nature (Nogueira, 1998) of follow-up, which brings ambiguity to the means of implementation (Matland, 1995), since each case is different, and seems to open up the space for discretion circumscribed by the rules.

On the subject of working conditions, we highlight work overload, which seems to be related to the prioritization of certain activities to the detriment of others, especially the failure to provide collective care, and the resistance to providing care on spontaneous demand. The perceived lack of clearer workflows with the network and the justice system leads to doubts about the referrals to be made, causing uncertainty and demands for specific training.

Furthermore, the implementation arrangement influences the work of the technicians because the cases referred to them by the justice system lead them to investigate and respond to the case. Two other pieces of evidence suggest that this organizational aspect intersects with the SLB’s internal relations, influencing their work in an aggregate way. Firstly, the expected (and necessary) relationship with the network leads to referral practices. Secondly, at the formal end of the case, which, according to the arrangement, is carried out jointly by the technicians and the coordination in the practice of case management, they resume monitoring and present justifications for the decision, based on interpretations of the family’s ability to respond to the monitoring process. The following quotes illustrate these influences:

Then I think about the follow-up itself, which includes getting the family involved, thinking about how the follow-up is going to take place, the links that are made with the network and it’s something that takes up a lot of time, because sometimes when we talk about follow-up in Paefi, people restrict the work to care (Interview 03).

When we joined, there was a meeting once in which a manager came to talk about the work that is written down. And what’s written down is what’s in the rules, but for all the rules you have to define flows, and that’s still a long way off (Interview 05).

There was a case I took that was a police report. Referred to by the Public Prosecutor’s Office, based on a police report, so I said: “yeah, I’ve become a cop” (Interview 02).

Influence of user attributes

The general profile of the Paefi users surveyed shows that the majority are elderly people, followed by children and adolescents, victims of violence and neglect (Table 1), and referred by rights defense bodies and the justice system.

Table 1.
Number of families monitored by Paefi by segment/condition and violation suffered, Belo Horizonte region surveyed, 2021

Rights violation	Teenagers	Children	Elderly	People with disabilities	Total
Sexual violence	1	1			2
Intra-familial sexual abuse	2	2			4
Extra familial sexual abuse	3	1			4
Psychological violence	3	2	16	1	22
Physical violence	10	5	5	1	21
Moral violence			1		1
Property violence			1		1
Neglect	10	23	14	2	49
Self-neglect			5		5
Abandonment			3		3
Child labor	1				1
Street situation			1	1	2
Total (n)	30	34	46	4	114
%	26.32%	29.82%	40.35%	3.51%	100%

• Source: compiled by the author, based on SIGPS/BH data.

In general, some practices and especially the construction of categorization schemes are influenced by user attributes. Faced with flesh-and-blood beneficiaries with real demands, Paefi’s street-level meetings are processes of co-production between bureaucrats and the public, which requires establishing consent and agreements. In this sense, the user’s motivation in the face of these agreements is decisive for the course of the follow-up, in other words, for deciding on the next activity or course of action to be taken and for classifying the user as participatory or resistant.

For example, if the user has resolved the outstanding prescription, the next step is to apply for the Benefício de Prestação Continuada (BPC) (Continuous Welfare Benefit), as observed in one case. If the user is unmotivated and classified as resistant, the technicians decide to make visits to understand the family “dynamics”. In addition, if the user “doesn’t comply”, they contact the network to try other means of adherence. This reveals a logic of action, within the case, in which the user’s motivation affects the carrying out of certain practices (visits, contact with the network), linked to the categorization of the resistant user. This logic seems to express judgments made by the technicians about what it means to be motivated or resistant in order to take the next step or drop the case.

Users’ ability to deal with and overcome their problems is another behavior considered by the technicians when classifying the organized user, but it is also a source for the interpretations and decisions made at the time of the case’s outcome. For example, in the case of an adolescent referred by the Guardianship Council as a victim of sexual exploitation, the technician offers a basic set of practices (remote care, individual care and contact with the network) and reports that, during follow-up, mother and daughter “denied sexual exploitation” and showed a “strong bond” between them. During their last appointments, they said that “their children were enrolled and doing their schoolwork”. These behaviors seem to have led the technician to interpret, in the words she recorded in the case file, which “the family situation is stable and the mother is trying to provide a supportive and protective environment for the children’s development”.⁸ The real case indicates that, after interpreting the family’s situation and judging that it was capable of showing itself to be organized, the technician decided to leave the service.

In addition to the motivation and capacity of users, other characteristics influence the performance. First, it is expected that the age group will define certain practices such as referrals. For example, some interviewees reported and it was observed in a service for adolescents that, when the target audience is children and adolescents, the technicians check the school situation to assess the need for some contact with the institution. If the user is a woman in a situation of violence, although there are doubts about the action flows, there will be referrals to specialized Defense Centers, as discussed in observed meetings. If the user is elderly and the technician observes that she depends on specialized care at home to move around or perform her personal hygiene, she disconnects the case and refers the user to the specialized service for this population, as identified in cases in the Medical Records. Thus, the stage of the user’s life cycle is related to certain types of referral practices. To meet the objective condition of age, the technicians articulate choices within the discretionary space of the available practices, responding almost automatically to them.

In the same vein, the demands that users present during the follow-up also affect referral practices, especially if they are more objective demands, that is, related to access to or receipt of services and benefits (single registry for those who have no income, documents for those who have lost them, psychological consultation, registration). On the other hand, when users stop presenting demands of this type, the technicians seem to interpret that the possibilities for intervention have been “exhausted” and this may be important for discontinuing the case.

Finally, the physical health and mobility conditions of users are added to these considerations. For some technicians, if the user is an elderly person, “bedridden” or has difficulty “moving” to the regional office, they choose to make visits, or when traveling is not an impediment, face-to-face care is provided at CREAS. The excerpts in Box 2 highlight the associations between user attributes and the performance of the technicians.

Box 2.
Influence of user attributes on the performance of techniques: the illustrative ones

User Attributes	Excerpts
Behavior	Motivation The visit depends on how they respond. Both to the interventions and to the openness they give, because there are people who don’t open up and I respect that, if the person doesn’t open up, I go straight to the point, I’m more objective (Interview 03). Capacity I think that in these cases, look, “what is the violation?”, a son who was attacking and he was able to rethink his practice, the coexistence there became lighter, so he’s overcome it (Interview 02).
Demands	When we know what resources, we have or don’t have to assist the family in terms of assistance, we choose some strategies. Many families already have this very clear, what their demands are and what they need, so we choose what to do based on that (Interview 05). For example, I took a family conflict that was more intense, and it was a conflict that actually caused a bit of psychological abuse in the elderly woman. In this case, I stayed a little longer, I mediated with the family, I involved the legal advisor with me and so on, the demand was complicated (Interview 02).
Physical condition	What I thought, what I planned for the intervention, I thought, guys, this is an elderly woman who doesn’t leave the house, she only leaves to go to the doctor, things like that, she has no connection with me, because I had only seen her children, her three children. I was already starting to provide some in-person care, we could already do visits, so I scheduled a visit with the negligent son (Interview 01). For home visits, I prioritize those who can’t come here to CREAS or who have difficulty connecting their family to the service (Interview 09).

• Source: Prepared by the author.

The influence of users’ attributes can also be explored through their intersection with the other factors considered. The social assistance orientation that the professionals share and that makes them act more as technicians than as social workers or psychologists is added to the behaviors (motivation and responsiveness) of the users, influencing the performance of the technicians. In this case, the two factors combined are decisive both for the practice of agreeing on the Monitoring Plan and for naming users as “adherent-resistant” and “organized-disorganized”. This is because, as seen, the users’ behaviors are evaluated by the SLBs to take the next step and to judge their capacity to carry out “movements” that lead to overcoming the situation of rights and, with this, to their process of “organization”.

These expectations are anchored in two operational guidelines of SUAS social assistance services: i) the family as the basic nucleus of intervention, which can be strengthened with the services and protect and care for its members (Brasil, 2012); ii) the need for active participation of users in the monitoring process for the “procedural dynamics” and almost never finished, of overcoming the violation of rights (Carmona, 2019).

Thus, those who adhere and become capable, that is, “organize” themselves, are removed from Paefi, and can be referred to other services when they continue to show some vulnerability. With those who resist and those who do not organize themselves, on the other hand, they continue, or as they say, “insist” on monitoring, which suggests a positive differentiation of cases by the techniques and implementation practices convergent with the objectives of Paefi. They insist until the family refuses to move forward, or until they “exhaust the possibilities” of intervention, reinforcing once again the influence of professional social assistance guidance on the practices of the technicians.

The following statements illustrate the evidence on how categorization schemes are influenced, in an aggregated way, by their behaviors and by the social assistance guidance shared by the SLBs:

in social assistance, if the family does not show up for the appointment, you actively search for them, you go to the house, you search, you insist, you send a telegram, you try to contact them by phone, you make a home visit in an attempt to make the person aware that you are there as a social assistance provider (Interview Tech. 06).

I tried the interventions and the family is not adhering to the intervention, which is a reason to close the case, if you have exhausted the intervention possibilities, you close it. Or if you have exhausted a case in which you see that the elderly woman was in a risky situation, but during the follow-up, what I noticed is that she is not an elderly woman who suffers violence, mistreatment, she is very well cared for (Interview Tech. 01).

It turns off when we observe that that family has already managed to understand the situation that led to that insertion in the service, perhaps it does not mean that that situation will not occur again, but that the person was able to make an interpretation and was able to organize themselves in a way that they experience today, as I would say, they have a different experience from the one that came to us (Interview Tech. 04).

The age of the users is combined with the professional knowledge of the technicians, derived from their training in psychology and social work, leading to a more specific practice of mediating the interaction between the technicians and the users during the consultations. This is revealed in the use of dynamics and pedagogical resources, such as drawing with children and playful games with teenagers, as illustrated in the following quotes.

I really like using the genogram, which is a graph that we make, it’s a drawing that I make with the family, for them, for example, “oh, tell me the story, what your family is like, from your grandparents, what do you know, where violence arises” (Interview 06).

Children and adolescents, sometimes the child uses drawings or some toy, we had a toy library, which has now been deactivated, they use some element to allow a conversation to flow. With adolescents, I try, with my minimum knowledge of updated things, to bring up some subject that interests them, to find out about their tastes, what they like, to learn about their experiences (Interview 03).

In short, the more or less motivated and active behavior in relation to what is agreed upon in the follow-up (attending the appointment, going to the consultation, obtaining the documentation) is essential to decide the step-by-step process, the referrals, when to visit the family and, in addition to the social assistance guidance, to define the “resistant” user and persist with the case. The motivation of the users is also an important attribute for the behavior of the SLBs, in light of Tummers (2017), Møller (2016) and Guul, Pedersen e Petersen (2021).

The behaviors of users judged as moves to overcome violations (enrolling their child in school, improving self-care, becoming aware of the violence suffered or perpetrated, creating emotional bonds) seem to be decisive, together with the professional logic of social assistance, in classifying the user as “organized”, which can lead to the termination of the case. Personal attributes, such as being elderly, a child or an adolescent, and the type of demands brought by them, especially those more concrete and related to access to services and benefits, are considered by the techniques for proposing referrals and for mediating the relationship with pedagogical resources. Physical conditions, mobility and health are considered for carrying out home visits or in-person care at CREAS.

Box 3 systematizes the findings on the isolated and aggregated relationship between the users’ attributes and the practices and schemes for categorizing the public in the implementation of Paefi by the SLBs.

Box 3.
The influence of user characteristics on the performance of SLB.

User characteristics			Element of performance influenced by the identified characteristics
General	Identified in the research
Behaviors	Motivation	Participate in monitoring activities, complete monitoring + Professional social assistance guidance	Practice of agreeing on the step-by-step follow-up; “Adherent-resistant user” categorization scheme Practice of insisting on the case
	Problem-solving ability	Understanding the violation suffered, carrying out behaviors to overcome the violation + Professional social assistance guidance	“Organized-disorganized” user behavior categorization scheme Case termination practice
Personal and physical attributes	Age	Child, teenager, elderly	Practice specific referrals
		Child, teenager, elderly + Professional knowledge of SLB	Mediation practice with pedagogical resources
	Mobility and health conditions (bedridden, difficulty moving around)		Practical home visits or face-to-face care at CREAS
Demands presented	Materials (medical consultation, transportation, food)		Practice specific referrals

• Source: Prepared by the author.

Finally, although it is not the focus of this article, it would be worth exploring the implications of the performance of the techniques on the objectives of Paefi or on the processes of reproduction of inequalities intertwined in the front line of implementation. Along these lines, for record purposes, the research whose excerpt is presented in this article also identified the processes of differentiation in the treatment of users engendered in their practices and in the categorization schemes. It was observed that, depending on the type of user followed, the techniques select the cases to insist or disconnect. Cases of resistant are chosen to insist, following the socio-assistance guidance of SUAS, which states that the user should be worked with until there is nothing more that can be done. Cases of organized people are selected for disconnection, because they were able to overcome the violations, and can still be referred to other services in the network if they still present vulnerability. In these cases, the differentiations seem to be positive and would indicate an action focused on the objectives of the service and with less risk of reproducing inequalities (Assis, 2023).

Conclusions

Since this is a single case study, with few techniques, with a relational profile of engagement in homogeneous social groups, in addition to the observation of standardized performance in terms of practices carried out, the analysis of the influences of the characteristics of the techniques on the variation of their performance was not adequate. The exercises carried out associated profession, length of service and professional satisfaction with the categorization schemes and no relevant variations were found.

However, given the impossibility of concluding categorically that there was no influence of professional factors or these subtle influences, an interpretative effort was undertaken to associate internal relationships, user attributes and organizational elements with the performance of practices and mobilization of users’ social categorization schemes, seeking to understand how they influence more general behavior in the process of implementing Paefi.

This interpretative effort showed that the specialized knowledge of the professions of social workers and psychologists seems to be diluted by the principles of the field of social assistance intervention built for SUAS, but still, to some extent, they to act in the service use them as references. Furthermore, the agents’ actions occur in a very collective context, in which decisions are made as a team, with local and network management, whose capacity to offer also seems to be fundamental for practices at street level. The same can be said of the organizational conditions, which are highly structured by the rules of the service, but also have challenges of work overload and the availability of certain resources that enable the field of action.

But these elements make up only a part of the work of the SLBs of Paefi. This is because, most of it seems to be conditioned by the characteristics of the users served, especially their demands, behaviors and some physical conditions, which can also intersect with the professional knowledge of the techniques to influence these practices and the construction of categorization schemes. These attributes, isolated or aggregated, are considered for referrals and, in particular, for the judgments made in the construction of the behavioral categorization schemes of the public. This reinforces the importance of these factors for the performance of SLB, reinforcing previous studies (Guul; Pedersen; Petersen, 2021; Møller, 2016). In short, the techniques are guided less by their individual factors, which seem to be diluted both in the structured and collective context in which the implementation takes place, and in the nature of a service that does not fit into complete formalization, because, in order to happen, it needs, much more than the techniques, the user.

The nature of SUAS and Paefi may be the key to explaining this result. Paefi is not very programmable because it is co-produced with users at each service, based on their engagement in the activities and their responses to the changes needed to overcome their problems. Thus, at street level, with the guiding “script” in hand, the service is provided on the spot, which requires flexible action adapted to the reality of the case and the network is offering. This is a contribution of this empirical case to the literature: the findings highlight the reduced room for discretion in a highly regulated service, combined with the need for flexibility in practices, as determinants for the performance of SLBs in policies that depend on high user participation for their implementation, as is the case with Paefi.

Another contribution to the SLB literature is the identification of the “organized” versus “disorganized” user scheme, still little explored in empirical studies, especially because it is wrapped up in a representation of the dependent user, going beyond the deservingness stereotypes already discussed. The article contributes as a research design that integrates diverse data sources and strategies, with an emphasis on ethnographic observation, highlighting methodological connections of social sciences in the analysis of public policies.

At a higher level of contribution, such findings also converge with assumptions of mixed implementation approaches, developed in the 1990s, with the “strong” elements of hegemonic perspectives breaking the “top-down and bottom-up” binomial. In general, such approaches perceive the implementation process in terms of a “scope of limitations for action (rules and roles) and the use made of this scope” (Barrett, 2004, p. 257), with both goals and resources being important and interconnected for performance, as well as the actions of operators and beneficiaries (Lotta, 2019; Pülzl; Treib, 2007). Therefore, the implementation of public policies that aim to change behavior and encourage the active participation of users are precisely more dependent on their demands and motivation to happen and be successful. This is more evident for Paefi and, perhaps, for SUAS services, which foresee the engagement and responsiveness of families as the basis for the co-production of intervention and overcoming violations. Such factors become central to the performance of SLB in the service, whether by inducing improvisations and ad hoc choices in the service rooms, or by supporting classifications of the public. Ultimately, confirming the assumptions of the literature, but refining them based on the case studied, it was observed that the general performance of Paefi agents results from a set of contextual factors in the implementation process and the cases they follow.

References

• ASSIS, Marcos Arcanjo de. Burocratas de nível de rua do Sistema Único de Assistência Social: o que fazem, como interagem e classificam os usuários? 2023. 282 f. Tese (Doutorado em Ciência Política) - Universidade Federal de Minas Gerais, Faculdade de Filosofia e Ciências Humanas, Belo Horizonte, 2023.
• BARRETT, Susan M. Implementation studies: time to revival? Reflections on 20 years of implementation studies. Public Administration, v. 82, n. 2, p. 249-262, 2004.
• BELO HORIZONTE. Secretaria de Assistência Social, Segurança Alimentar e Cidadania. Orientações técnicas para o aprimoramento dos processos de trabalho e de gestão do Centro de Referência Especializado de Assistência Social (CREAS) no município de Belo Horizonte. Belo Horizonte, 2020.
• BRASIL. Ministério do Desenvolvimento Social e Combate à Fome. Secretaria Nacional de Assistência Social. Norma Operacional Básica/SUAS. 12 dez. 2012. Brasília: MDS, 2012.
• BRASIL. Ministério do Desenvolvimento Social e Combate à Fome. Secretaria Nacional de Assistência Social. Tipificação Nacional de Serviços Socioassistenciais. 13 maio 2014. Brasília: MDS, 2014.
• BRASIL. Ministério do Desenvolvimento Social e Combate à Fome. Secretaria Nacional de Assistência Social. Fundamentos ético políticos e rumos teórico metodológicos para fortalecer o Trabalho Social com Famílias na Política Nacional de Assistência Social. Brasília: MDS, 2016.
• BRASIL. Ministério do Desenvolvimento Social e Combate à Fome. Secretaria Nacional de Assistência Social. Norma Operacional Básica de Recursos Humanos NOB-RH/Suas. 13 dez. 2006. Brasília: MDS, 2006.
• BRASIL. Ministério do Desenvolvimento Social e Combate à Fome. Secretaria Nacional de Assistência Social. Orientações Técnicas: Centro de Referência Especializado de Assistência Social- CREAS. 29 nov. 2011. Brasília: Editora Brasil, 2011.
• BRODKIN, Evelyn Z. Policy work: Street-level organizations under new managerialism. Journal of Public Administration Research and Theory, v. 21, n. 2, p. i253-i277, 2011. [supplement issue]. Disponível em: doi:10.1093/jopart/muq093. Acesso em: 22 abr. 2022.
  » https://doi.org/10.1093/jopart/muq093
• BUNDGAARD, Sofie; NØRGAARD Rebecca. Ethnic disparities in healthcare treatment? 2018. No prelo.
• CARMONA, Daniele Rodrigues Souza. A abordagem da violência e sua superação na política pública de Assistência Social: concepções e práticas dos profissionais do Serviço de Proteção e Atendimento Especializado a Famílias e Indivíduos - PAEFI. 2019. 162f. Dissertação (Mestrado em Promoção de Saúde e Prevenção da Violência) - Faculdade de Medicina, Universidade Federal de Minas Gerais, Belo Horizonte, 2019.
• CFESS - Conselho Federal de Serviço Social. Parâmetros para Atuação de Assistentes Sociais na Política de Assistência Social. Brasília, DF: CFESS, 2011.
• CFP - Conselho Federal de Psicologia; CONPAS - Comissão Nacional de Psicologia na Assistência Social. Nota técnica com parâmetros para atuação das (os) profissionais de psicologia no âmbito do Sistema Único de Assistência Social (SUAS). Brasília, DF: CFP, 2016.
• DOMENICO, Silvia Marcia Russi de. Análise Temática. In: BRUNSTEIN, Janette; GODOY, Arilda Schmidt; BRITO, Eliane Pereira Zamith; ARRUDA, Emílio José Montero (orgs.). Análise de dados qualitativos em pesquisa: múltiplos usos em administração. 1 ed. São Paulo: Editora Mackenzie; FGV Editora, 2020. p. 60-82. [volume 1].
• EVANS, Tony. Organisational rules and discretion in adult social work. British Journal of Social Work, v. 43, n. 4, p. 739-758, 2013.
• GALE, Nicola; DOWSWELL, George; GREENFIELD, Sheila; MARSHALL, Tom. Street‐level diplomacy? Communicative and adaptive work at the front line of implementing public health policies in primary care. Social Science and Medicine, v. 177, p. 9-18, 2017.
• GOFEN, Anat; SELLA, Shelly; GASSNER, Drorit. Levels of analysis in street-level bureaucracy research In: HUPE, Peter L. (ed.). Research handbook on street-level bureaucracy. Cheltenham: Edward Elgar Publishing, 2019. p. 336-350.
• GRAU, Nuria Cunill. La intersectorialidad en las nuevas políticas sociales: Un acercamiento analítico-conceptual. Gestión y política pública. Ciudad de México, v. 23, n. 1, p. 5-46, 2014.
• GUUL, Thorbjørn Sejr; PEDERSEN, Mogens Jin; PETERSEN, Niels Bjørn Grund. Creaming among caseworkers: effects of client competence and client motivation on caseworkers’ willingness to help. Public Administration Review, v. 81, n. 1, p. 12-22, 2021. Disponível em: https://doi.org/10.1111/puar.13297 Acesso em: 16 abr. 2025.
  » https://doi.org/10.1111/puar.13297
• HARRITS, Gitte Sommer; MØLLER, Marie Østergaard. Prevention at the front line: how home nurses, pedagogues, and teachers transform public worry into decisions on special efforts. Public Management Review, v. 16, n. 4, p. 447-480, 2014.
• HARRITS, Gitte Sommer. Reproducing class? exploring class-based cultural practices among Danish and American teachers. Paper presented at Mini-conference on class and culture, Eastern Sociological Association, Baltimore, February 23rd-24th, 2018.
• HARRITS, Gitte Sommer; CECCHINI, Mathilde. Professional agency in street-level work. In: Annual Conference of Europe’s leading forum for the study of public administration and policy, 2019, Belfast. Anais […]. Belfast: European Group for Public Administration, 2019.
• HUPE, Peter. Dimensions of discretion: Specifying the object of street-level bureaucracy research. Der Moderne Staat. Zeitschrift für public policy. Recht und Management, v. 6, n. 2, p. 425-440, 2013.
• HUPE, Peter; BUFFAT, Aurélien. A public service gap: capturing contexts in a comparative approach of street-level bureaucracy. Public Management Review, v. 16, n. 4, p. 548-569, 2014.
• HUPE, Peter; HILL, Michael. Street-level bureaucracy and public accountability. Public Administration, v. 85, n. 2, p. 279-299, 2007. Disponível em: https://doi.org/10.1111/j.1467-9299.2007.00650.x. Acesso em: 5 maio 2022.
  » https://doi.org/10.1111/j.1467-9299.2007.00650.x
• HOFF, Timothy. Medical home implementation: a sensemaking taxonomy of hard and soft best practices. Milbank Q, v. 91, n. 4, p. 771-810, dez. 2013. Disponível em: https://onlinelibrary.wiley.com/doi/10.1111/1468-0009.12033. Acesso em: 5 maio 2022.
  » https://doi.org/10.1111/1468-0009.12033
• KEISER, Lael R. Understanding street-level bureaucrats’ decision making: Determining eligibility in the social security disability program. Public Administration Review, v. 70, n. 2, p. 247-257, 2010.
• LIMA, Luciana Leite; D’ASCENZI, Luciano. O papel da burocracia de nível de rua na implementação e (re)formulação da Política Nacional de Humanização dos serviços de saúde de Porto Alegre (RS). Revista de Administração Pública, Rio de Janeiro, v. 51, n. 1, p. 46-63, fev. 2017.
• LIPSKY, Michael. Burocracia de nível de rua: dilemas do indivíduo nos serviços públicos. Tradução: Arthur Eduardo Moura da Cunha. Brasília: ENAP, 2019. 430 p.
• LOTTA, Gabriela S.; MARQUES, Eduardo Cesar. How social networks affect policy implementation: an analysis of street-level bureaucrats’ performance regarding a health policy. Social Policy & Administration, v. 54, n. 3, p. 345-360, 2020. Disponível em: doi:10.1111/spol.12550. Acesso em: 5 maio 2022.
  » https://doi.org/10.1111/spol.12550
• LOTTA, Gabriela. A política pública como ela é: contribuições dos estudos sobre implementação para a análise de políticas públicas. In: LOTTA, Gabriela (org.). Teorias e análises sobre implementação de políticas públicas no Brasil. Brasília: Enap, 2019. p. 11-38. [Capítulo 1].
• LOTTA, Gabriela. Burocracia e implementação de políticas de saúde: os agentes comunitários na Estratégia Saúde da Família. Rio de Janeiro: Editora Fiocruz, 2015. 321 p.
• LOTTA, Gabriela; SANTIAGO, Ariadne. Autonomia e discricionariedade: matizando conceitos-chave para o estudo de burocracia. BIB, São Paulo, v. 83, n. 1, p. 21-42, 2017.
• MAROULIS, S. The role of social network structure in street-level innovation. The American Review of Public Administration, v. 47, n. 4, p. 419-430, 2017. Disponível em: https://doi.org/10.1177/0275074015611745. Acesso em: 7 set. 2022.
  » https://doi.org/10.1177/0275074015611745
• MATLAND, Richard E. Synthesizing the implementation literature: the ambiguity-conflict model of policy implementation. Journal of Public Administration Research and Theory, v. 5, n. 2, p. 145-174, 1995.
• MAYNARD-MOODY, Steven Williams; MUSHENO, Michael Craig. Cops, teachers, counselors: stories from the front lines of public service. Ann Arbor: University of Michigan Press, 2003.
• MAYNARD-MOODY, Steven Williams; MUSHENO, Michael Craig. Social equities and inequities in practice: street-level workers as agents and pragmatists. Public Administration Review, v. 72, p. 16-23, nov./dec. 2012. [special issue].
• MEYERS, Marcia K.; VORSANGER, Susan. Street-level bureaucrats and the implementation of public policy. In: PETERS, Guy; PIERRE, John (eds.). Handbook of public administration. Thousand Oaks: Sage, 2003. p. 245-255.
• MØLLER, Marie. She isn’t someone I associate with pension: a vignette study of professional reasoning. Professions & Professionalism, v. 6, n. 1, p. 1-20, 2016.
• MØLLER, Marie; HARRITS, Gitte. Constructing at-risk target groups, Critical Policy Studies, v. 7, n. 2, p. 155-176, 2013.
• NIELSEN, Vibeke. Are street-level bureaucrats compelled or enticed to cope? Public Administration, v. 84, n. 4, p. 861-889, 2006.
• NOGUEIRA, Roberto Martinez. Los proyectos sociales: de la certeza omnipresente al comportamiento estratégico. Santiago de Chile: CEPAL, 1998. [Serie Políticas Sociales, 24].
• NUNES, João; LOTTA, Gabriela. Discretion, power and the reproduction of inequality in health policy implementation: Practices, discursive styles and classifications of Brazil’s community health workers. Social Science & Medicine, v. 242, p. 1-8, 2019. [112551]. Disponível em: doi: 10.1016/j.socscimed.2019.112551. Acesso em: 22 abr. 2022.
  » https://doi.org/10.1016/j.socscimed.2019.112551
• PIRES, Roberto; SANTOS, Maria Paula. Desafios do multiprofissionalismo para a redução de danos em Centros de Atenção Psicossocial Álcool e outras Drogas. Saúde e Sociedade, São Paulo, v. 30, n. 2, p. 1-12, 2021. [200072].
• PÜLZL, Helga; TREIB, Oliver. Implementing Public Policy. In: FISCHER, Frank; MILLER, Gerald. J.; SIDNEY, Mara S. (eds.). Handbook of public policy analysis: theory, politics and methods. London; Nova York: CRC Press; Taylor & Francis Group, 2007. p. 89-107. [chapter 7].
• RAAPHORST, Nadine. How to prove, how to interpret and what to do? Uncertainty experiences of street-level tax officials. Public Management Review, v. 20, n. 4, p. 485-502, 2018. Disponível em: doi 10.1080/14719037.2017.1299199. Acesso em: 22 abr. 2022.
  » https://doi.org/10.1080/14719037.2017.1299199
• SICILIANO, Michael D. Professional networks and street-level performance: how public school teachers’ advice networks influence student performance. The American Review of Public Administration, v. 47, n. 1, p. 79-101, 2015.
• THOMANN, Eva. Is output performance all about the resources? A fuzzy‐set qualitative comparative analysis of street‐level bureaucrats in Switzerland. Public Administration, v. 93, n. 1, p. 177-194, 2015.
• TUMMERS, Lars; BEKKERS, Victor; VINK, Evelien; MUSHENO, Michael. Coping during public service delivery: a conceptualization and systematic review of the literature. Journal of Public Administration Research and Theory, v. 25, n. 4, p. 1099-1126, 2015. Disponível em: doi:10.1093/jopart/muu056. Acesso em: 6 fev. 2022.
  » https://doi.org/10.1093/jopart/muu056
• TUMMERS, Lars. The relationship between coping and job performance. Journal of Public Administration Research and Theory, v. 27, n. 1, p. 150-162, 2017.
• WEBER, Max. Ciência e política: duas vocações. São Paulo: Cultrix, 2011.
• WEBER, Max. Economia e sociedade. Brasília: Editora da UnB, 2000.

• ³
  It consists of the articulation between all the provision units of the various offerings of SUAS in the territories, meaning, with other SUAS services at different levels of protection (Brasil, 2011).
• ⁴
  The identification numbers of the two facilities with an equal score closest to the average are: 13106201882 and 13106202206, the last of which was drawn.
• ⁵
  From this point onwards, the term “techniques” will be used to refer to the work of the SLB studied, who identify themselves and are called this way on a daily basis, regardless of their position or professional training (social work, psychology, law, arts).
• ⁶
  The research was approved by the Comitê de Ética em Pesquisa vinculado à Plataforma Brasil (Research Ethics Committee linked to the Brazil Platform), according to the Final Report (CAAE number: 36201720.1.0000.5149), which provided for the use of Informed Consent Forms (ICFs) for the interviews, which were duly applied.
• ⁷
  Access to the medical records was authorized by Termo de Compromisso de Utilização de Dados (TCUD) (Data Use Commitment Agreement), signed by the sector responsible for the municipal management of SUAS.
• ⁸
  Source: Socio-assistance Report from Prontuário 5.
• ¹
  The author thanks the Fapemig (Minas Gerais Research Support Foundation) for the funding in the writing of the article under the project APQ 05331-23, Call for proposals 09/2023.

• Editors
  Débora Rezende de Almeida
  Rebecca Neaera Abers

Publication Dates

• Publication in this collection
  04 July 2025
• Date of issue
  2025

History

• Received
  24 Jan 2024
• Accepted
  24 Jan 2025