When autonomy is necessary for performance: Brazilian public defenders’ offices

Public Defender’s Offices (PDOs) are responsible for providing legal assistance to citizens in situations of social and economic vulnerability. These organizations litigate directly against the state to guarantee the provision of public services and prevent government abuses or deviations. Therefore, these organizations require autonomy to fulfill their mission. This work analyzes the relationship between autonomy and performance in Brazilian PDOs. The work tests the perception of the impact between autonomy and performance of Brazilian PDOs. A survey was carried out with bureaucrats from 27 Brazilian PDOs, and the data was analyzed through structural equation modeling. The variables observed include, in addition to performance and autonomy, availability of resources, competencies, and accountability. The results point out a perception that PDOs must be autonomous and free from political influences to perform their mission properly. The bureaucrats’ skills and accountability were identified as relevant factors for the organizations’ performance. However, there was no significant relationship between resource availability and perceived performance. The contributions of this study include obtaining evidence of empirical validity to measure the constructs “autonomy,” “availability of resources,” “skills,” and “responsibility” and their impact on the performance of PDOs.


INTRODUCTION
Performance and autonomy are important topics in the study of Justice Administration, a research field with high practical and political salience but still neglected by the broader field of Public Administration (Guimaraes, Gomes, & Guarido, 2018). Understanding these concepts and how they possibly relate is fundamental to solving a paradoxical problem in the Judiciary of many countries, such as expanding access to justice and, at the same time, meeting the demand for better and faster justice services (Guimaraes et al., 2018).
Performance and autonomy are multilevel and multidimensional constructs, which have been observed in various public organizations through different aspects. This study observes these constructs at individual level through the perception of bureaucrats. Therefore, performance corresponds to the perception of public defenders and PDOs' managers about the results and policies produced by the PDO. In turn, autonomy corresponds to the perception of these bureaucrats about the discretion in carrying out their work functions.
We have not been successful in finding studies that empirically associate these dimensions in Justice Organizations. However, there are studies in the public administration literature that examine such relationship, but these studies are not univocal. In fact, there is still no compelling evidence that indicate greater performance of autonomous agencies (Maggetti & Verhoest, 2014).
On the one hand, autonomy would tend to influence performance by to promoting organizational and entrepreneurial capacities. Autonomous agencies tend to have a great deal of discretion in the decision-making process and in the management of resources, which allows these organizations to anticipate environmental changes (D' Almeida & Klingner, 2008;Silver, 1993). On the other hand, high levels of autonomy can affect performance, as it might represent weaknesses in the accountability mechanisms. Thus, bureaucrats would tend to maximize their own interests, overriding the public interest, and therefore opting to sub-optimal performance (Han & Hong, 2019;Kim & Cho, 2014).
Autonomy may be relevant to explain the performance of Public Defender's Offices (PDOs), an organization responsible for providing legal assistance to socially and economically vulnerable citizens. In Brazil, this organization is an executive agency that is not part of the Judiciary but orbits the Justice System. To carry out their mission, PDOs must directly litigate against the State in order to ensure the provision of public services and to prevent abuse or deviation by the government in applying the law. Therefore, elected politicians could intentionally stifle PDOs, depriving them of the resources and means necessary for their performance.
Therefore, we can ask: does greater perception of autonomy by PDOs' bureaucrats leads to higher levels of perceived performance? Having this question as a guide, this study aims to test the impact of perceived autonomy on the perceived performance of Brazilian PDOs. Other factors capable of influencing performance are also observed, such as the availability of resources, the skills, and characteristics of bureaucrats. This research was operationalized through the application of a survey to observe the perception of the Brazilian PDOs' bureaucrats. The analyses were performed using structural equation modeling. The results demonstrate that autonomy, skills of bureaucrats, and accountability directly influence performance. However, there is no significant relationship between resources and performance.

THEORETICAL FRAMEWORK
The relationship between the autonomy of public organizations and their performance can be understood through the principal-agent approach whereby principals are political representatives who delegate authority to autonomous agencies, and agents are those who use delegated authority to govern such agencies. This theoretical perspective is relevant, as there is a link between the internal complexity of organizations and their external environment, which involves the relationship with the government (Talbot, 2010).
Relations between political representatives and bureaucrats are complex and largely conditioned by cultural values and assumptions. Political representatives can only realize the benefits of delegation by granting discretion to the bureaucrats. In turn, bureaucrats tend to maximize their own interests, which may conflict with the interests of the politicians. This situation can produce a loss of agency, that is, the results generated by the bureaucrats can be different from the preferences of the political representatives. In this sense, the institutional design of the delegation must balance the control mechanisms and the discretion in order to allow the bureaucrats to produce the results desired by the political representatives (Pollitt & Bouckaert, 2017;Sweet & Thatcher, 2002).

Autonomy and performance of PDOs
Governance is a multidimensional construct, of which autonomy (often referred to as independence) is a fundamental dimension. This dimension can be expressed, for example, through the dependence of Justice Organizations on State resources; the ability of law enforcement officials to act in accordance with laws and the constitution; possible interference by members of other powers in judicial decisions; interests and pressures of third parties (Akutsu & Guimarães, 2015).
In the specific case of PDOs, there is an immanent conflict between principal and agent, as the PDO litigates directly against the State in cases of public policy failures and government arbitrariness. Thus, PDOs need certain levels of discretion to produce policies that are often contrary to government interests. However, there is indication that that the performance levels of the Brazilian federal PDO reduced after that agency constitutionally acquired the condition of political and administrative autonomy (Buta, Gomes, & Lima, 2020). However, this has not yet been empirically tested.
In this regard, it is noteworthy that the Brazilian PDO is practically unaccountable to other administrative agencies, and only loosely supervised by the executive and legislative branches. From an individual perspective, this has generated an expansion of rights and prerogatives of public defenders and, consequently, greater freedom of functional action. However, from an institutional perspective, it made legal assistance services more expensive, which made unfeasible the universalization of the legal assistance provision to the needy, hampering the use PDO as an effective tool to reduce the levels of social exclusion and marginalization (Cunha, 2021).
Performance is also a multidimensional and multilevel construct, which has been observed in the most diverse public organizations under varied aspects. With regard to Justice Organizations, performance has been linked to: speed of judicial proceedings (e.g., Abramo, 2010), workload and output (e.g., Beenstock & Haitovsky, 2004;Jonski & Mankowski, 2014), efficiency (e.g., Yeung, 2020), access to justice (e.g., Cunha, Oliveira, & Glezer, 2014;Oliveira & Cunha, 2016), quality, and effectiveness (Gomes & Guimarães, 2013). In the present study, performance is observed through the perception of bureaucrats about the results and policies produced by the PDO.
Thus, the direction of the relationship between autonomy and performance may vary. Some studies indicate that autonomy and performance are directly related (Anand, Exworthy, Frosini, & Jones, 2012;Bertelli, 2006;Braadbaart, Van Eybergen, & Hoffer, 2007;Silver, 1993;Vining, Laurin, & Weimer, 2015). Other studies point to an inverse relationship, that is, higher levels of autonomy would be related to lower levels of performance (Kim & Cho, 2014;Overman & van Thiel, 2016). There are also articles that bring a curvilinear relationship (Fukuyama, 2013;Rainey & Steinbauer, 1999). In this case, balanced levels of autonomy may positively influence performance, but the excess or lack of autonomy would contribute to lower levels of performance. There are even studies that indicate that autonomy is not such a relevant factor in defining the performance levels of public organizations (Han & Hong, 2019;Verhoest & Wynen, 2016), or even do not find a significant relationship between autonomy and performance (Yamamoto, 2006).

Research hypothesis
Considering these conclusions, some hypotheses can be formulated, in order to highlight the main constructs and respective theoretical relationships arising from the theoretical-empirical references discussed. Testing these hypotheses allows adding empirical evidence to the relationship between the autonomy and performance constructs, as well as bringing empirical evidence to resolve controversies in the collected studies.
The performance of public organizations depends on several factors, such as supportive behavior by external stakeholders, as politicians for example; agency autonomy with regard to the implementation of its mission; and certain leadership behaviors (Rainey & Steinbauer, 1999). In the case of PDOs, it is worth remembering that they litigate directly against the state, for example, in cases where public policies are not being satisfactorily delivered to the vulnerable population. Furthermore, these agencies are also sometimes anathema when they are seen as defenders of criminals. Thus, elected politicians may have vested interests in stifling a legal aid service that is causing them some sort of embarrassment. Consequently, PDOs with higher levels of political and administrative autonomy would tend to perform better. Therefore, here is the following hypothesis:

H1: Higher levels of autonomy perceived by PDO bureaucrats lead to higher levels of performance.
Autonomy, however, should not be the only factor capable of explaining the performance of PDOs. The availability of resources and the capabilities of bureaucrats must also be considered. The availability of resources is a factor that has been studied as an explanatory variable for judicial performance, but the way in which this variable influences performance is still controversial.
For example, there are indications that the importance of personnel resources for performance of Justice Organizations is relative. The surplus of personnel may cause a reduction in the level of performance, taken as productivity (Beenstock & Haitovsky, 2004;Gomes et al., 2019;Gomes & Freitas, 2017;Gomes & Guimarães, 2013). Investment in technology, however, has been positively related to increased productivity in courts. Nevertheless, it should not be assumed that performance will increase whenever technological resources expand. The productive capacity has limitations, as it depends on the capacities of judges and their assistants .
In fact, there must be a limit between the ratio of resources to performance, especially when looking at issues of efficiency (cost/benefit) and average productivity of employees. However, in absolute terms, the expansion of available resources (inputs) must lead to an increase in the outputs (Overman & van Thiel, 2016). It is worth thinking that an increase in the number of public defenders, for example, will generate an increase in the number of citizens assisted by the PDO, even if this means that each of the public defenders will start to show less productivity. Therefore, we expect to find a positive relationship between resources availability and performance: H2: Higher availability of resources leads to higher performance.
Since the autonomy of public agencies also covers issues related to obtaining resources by their own means (Gilardi, 2002), which, in the specific case of PDOs, may involve attorney's fees, for example, it is possible that, in a condition of greater autonomy, public agencies will have access to a greater amount of resources, and, therefore, there would have greater performance. Therefore:

H3: The relationship between autonomy and performance is mediated by the availability of resources.
This relationship might be tricky if the autonomy represents a greater distance from the politicians responsible for the distribution of public resources. In other words, autonomy can lead to distance from the principal and, consequently, lower levels of resources (van Thiel & Yesilkagit, 2011).
The bureaucracy skills are also a factor that explains their performance. Indeed, capacities and performance are intrinsically related concepts, which, in a broad sense, describe the inputs and outputs of management systems (Christensen & Gazley, 2008), and the analytical, managerial, and political skills are important dimensions of the concept of capacity (Howlett & Ramesh, 2016;Wu, Ramesh, & Howlet, 2015).
What organizations can actually do and the likelihood of success in what they do depends critically on the skills they have to diagnose problems and develop appropriate strategies to address those problems. Successful organizations must be composed of a body of bureaucrats with the ability to acquire and use knowledge, as well as the ability to access and apply scientific and technical knowledge, in addition to mastering analytical techniques (Howlett & Ramesh, 2016). Therefore, here is the following hypothesis: H4: Skill levels are positively related to performance.
In the specific case of Justice organizations, the observation of autonomous judicial councils in Eastern European countries has shown that the mere granting of autonomy, without assuring that social interests are served, does not guarantee the proper functioning of these organizations in practice (Beers, 2012). The potential for implementing quality public policies can certainly come from a professional and insulated bureaucracy (Cingolani, Thomsson, & Denis, 2015).
However, a situation of greater autonomy can influence performance to a greater or lesser extent depending on the capabilities of the bureaucracy. Thus, on the one hand, the impact on performance tends to be negative if an organization has a high degree of autonomy, but is full of professionals without technical competence, or is formed by politicians appointed by vested interests. On the other hand, an organization with the same high levels of autonomy, but with a qualified professional bureaucracy, with the necessary skills and knowledge, would tend to have a positive impact on performance (Fukuyama, 2013). In this sense, the relationship between autonomy and performance would be moderated by the skills of the bureaucracy: H5: The skills of PDOs bureaucrats moderate the relationship between autonomy and performance.
Last, but not least, institutional arrangements in which agents have high levels of discretion tend to cause a loss of agency, that is, the results generated by agents tend to be different from those preferred by the principals. In such cases, the powers delegated to the agents are vast, but the control instruments that the principals have to restrict or reverse the results of unwanted policies are scarce (Sweet & Thatcher, 2002). In such cases, the leaders of the autonomous agency tend not to worry about whether their decisions violate the interests of the principals, thus leading to sub-optimal political choices (Tomic, 2018).
In this sense, maintaining mechanisms capable of keeping agents accountable for their actions would tend to guarantee the performance of autonomous organizations. In fact, multiple accountability channels decrease the chances of failure, correcting unwanted behavior (Schillemans, 2010). Thus, it is expected that higher levels of accountability are related to higher levels of performance, leading to the following hypothesis: H6: Higher levels of accountability lead to higher levels of performance.
Furthermore, the accountability mechanisms influence the behavior of bureaucrats. The closeness between the parent ministry and the agency, with greater trust relationship between principal and agent, brings a stronger perception of accountability (Schillemans et al., 2021) which might influence the agency's performance. It is worth remembering that the absence of accountability mechanisms can lead to sub-optimal performance (Tomic, 2018). Thus, it can be expected that the level of accountability moderates the relationship between autonomy and performance, as stated in the following hypothesis: H7: The level of accountability moderates the relationship between autonomy and performance.

METHODS
A survey was carried out to capture the perception of bureaucrats from Brazilian PDOs about aspects related to the autonomy, skills, resources, accountability, and performance of these agencies. It is assumed that bureaucrats have relevant considerations based on their experience, so that their perception is important to indicate the functioning of the PDOs. It is worth mentioning that this research has a cross sectional design, and the theoretical testing was carried out though a quantitative approach.
Data collection took place via a survey applied to bureaucrats of PDOs throughout the country. The survey was applied via internet to PDOs' bureaucrats all over Brazil in September and October 2019, using the Google Forms tool. The link to the questionnaire and an explanatory text inviting the bureaucrats to be part of the survey were shared in internal communication channels of PDOs and professional associations of public defenders.
The research universe is composed of PDOs' bureaucrats. This includes the 28 PDOs, one in each Brazilian federative unit plus the Federal Public Defender's Office. The IV Diagnosis of the Public Defenders' Offices in Brazil points to a population of 6,062 public defenders in 2015, counting all PDOs. Data for the population of other types of public servants, such as managers and advisors, is not known.
The research sample was not obtained randomly, since the form of application depended on the institutional support of PDOs and professional associations of defenders. Data were collected from 27 PDOs, the PDO of the State of Pernambuco was the only one from that we could not obtain responses. After processing the data, in which responses from interns and outsourced employees were excluded, as well as a response recorded in duplicate, 325 response records remained, which is sufficient for the analysis, for which samples greater than 300 are reliable (Tabachnick & Fidell, 2013).
The questionnaire, that seeks to measure comparatively state capacities, was based on a scale developed by the Center on Democracy, Development, and the Rule of Law, from Stanford University, in partnership with the Brazilian Institute for Applied Economic Research . Recent studies show evidences of validity of this questionnaire (Cavalcante & Lotta, 2021;Gomide, Machado, & Albuquerque, 2019). Furthermore, the Box 1, below, presents factors, concepts and the theoretical framework that supports them. As the survey was designed for public organizations linked to national governments, some changes in the items were necessary to adapt it to the context of subnational PDOs. Before applying the survey, a pilot was carried out with a group of representatives of the target audience in order to validate the items semantically. Semantic validation aims to verify that the items are understandable to the target population (Pasquali, 2010). This step was carried out by administering the questionnaire to eight members of the Federal PDO, the PDO of the Federal District and the PDO of the State of Ceará, selected by convenience. After each answer, the doubts raised about the survey were noted, and the response time was estimated. The application of the pilot allowed the identification of items with confusing wording, to which explanatory texts were added in order to solve any doubts.
The adapted questionnaire consists of ten sections and 72 items. The initial section includes a presentation of the questionnaire, with information about the target audience, the average response time, the commitment to confidentiality and the contact details of the research author. Next, there is a section with items on the sample's socio-demographic profile. The following sections cover aspects related autonomy, performance, and capacities of PDOs.
For the present study, 38 items were used, as the scope of this article covers only the following dimensions: performance (dependent variable), and autonomy, accountability, skills, and resources (independent variables). The responses to the items were organized on a Likert-type scale, with five points of agreement.
Data analysis was performed using structural equation modeling and confirmatory factor analysis. We used RStudio and Onyx (both free software) to measure and build analysis models. Such methodology allows estimating relationships between latent and observed variables (Bollen, 1989). In this sense, we sought to measure the latent variables, performance, autonomy, skills, resources, and accountability through the observed variables represented by the attributes contained in each of the items in the questionnaire. Then, the relationships between each of the latent variables were measured.
Seven models were created for the analyses, one for each hypothesis. For the elaboration of the models 5 and 7, it was necessary to create an interaction variable. This variable was created by the multiplication between one of the indicators linked to the factor Skills (x53) and Autonomy (x10), as well as Accountability (x22) and Autonomy (x10). It is noteworthy that the multiplication of just one indicator of each factor is considered sufficient for the creation of the multiplicative term and prevents measurement error. The result of this multiplication was centralized in order to avoid multicollinearity of the new variable with the variables that formed it (Cortina, Chen, & Dunlap, 2001).

RESULTS AND DISCUSSION
The socio-demographic profile of the sample has the following characteristics. There is a prevalence of public defenders, which correspond to 73% of the sample; the other 27% of the answers are from management employees and advisers. The average age of respondents is 38 years, ranging from 25 to 66 years old. The average working time in public administration is 11 years. The male gender is predominant, representing 55% of the sample, compared to 45% of female respondents. Most respondents, 27% have undergraduate degree as the highest level of education, 53% of respondents have a lato sensu graduate degree, 18% have a master's degree, 1% has a doctorate degree, and 1% has only high school level. Finally, 97% of respondents are not affiliated to a political party.
Initially, the adequacy of the sample for statistical analysis was verified. The sample size is satisfactory, as stated earlier. The Kaiser-Meyer-Olkin test, capable of measuring the sample's adequacy for the analyses, indicates an adequacy value of 0.89, which is acceptable. Bartlett's test was significant, indicating significant relationships between variables. It is concluded, therefore, that the data is adequate for the analyses.
All the relationships between the latent variables and their observable aspects, that is, the questionnaire items, are significant for all models analyzed. A supplementary Box (Appendix) contains the items corresponding to the factors, as well as the factor loads. Factor loads were obtained through confirmatory factor analysis for the construction of this box.
The hypothetical models presented good fit indexes. The global adjustment indexes indicate that the models are representative. It can be seen through the comparative fit index (CFI) and the Tucker-Lewis index (TLI). The closer to one, the better adjusted the model is (Meade, Johnson, & Braddy, 2008). The Root Mean Square Error of Approximation (RMSEA) and the Standardized Root Mean Square Residual (SRMR) deal with errors or residuals of the model. The lower the RMSEA and the SRMR, the better the model fit (Meade et al., 2008). Therefore, they also indicate that the models have good fit. The relationships between the latent variables and the fit indexes of each model can be seen in Table 1.  Autonomy and performance are positively related in all models. This result corroborates H1, which states that higher levels of autonomy perceived by PDO bureaucrats are related to higher levels of performance. It is in line with part of the Public Administration literature that indicates that autonomy positively influences the performance of public organizations (e.g., Anand et al., 2012;Bertelli, 2006;Braadbaart et al., 2007;Silver, 1993;Vining et al., 2015). It is necessary to take into account that this research observes autonomy and performance from an individual perspective. In this regard, the autonomy of public defenders represents a set of prerogatives and rights that lead to a greater capacity for functional action but makes PDOs more expensive and prevents their services from being expanded (Cunha, 2021).
Model 2 shows no significant relationship between resource availability and performance, which undermines H2. It is noteworthy that the majority of respondents strongly agree that human resources, budget and technological resources, as they are currently available, are obstacles to the PDO's performance. In addition, most respondents strongly disagree with the statement that the resources available are sufficient for the PDO to fulfill its mission. Yet, most respondents realize that the results of the PDO in which they work have improved over time. This result, however, must be viewed with caution, as it could represent a general perception of bureaucrats that there is a lack of resources in the PDOs.
It is worth mentioning the indications in the literature that put in doubts the importance of increasing the amount of resources for the performance of Justice Organizations. There are studies that indicate that the increase in personnel resources, for example, may even cause a reduction in the level of performance (Beenstock & Haitovsky, 2004;Gomes et al., 2019;Gomes & Freitas, 2017;Gomes & Guimarães, 2013).
In turn, Model 3 indicates that the effect of autonomy on performance is direct, and not mediated by the availability of resources. It leads us to reject H3, which states that the relationship between autonomy and performance is mediated by the availability of resources. In fact, the relationship between autonomy and resources is negatively related. There is evidence that greater autonomy leads to less availability of resources, and not the other way around. This result goes against what was found by Buta (2019), when studying Latin American PDOs. It is worth mentioning, however, that autonomy in that study is taken in a broad way, that is, the institutional arrangement of the organization as a whole was observed. In the present study, autonomy is related to the individual perception of bureaucrats. It is also noteworthy that Brazilian PDOs have much higher levels of autonomy than PDOs in other Latin American countries (Buta, 2021).
A possible explanation for this result is that the condition of greater autonomy leads to a great distance between the bureaucracy of PDOs and the elected politicians, who are responsible for the distribution of public resources. In this way, these politicians would tend to rely less on this bureaucracy and, hence, reduce the amount of resources allocated to it (van Thiel & Yesilkagit, 2011). Therefore, PDOs operate in a complicated context, in which they must have reasonable levels of autonomy to be able to exercise their mission in the best way, but this ends up alienating them from elected politicians. Accordingly, the autonomy of these agencies should represent the guarantee that they would have the amount of resources necessary for their proper functioning.
The results concerning Model 4 indicate that skills also positively and significantly influence performance. This corroborates H4, which states that skill levels are positively related to performance. This result was expected, since it is well known that organizations that have a body of more qualified bureaucrats, endowed with skills that allow them to acquire and apply knowledge will definitely perform better (Howlett & Ramesh, 2016). These findings are similar to those found by Cavalcante and Pereira (2019), even using different databases and methods.
Model 5 indicates that skills also moderate the relationship between autonomy and performance, corroborating H5. The greater or lesser degree of skills influences the relationship between autonomy and performance. This finding is in agreement with the theorized by Fukuyama (2013). That is, if the PDO is endowed with skilled bureaucrats, granting greater levels of autonomy to this agency will tend to improve the level of performance.
Finally, accountability is another important performance-inducing variable. Model 6 indicates that accountability positively and significantly influences performance, which corroborates H6. In the presence of this variable, the relationship between autonomy and performance is weaker than in the other models. Although both variables are important predictors of performance, the test performed on Model 7 did not indicate that the interaction between autonomy and accountability influences performance, suggesting H7 rejection. In other words, autonomy and accountability are both necessary factors for the performance of PDOs, but they operate through different means. Figure 1 summarizes the hypotheses results.
PDOs' bureaucrats, on average, perceive low to medium levels of accountability. This is a reflection at the individual level of what occurs at the institutional level. The PDO has no hierarchical link with the executive branch, that is, it is an independent body. In addition, the head of the PDO is elected by his or her peers, appointed by the chief executive and approved by the legislature for a fixed term. He manages under the guidance of a board formed exclusively by public defenders. All these factors lead to loose supervision and lack of accountability (Cunha, 2021).

CONCLUSION
In order to test the impact of perceived autonomy on the perceived performance, this work took place through the application of a survey to bureaucrats from Brazilian PDOs. The results indicated that autonomy is positively related to performance. PDOs are organizations that often litigate directly against the State to ensure the provision of quality public services to the most vulnerable population and to avoid discriminatory application of the law by Governments. Therefore, these results indicate that such organizations should operate at certain levels of autonomy. It was also observed that bureaucrats' skills influence performance and moderate the relationship between autonomy and performance. It is also noteworthy that the availability of resources did not present a statistically significant relationship with performance. This result deserves to be better evaluated in future studies, as it may come from a general perception of bureaucrats that there is a lack of resources for the PDOs. In addition, the resources availability was negatively related to autonomy. In other words, greater autonomy is related to the lower availability of resources and may also be related to the distance between the PDOs and the politicians responsible for the distribution of public resources.
It should also be noted that accountability has stronger influence on performance than autonomy. This indicates that caution should be exercised when granting autonomy to PDOs. High levels of autonomy, without mechanisms that allow PDOs to be held accountable for poor performance, will tend to lead to behaviors in which the agent satisfies his own interests, to the detriment of the PDO's mission.
This article presents theoretical contributions. First, this research shows a positive relationship between the autonomy and performance, thus, adding to a group of studies that verify the positive relationship between autonomy and performance. This is evidence that public agencies with a mission related to promoting access to justice and controlling public policies aimed at the vulnerable population need greater autonomy to perform better. It is also important to highlight the novelty of this type of approach for PDOs. Moreover, the study proposes explanations about the mechanisms by which autonomy is able to influence performance, thus indicating that there is an interaction between autonomy and skills of these bureaucrats. In addition, higher levels of autonomy are not related to higher levels of availability of resources.
There are also practical contributions, which can be applied in the management of PDOs, and by extension, to other organizations of the Justice System. The increase in the skills of bureaucrats, as well as the guarantee of autonomy in the exercise of their functions and establish accountability mechanisms may be relevant ways to the expansion of their performance.
This study has limitations, among which it is worth mentioning that the data analysis was carried out in an aggregate manner, which can hide the regional differences between the PDOs. Data collection was not done at random, so that the PDOs and professional associations of public defenders who supported the research by disseminating the survey in their internal communication channels have greater representation in the data. Thus, the data have only internal validity, and cannot be extrapolated to the population.
Future studies may bring complementary approaches that help to better understand the relationship between autonomy and performance, especially from Justice Organizations, which naturally need greater autonomy in their operations. Furthermore, a more detailed analysis is needed specifically on the relationship between availability of resources and their possible interactions with autonomy and performance.

Factor
Introduction and possible answers Item Factor Load Mode Median