Justice in Black and Brown: The impact of political control and representative bureaucracy on street level outcomes

This study tests the explanatory power of two prominent public administration theories — political control and representative bureaucracy — in understanding disparities in public service provision. While prior research focused on street-level bureaucracy, we study here how political and group identities of top elected law enforcement officials affect the distribution of youth arrests, alternatives, and prosecutions among minority groups. Data from Florida ’ s 67 counties between 2015 and 2020 demonstrate that sheriff ’ s and state attorney ’ s party affiliation, race, and ethnicity do affect street-level outcomes. However, the effects are more pronounced for race than ideology. Also, representative bureaucracy appears more relevant for race than ethnicity and explains the behavior of Black sheriffs but not Black state attorneys.

Evidence for practice • The social and political identities of top elected officials affect policy implementation at the street level.• Representative bureaucracy explains the distribution of youth arrests among minority groups better than political control.Yet, the partisanship of state attorneys explains prosecutions better than their group identity does.• Race appears more salient than ethnicity as a mechanism of representation among minority sheriffs.• Minorities are not monolithic-Black and Hispanic juveniles are subject to differential treatment and see disparate justice outcomes.
Prior research has extensively studied unequal justice outcomes among social groups, including police encounters, arrests, prosecutions, and court decisions (Headley & Wright II, 2020;Mitchell & Caudy, 2015;Stolzenberg et al., 2013).Scholars agree that social and contextual factors alone cannot explain these disparities (Bishop, 2005).
Research on representative bureaucracy has provided valuable insights into how street-level officers could produce favorable justice outcomes for social groups with whom they identify (Ding et al., 2021;Meier & Nicholson-Crotty, 2006;Theobald & Haider-Markel, 2009;Wilkins & Williams, 2008, 2009).Less understood is whether and how the characteristics of elected law enforcement officials influence street-level outcomes.This group of public officials warrants investigation because sheriffs and state attorneys belong to a class of officials known as elected executives (Miller, 2013).While selected by a popular vote, they perform administrative functions.As agency heads, they can influence street-level bureaucrats directly through the chain of command and indirectly through their political and social identities.Focusing on elected executives in law enforcement allows us to investigate how their partisanship and race/ethnicity affect bureaucratic behavior on the ground and assess two prominent public administration theories-political control of the bureaucracy and representative bureaucracy.
We examine these effects within the juvenile justice system, where outcomes often have life-changing consequences, and disparities among social groups at this level are perpetuated in the adult criminal justice system.To assess the distribution of justice among social groups, we study the overrepresentation of Blacks and Hispanics in arrests, alternatives (that is, diversion to other programs to avoid arrest), and prosecutions.Data are from all 67 counties in Florida (2015-2020), a state with significant racial diversity and differences in group political mobilization (Bishin & Klofstad, 2012;Meier & Stewart Jr, 1991;Rocha & Hawes, 2009).
The analysis shows that Black youth experience disparate justice outcomes compared to Hispanic youth.Demographic characteristics of the sheriff, especially their race, have a more pronounced effect on the disparities in Black youth arrests than the sheriff's ideology.Having a Black sheriff mitigates disparities in Black youth arrests; however, we document increasing disparities in Black youth prosecutions among counties with Black state attorneys.
The study makes three main contributions.First, it examines an important but understudied question about the impact of top elected officials' identities and whether these predict street-level bureaucratic outcomes in an essential local government service-justice provision.Specifically, the political affiliation and race/ethnicity of the most powerful law enforcement actors at the local level-the sheriff and state attorney-are associated with outcomes for minority youth.Although police officers make the actual arrests, their behavior corresponds with the messages they receive from the upper echelons of their agencies.
Second, top law enforcement officials are principals within their respective agencies and can affect the behaviors of their subordinates in direct and indirect ways.Studying elected executives allows us to compare the relative explanatory power of political control and representative bureaucracy.While the former assumes that the outcomes produced by frontline officers will reflect the political preferences of elected principals, the latter posits that the social identities of office holders will generate benefits for clients with similar social identities.Our analysis suggests that representative bureaucracy explains the distribution of Black youth arrests better than political control, and race appears more salient than ethnicity as a mechanism of representation among minority sheriffs.In contrast, the partisanship of state attorneys explains Black youth prosecutions better than their group identity does.
Third, the research examines the disparities by comparing the justice outcomes of the two main minority groups-Blacks and Hispanics-rather than between minority and majority groups.Our findings reveal that minorities are far from being monolithic-Black and Hispanic juveniles are treated differently and experience disparate outcomes in arrests, alternatives, and prosecutions.Such differences likely also reflect other differences in political power and social status.In this sense, our work speaks to the comparability of minority groups' experiences, with important implications for the provision of public services.
We next discuss the elected executive status of sheriffs and state attorneys and the power structure in Florida.The following section outlines the tenets of the two theories-political control and representative bureaucracy-identifies the potential pathways of influence for each theory, and presents hypotheses for the empirical tests.We continue with the research setting and why Florida is an appropriate test ground.After reporting the results, we discuss the implications of our findings and avenues for future research.

ELECTED EXECUTIVES IN LOCAL LAW ENFORCEMENT
Sheriffs and state attorneys belong to a class of public officials elected through a direct vote but charged with administrative responsibilities (Miller, 2013).Like legislators, they run for office and feel the pressure of electoral accountability, yet they administer public agencies like other top-level managers.
In Florida, sheriffs and state attorneys are part of the judicial branch.County sheriffs are responsible for enforcing the Florida Constitution, state laws, and statutes.Their role is to provide for citizens' security, safety, and well-being.According to Chapter 30 of Florida Statutes, the sheriff has exclusive power to appoint, promote, and terminate a deputy sheriff.The sheriff and the sheriff's deputies execute all processes of the Supreme Court, circuit courts, county courts, and boards of the county commissioners.
Florida's state attorneys are local officials responsible for appearing in circuit and county courts within their jurisdictions to prosecute or defend on behalf of the state all suits, applications, or motions, civil or criminal, in which the state is a party.Chief prosecutors appoint assistant state attorneys (state prosecutors) to serve at their pleasure.The assistant state attorney assumes all the powers and duties of the appointing state attorney.
In the juvenile justice system, when a minor violates the law, there are three possible outcomes of their encounter with a sheriff or sheriff's deputy.Depending on the crime's severity and the officer's discretion, the youth can receive a warning, an arrest, or a civil citation (an alternative to an arrest).States utilize various pre-arrest diversion initiatives.Prior research uses "alternative" as an umbrella term (Nadel et al., 2021).In Florida, an alternative equates to a civil citation.As Smith, Visher, and Davidson (1984, 234) explain, "[t] he discretionary nature of policing, coupled with the powerful implications of police discretion in the justice system, defines the primary issue of concern: whether decisions to arrest are influenced by suspect characteristics such as race, sex, and age." If the minor is arrested, the case is referred to the state attorney's office.The state attorney (or assistant state attorney) then decides whether or not to prosecute.Because prosecutors possess vast discretion, they are considered the most powerful actors within the justice system (Baker & Hassan, 2021).As Gordon and Huber (2009) put it, "criminal prosecutors exercise enormous discretion in coordinating investigations with law enforcement agencies, deciding which cases to bring to trial and conducting plea negotiations with defendants.They also enjoy considerable autonomy in setting the law enforcement priorities for the jurisdictions in which they operate" (136).Davis (2005) shares a similar sentiment about the prosecutors' role in the system, contending that life-anddeath decisions are "totally discretionary and virtually unreviewable" (56).
The two agencies constantly interact and influence one another.The State Attorney's Office can only prosecute the offenders referred by the Sheriff's Office and, thus, depends on who gets arrested.On the other hand, state attorneys send strong signals to law enforcement within their jurisdictions by deciding who to prosecute (Baker & Hassan, 2021).If the sheriff and sheriff's deputies know that the State Attorney's Office will not prosecute certain crimes, they will put less effort into arresting offenders for such violations.

THEORETICAL FRAMEWORK
Both political control and representative bureaucracy are mechanisms to reconcile public bureaucracies with democratic principles.Yet, they stem from different philosophies, follow distinct paths, and produce disparate predictions.

Political control of the bureaucracy
Political control of the bureaucracy, known also as overhead democracy, focuses on how elected officials can control the bureaucrats' actions.In a democracy, citizens elect representatives who set the policies and appoint or oversee agents (bureaucrats) to implement them.Political principals devise various mechanisms and procedures to ensure that agents enforce the policies congruent with the principals' preferences.The theory builds on the concept of political power, where elected representatives direct street-level bureaucrats to produce outputs and outcomes they might not otherwise (Calvert et al., 1989;Ringquist, 1995).
Bureaucrats who implement the policies on the ground, however, develop specialized knowledge that principals usually lack.This information advantage can be used to limit political control or hide actions from elected officials.In a study of the Environment Protection Agency enforcement activity after Reagan's inauguration, Ringquist (1995) demonstrates that political control can affect the agency's outputs-and seemingly make administrative actions compatible with the preferences of political principals-but cannot alter agency values.Overall, scholars find that administrative agencies do not necessarily operate as the theory of political control postulates (e.g., Balla, 1998;Meier and O'Toole, Jr., 2006;Wood & Waterman, 1991).
Studies of the police have documented some support for political control.Chaney and Saltzstein (1998) examine police responsiveness to direct orders and find that political principals influence the agents' behavior, even in settings that are hard to control.They found that state and local laws requiring an arrest for domestic violence effectively altered police practice.In this case, however, bureaucratic discretion coexisted with political control.Meier (1994) reports similar findings for drug enforcement, where principals led the bureaucracy in the direction it wanted to go anyway.
Because the county sheriff and chief prosecutor run for office, they are political actors with specific policy preferences and agendas.They are both principals and agents (Gordon & Huber, 2009)-principals to subordinates in the organization and agents to voters.Their subordinatessheriff's deputies and state prosecutors-are typical streetlevel bureaucrats (Lipsky, 1980), with vast discretion to implement laws.They operate "at the boundary between citizens and the state, and they profoundly shape the definitions of both through the actions they take and the norms they invoke" (Maynard-Moody & Musheno, 2000, 332).In the juvenile justice system, the frontline workers act as gatekeepers (Jackson, 2022).When deciding on the appropriate action, they consider the circumstances of the crime against the letter of the law.Their actions are also shaped by the organization's rules, informal culture, and their own beliefs (Scott, 1997).How do elected law enforcement officials affect justice outcomes on the ground?We argue that influence goes through direct and indirect channels.Figure 1 illustrates these pathways.
County sheriff's offices are hierarchical, often depicted as paramilitary organizations where sheriffs influence the rank and file directly through the chain of command.The same applies to state attorney offices, albeit the military comparison is less relevant.The direct orders of the elected executives in law enforcement agencies establish priorities that likely reflect their ideological positions previously expressed in their campaigns.Political control can also indirectly influence street-level justice outcomes through the partisan cues that deputies receive regarding the approach to crime.
Florida sheriffs and state attorneys are elected on partisan ballots.The two political parties have clearly marked out different approaches to law enforcement.Republicans are more likely to support a more punitive approach to juvenile delinquency, whereas Democrats subscribe to a more preventive and rehabilitative approach to juveniles.Republicans place a greater emphasis on individual responsibility (Farris, 2017), which translates to more arrests and prosecutions for those who violate the laws.Democrats emphasize societal issues that may lead to delinquent behavior.Electing a Democrat suggests that the community leans liberal, which translates into providing more alternatives to arrests investing in preventive measures such as after school programs and youth employment (Pickerell, 2020).In Florida, most sheriffs and state attorneys identify as Republicans.
Drawing on principal-agent theory, we expect non-Republican sheriffs and Democrat state attorneys to directly advocate for or indirectly signal a less punitive approach to crime to frontline workers than Republican sheriffs and state attorneys. 1Thus, our political control hypotheses explore the party differences in crime control for Black and Hispanic youth.The justice outcomes reflect the decisions made by law enforcement officers on whether to arrest a juvenile or issue a civil citation and by state prosecutors on whether to prosecute the detained youth.In effect, we have six hypotheses, organized by outcome and minority group:

Representative bureaucracy
Representative bureaucracy is another mechanism to ensure that administrators promote the public interest.The theory posits that the citizenry is better served when administrators share similar demographic characteristics (Kingsley, 1944).Representation is passive (descriptive) when it comes from shared demographic features such as sex, race, income, class, and religion, and active (substantive) when bureaucrats purposefully advance the interests of the group with which they identify (Mosher, 1968).Prior research has documented benefits for clients who share similar characteristics with frontline staff (Maynard-Moody & Musheno, 2003;Riccucci, 2005;Rutherford & Mee, 2022;Scott, 1997;Selden et al., 1998;Wilkins & Keiser, 2004).Representative bureaucracy provides a counterpoint to the principal-agent theory because it predicts how the bureaucracy would act in the absence of political pressure (Meier & O'Toole Jr, 2006).Long (1952) depicts it as a mechanism that heals a constitutional defect in representation and advances democratic principles (Selden et al., 1998).
Besides active and passive representation, symbolic representation reflects how others perceive the official.Theobald and Haider-Markel (2009) contend symbolic representation occurs at the cognitive level and affects the attitudes of and outcomes for the represented group just by the official holding a public post.Blacks perceive police actions as more legitimate if conducted by Black officers, and Whites consider police actions more legitimate when performed by White officers.Similarly, Black citizens consider the police to be better performing, trustworthy, and fair when the police force is composed mainly of Black officers (Riccucci et al., 2018; but see Headley et al., 2021).
In this study, we contend that the same signaling effect occurs for street-level bureaucrats and how they perceive their leaders.Sheriffs' and state attorneys' social identities serve as cues for the behavior of deputies and prosecutors on the ground.
Similar to political control, representative bureaucracy has direct and indirect pathways (Figure 2).In this case, the direct orders of agency heads would reflect their social identity preferences (minority representation) rather than ideological ones.Specifically, sheriffs and state attorneys can actively represent the groups with whom they identify by issuing directives or setting procedures that benefit those groups.
Indirectly, as suggested above, the social identities of agency heads send signals about the preferences of the communities that directly elected them.The election of a minority sheriff or state attorney indicates a community preference for greater inclusiveness and equity and has a symbolic impact at the street level without any deliberate actions on the part of elected officials.Instead, they stem from how street-level officers read the signal of their minority status.Thus, deputies and state prosecutors may be less rigid when deciding whether to arrest or whether to prosecute a youth from the same racial/ethnic background as the respective county sheriff or chief prosecutor.In turn, this may mitigate disparities in justice outcomes for minorities (Gunderson, 2022).The six hypotheses test the premises of representative bureaucracy, organized again by outcome and minority group:

The case for the null hypothesis
Although both the political control and representative bureaucracy literatures are well established and provide credible hypotheses, each considers a variety of factors that facilitate or limit the impact of each process.This section proposes that the case of juvenile justice in Florida involving sheriffs and state attorneys is a hard case for either demonstrating political control or finding the impact of bureaucratic representation.While hard cases are important to establish the generality of any theory, they also showcase the myriad other factors that can influence bureaucratic behavior.Hypothesis 1 holds only if the street-level bureaucratssheriff's deputies and state prosecutors in our case-are receptive to signals from their political bosses.A line of research paints street-level bureaucrats as rather autonomous from their superiors in the exercise of discretion (Brehm & Gates, 1999;Riccucci, 2005).Frontline workers often pursue their own goals, which might run contrary to the goals of their superiors (Lipsky, 1980).May and Winter (2009), for example, register only a weak influence of politicians and managers on policy implementation.They find frontline workers act in a way consistent with their own policy goals, professional knowledge, and policy predispositions.
The extant evidence regarding the effect of representative bureaucracy (Hypothesis 2) is even more nuanced.Substantively, the area of criminal justice and law enforcement has consistently found evidence of representative bureaucracy in the case of gender (Meier & Nicholson-Crotty, 2006;Shoub et al., 2021) but decidedly mixed or even negative findings for race (Gilad & Dahan, 2021;Headley & Wright II, 2020;Hong, 2017;Kennedy et al., 2017;Wilkins & Williams, 2008, 2009;Wright et al., 2022) except at very high levels of representation (Nicholson-Crotty et al., 2017).Few studies have examined the impact of law enforcement leadership, the focus of this study, and those have shown little impact (Johnston & Houston, 2018).How and when leadership matters remain an open question, particularly in the area of juvenile justice.
The literature on representative bureaucracy among prosecutors is significantly less extensive than that on law enforcement.While many studies show that Black and other minority defendants fare worse in prosecutorial decisions (McCoy et al., 2012), the results often reflect the disproportionate prior contacts with the criminal justice system (e.g., arrests, prior records).An experimental vignette study using several hundred actual prosecutors (Robertson et al., 2019) found no racial or class biases.We could not locate a single study that related the race of the prosecutor with racial outcomes; one study (Baker & Hassan, 2021) did show that more experienced female prosecutors were more likely to prosecute cases of domestic violence and sex offenses.
Several theoretical reasons suggest that sheriffs and state attorneys are cases where the race of the top individual might have limited impact.Minority elected officials face multiple cross-pressures that affect their ability to act as a racial representative.We identify several: • Identity pressures.Minority leaders face expectations to represent the group they come from, but as elected officials, they need to represent everyone.Socialization has been a long-acknowledged counterweight to the development of a representative bureaucracy (Wilkins & Williams, 2009).Sheriffs generally come through the ranks and may have spent decades in law enforcement, adopting the norms of the organization and profession.Similarly, state attorneys are trained as lawyers, and their professional reputation is linked to successfully prosecuting cases.Herbert (1974) indicates that minority managers must conform to the organization's demands, such as mission and orientation, to move up into leadership.Carroll et al. (2019) argue these organizational pressures are why minority school superintendents hold values and take actions similar to White superintendents.
All these factors, combined with the relative rarity of minority sheriffs and state attorneys (see Table 1), might only affect the racial distribution of outcomes at the margins, if at all, and support the null hypothesis credibility.

Research setting
Florida has had one of the most volatile records on race relations over the last 200 years.Examples include the Seminole Indian Wars, the enslavement of African Americans, school segregation, and, currently, mass incarceration of Black and Brown people (Alexander, 2012;Hankerson, 2003;Jackson, 2022;Rivers, 2000;Shofner, 1976;Work, 2001).Florida is the third-largest populated state, and in the last 5 years, saw an increase in the youth population, particularly among the Hispanic and Black populations.Whereas Black youth represent about 21% of Florida's youth in 2019-2020, they accounted for 46% of arrests and 49% of prosecutions.Prior research has acknowledged the overrepresentation of Black and Brown youth prosecuted after a police encounter.Per Stevenson (2014), "by 2010, Florida had sentenced more than a hundred children to life imprisonment without parole for nonhomicide offenses, several of whom were thirteen years old at the time of the crime.All of the youngest condemned children were thirteen or fourteen years of age and were Black or Latino, with Florida having the largest population in the world of children condemned to die in prison for non-homicides" (154).

Dependent variables
The study has three dependent variables-one for each outcome for the two main minority groups, Black, and Hispanic.The Florida Department of Juvenile Justice data reflect the main decisional junctures in the justice system: Whether a youth receives an arrest or an alternative, and whether the arrest results in prosecution or not.The first pair of dependent variables, Black Youth Arrests and Hispanic Youth Arrests, reflects the proportion of Black and Hispanic youth arrested by a county sheriff's office, respectively.For instance, in Black Youth Arrests, the numerator is the number of Black minors that received an arrest, and the denominator is the total number of youth arrests in a county.The second pair, Black Youth Alternatives and Hispanic Youth Alternatives, represents the proportion of Black and Hispanic youth receiving civil citations by a county sheriff's office.The third pair, Black Youth Prosecutions and Hispanic Youth Prosecutions, captures the proportion of Black and Hispanic youth prosecuted by a county/circuit state attorney's office.
Table 1 reports the descriptive statistics for all variables.Appendix A provides racial/ethnic groups as per the U.S. Census.Appendix B outlines the decisions of law enforcement and state attorneys when a minor is taken into custody.Appendix C contains the variables' names, operationalizations, and sources.Appendix D lists Florida counties and judicial circuits.

Main explanatory variables
To test the association of political control and representative bureaucracy with the distribution of outcomes, we code the party and racial/ethnic identity of county sheriffs and state attorneys.Specifically, Non-Republican Sheriff equals 1 when the sheriff identifies as a Democrat or Independent, is appointed, or has no party affiliation, and zero otherwise.Democrat State Attorney is 1, when the state attorney runs on a Democratic ticket, and zero otherwise.
Race/ethnicity are operationalized as dichotomous variables.Black Sheriff and Black State Attorney is 1 if these officials identify as Black, and zero otherwise.Similarly, Hispanic Sheriff and the Hispanic State Attorney take on a value of 1 if officials identify as Hispanic.

Control variables
To ensure that the effects of party and group identities on the distribution of justice outcomes are not driven by other factors, the models include a set of control variables.Prior research shows that bureaucrats' gender affects their behavior (Meier & Nicholson-Crotty, 2006).This may also be the case for women in leadership roles within law enforcement agencies or state attorney's offices.Thus, the models include two dichotomous variables: Female Sheriff and Female State Attorney.Although the examination of intersectionality and representative bureaucracy has produced significant findings in recent years (see Fay et al., 2021;Whitebread et al., 2022), there are too few female officials in Florida to examine the combined impact of race and sex.To account for organizational socialization of top law enforcement officials (Oberfield, 2010), we use Young Sheriff and Young State Attorney, indicators equal to 1 for job tenure less than 10 years and zero otherwise.
Minority groups might face disparate outcomes if the median voter shares a conservative ideology that prescribes a tougher approach to crime.To tap the median voter preference (Clifford 2019), we use the percent of Trump Vote in the 2016 election.Prior research also shows that communities with lower educational attainment are likely to experience more crime (Claus et al., 2018).Education reflects the percentage of adults above 25 with a high school diploma.Relatedly, racial disparities in arrests increase in high-poverty areas (Wilkins & Williams, 2008).Our models include the Poverty Rate, measured as the percent of individuals below the poverty line.Extant research also links the area's crime rate with more instances of racial profiling (Smith & Petrocelli, 2001).Thus, we add to the model the natural logarithm of the number of crimes per 100,000 residents (Crime Rate).
Finally, we control the community makeup.Earlier research demonstrates that the racial composition of the population affects the perceptions, attitudes, and behavior of police officers (Meier & Nicholson-Crotty, 2006).To account for these effects, we include the percent of Black and Hispanic residents in a county-Black Population and Hispanic Population.

Estimation routine
Our analyses use data from the 67 Florida counties from 2015 to 2020 for a total of 335 county-year observations.Panel data sets can violate the assumptions of ordinary least squares due to both serial correlation and heteroscedasticity.We include year fixed effects in all models to account for any trends in the data.Heteroscedasticity can be particularly problematic in cross-sectional dominant panels with highly divergent units.Florida counties vary greatly by population, racial composition, urbanicity, and a variety of other factors, which could potentially produce a non-Gaussian distribution that is susceptible to outliers and high leverage points that might bias our inferences.To deal with this concern, we employ a robust regression estimator in STATA that uses an iterative weighting process that combines Huber weights and biweights.The results converge to OLS estimates if the data distribution meets the assumptions of the latter.

Disparities in justice outcomes among minority groups
To assess the disparities in the distribution of arrests, alternatives, and prosecutions among minorities, we regress the justice system outcomes for each youth group on its respective population in a county (Black Youth and Hispanic Youth).If the coefficient is close to or equal to 1, the outcomes are proportional to the population.A slope coefficient greater than 1 suggests that the group over-experiences the outcome, while a slope less than 1 means that the group under-experiences that outcome.The process implicitly assumes an equal rate of offenses among social groups.We denote the difference between 1 and the point estimate of the coefficient as disparity and interpret our results using this term.
The results point to significant disparities in justice outcomes among minority groups.Table 2 shows the overrepresentation of Black youth in arrests, alternatives, and prosecutions, in Models 1, 2, and 3, respectively.Model 1 demonstrates that, all else equal, a 1% point increase in Black Youth is associated with a 1.54 percentage point increase in Black Youth Arrests.In other words, the disparity for Black youth is 54% in terms of arrests.Model 2 shows that all else equal, a one percentage point increase in Black Youth is associated with a 1.80 percentage point increase in Black Youth Alternatives, that is, Black youth are 80% more likely to experience alternatives than their population proportion would suggest.Finally, Model 3 probes for disparities in prosecutions of arrested Black youth and indicates that, all else equal, a one percentage point increase in Black Youth is associated with a 1.63 percentage point increase in Black Youth Prosecutions.
Models 1-3 in Table 3 examine disparities in the justice outcomes for Hispanic youth.Unlike for Blacks, the results show that Hispanic youth are underrepresented along all three outcomes-the point estimates of the coefficients of Hispanic Youth are less than 1 in all three models.Specifically, Model 1 reveals that for each percentage point increase in Hispanic Youth there is a 0.67 percentage point increase in Hispanic Youth Arrests, an underrepresentation of 33% relative to their population.Hispanic juveniles also receive disproportionately fewer civil citations (Model 2).The coefficient of 0.49 implies underrepresentation by 51% in terms of alternatives.Finally, Hispanic juveniles are underrepresented in prosecutions, as the coefficient for Hispanic Youth in Model 3 is 0.73.

Testing the effects of political control and representative bureaucracy
To compare the explanatory power of political control theory and representative bureaucracy, we test if the outcomes for minority groups are conditional on political and group identities of top law enforcement officials.We interact Black Youth and Hispanic Youth with the variables capturing the sheriff's and state attorney's party affiliation and race/ethnicity.Models 4-6 in Tables 2 and 3 report the results for Black and Hispanic youth, respectively.Starting with Black youth (Table 2), the sheriff's political affiliation is not related to disparities in Black youth arrests and alternatives-neither of the coefficients of the interaction terms Black Youth Â Non-Republican Sheriff in Models 4 and 5 is significant.Yet, we document a statistically significant effect for state attorneys (Model 6) in the expected direction.The coefficient of Black Youth Â Democrat SA is À0.43, indicating a 43% decrease in the disparity of Black youth prosecutions when the state attorney is a Democrat.In other words, in counties with a Democrat state attorney, the prosecution rate for Blacks drops to 1.38 versus 1.81 elsewhere.
Regarding the effects of representative bureaucracy, our estimations show a strong representative effect for Black sheriffs and the opposite effect for Black state attorneys.The interaction coefficient of Black Youth Â Black Sheriff indicates that counties with a Black sheriff have a statistically significant lower arrest rate.Specifically, in counties with a Black sheriff, Black youth are only 14% more likely to be arrested relative to their representation in the overall population, compared to 79% in other counties.Figure 3 illustrates this effect.
Having a Black sheriff does not affect the distribution of alternatives to Black youth-the interaction coefficient fails to reach significance at conventional levels.Yet, having a Black state attorney has a significant impact on the rate of prosecution of Black youth that runs contrary to the expectations of representative bureaucracy theory.In Model 6, the coefficient of the interaction term Black Youth Â Black SA is 1.57 and significant at the 1% level; Black youth are 57% more likely to be prosecuted if there is a Black state attorney than in other counties.Figure 4 visualizes the effect.
The results in Table 3 allow us to compare the explanatory power of agency theory and representative bureaucracy for Hispanic youth.Having a non-Republican sheriff does not affect the rate of Hispanic youth arrests, as suggested by the insignificant, albeit negative, interaction term coefficient in Model 4. Yet in counties with a non-Republican sheriff, Hispanic juveniles receive alternatives at a higher rate than in counties with other sheriffs.The interaction coefficient Hispanic Youth Â Non-Republican Sheriff in Model 5 is 0.22.Counties with a non-Republican sheriff have an alternative rate for Hispanic youth of 0.76 versus 0.54 in other counties.Finally, having a Democrat state attorney does not affect the prosecution rate of Hispanic youth-the coefficient of the interaction term in Model 6 is positive but not significant.
We find no support for representative bureaucracy based on Hispanic identity.None of the interaction coefficients for Hispanic Sheriff or Hispanic SA in Models 4-6 is significant.
Given that state attorneys can prosecute only already arrested youth, we estimate the prosecution models while controlling for arrests within each minority group.Table 4 reports the baseline specification and the models with interactions for Black youth (Models 1 and 2) and Hispanic youth (Models 3 and 4).These estimations allow us to focus on the discretion of state attorneys in the justice process.The results paint a slightly different picture than those in Tables 2 and 3. Now the coefficient for Black Youth in model 1 is 0.83, indicating that a one percent increase in Black Youth is associated with 0.83% Black Youth Prosecutions, which is an underrepresentation of F I G U R E 3 Predicted Black youth arrests at the level 17%.For Hispanic youth, the underrepresentation in prosecutions is 77% when controlling for arrests (Model 3).
From Model 2, we infer that Democrat state attorneys exhibit a significant effect on the rate of prosecutions experienced by Black youth.The coefficient of À0.15 on the interaction term Black Youth Â Democratic SA (statistically significant at the 10% level) shows the rate decreasing to 0.77.The rate of Hispanic youth prosecutions is also lower when the state attorney is a Democrat, as indicated by the negative and statistically significant coefficient of À0.07 on the interaction term Hispanic Youth Â Democrat SA in Model 4, a reduction from 0.20 to 0.13.
In terms of representative bureaucracy, we note again significant effect of having a Black state attorney, which runs contrary to the theoretical expectations.In Model 2, the interaction term Black Youth Â Black SA is 0.75, indicating a higher prosecution rate, 1.67 versus 0.92 in other counties.Figure 5 depicts this effect in graphical form.As with the previous models, we find no support for a representative bureaucracy for youth, as indicated by the insignificant coefficient of the interaction term Hispanic Youth Â Hispanic SA in Model 4.
Among the control variables, counties with female sheriffs exhibit statistically significant higher rates on the three justice outcomes for Black youth in Table 2.This result indicates no reduction in disparities among sheriffs based on gender.Table 5 summarizes the main findings.

DISCUSSION AND CONCLUSION
Florida has a challenging past in race relations and inequities among people of color.Our analysis demonstrates the presence of significant disparities for Black youth in arrests and prosecutions.This group also receives more alternatives in the form of civil citations.In contrast, Hispanic youth are not over-represented in the three outcomes compared to their population numbers.Overall, the distribution of justice outcomes varies among minority youth, with Black juveniles seeing more pronounced disparities.
The data also provide more support for representative bureaucracy than political control for the outcomes produced by the Sheriff's Office, but not for the State Attorney's Office.Black sheriffs are associated with fewer arrests of Black youth and their presence in the county continues through the judicial process with fewer prosecutions for Black youth.Interestingly, the effect of Black state attorneys works in the opposite direction-Black youth get prosecuted at a higher rate, which persists even after controlling for the arrests, indicating that the higher prosecution rate is attributable to the Black state attorney.
We find some effect of political control.Counties with Democrat state attorneys experience fewer disparities in prosecutions among Black youth than counties with Republican state attorneys.This effect remains after controlling for Black youth arrests.We also register a similar impact for Hispanic youth-Democrat state attorneys decrease the rate of prosecutions, but only after controlling for the arrests.Thus, the evidence supports Hypotheses 1C 1 and 1C 2 about the effect of the state attorney's party affiliation.
Although the evidence supports only one out of the six hypotheses about the effect of the sheriff's and state attorney's race/ethnicity on disparities in youth justice outcomes (Hypothesis 2A 1 ), it is solid and consistent.Counties with Black sheriffs experience fewer disparities in arrests among Black youth than other counties, consistent with the expectations of representative bureaucracy.Given that top law enforcement officials need to represent everyone in the community, demonstrate considerable output before the electorate, and adapt to the organizational norms of their agencies, it is unlikely that the effect of their social identity occurs through direct pathways.Pursuing policies that benefit specific groups might cost them the office in the next election.In this sense, our finding that counties with Black sheriffs experience fewer Black youth arrests is more consistent with indirect pathways.While disentangling the specific underlying mechanism remains an avenue for future research, the effect on Black sheriffs likely occurs through symbolic representation, whereby deputies alter their behavior based on the identity of the sheriff.Finally, representative bureaucracy appears more relevant for race than ethnicity.While Black sheriffs are associated with lower Black youth arrests rates, there is no association for Hispanic sheriffs and Hispanic youth.The analysis also produced null results for Hispanic state attorneys.Representation does not work the same way for all social groups, and being Hispanic is not the same politically as being Black (Meier et al., 2004;Rocha & Hawes, 2009).Minorities experience differential treatment, and the justice outcomes between Hispanics and Blacks are far from similar.Yet, much of the extant literature on justice provision among Blacks and Hispanics treats minorities as monolithic (Stevens & Morash, 2015).Blacks are more cohesive as a social group.They are more likely to share similar values and have a stronger sense of linked fate (Dawson, 1994).Politically, they have been more active in pressing their concerns (Meier & Stewart Jr, 1991).
Hispanics are less cohesive as a social group; many are first-generation immigrants, with their values reflecting the different national origins and processes of arrival in the United States (Monforti & Sanchez, 2010).In this F I G U R E 5 Predicted Black youth prosecutions at the county level when controlling for Black youth arrests sense, the Hispanic community is much more heterogeneous, consisting of multiple populations with Cuban, Puerto Rican, Mexican, South American, or Central American origins (U.S. Census, 2020).At the same time, the Cuban American community is especially prominent, conservative, and rapidly moved into positions of political power (Bishin & Klofstad, 2012).The lack of outcome disparities is likely to lessen representation pressures, whether from representative bureaucracy or political control.
This study has implications for the theory and practice of public administration.First, the finding that the race of top elected officials affects the outcomes of the juvenile justice system indicates the need to account also for principals' identity as well as agents' identity.Such studies likely have ramifications for other government agencies and for non-profit organizations that deliver public services (LeRoux & Medina, 2022).Second, we find that sheriffs affect street-level outcomes more through race than ideology; that is, the theory of representative bureaucracy explains street level arrests better than political control.Yet, the state attorney's partisanship influences frontline workers' behavior more than their racial/ethnic identity.Further research is warranted to understand why the effect changes at different points in the juvenile justice system.Third, race appears to be more salient than ethnicity as a mechanism of representation among minority sheriffs in Florida.Future studies could verify these inferences within other policy and geographical settings.Lastly, our results show that the minorities are not a monolithic group.Rather, minority groups are subject to different experiences and disparate outcomes.In justice provision, Hispanics share more similar experiences with Whites than Blacks, likely due to Florida politics.Our findings underscore the need for practitioners to balance the scales in service provision and for scholars to explain better the disparate dynamics across minority groups.

ORCID
William T. Jackson https://orcid.org/0000-0003-4862-0260ENDNOTE 1 Non-Republican sheriffs are those who identify as Democrats, Independents, are appointed, or have no party affiliation.Within our sample, there are three independent sheriffs, one appointed sheriff, and four with no party affiliation.All state attorneys are politically affiliated and identify as either Democrat or Republican.
Sheriff/ State Attorney Direct: Hierarchical Chain of Command Street Level Bureaucrats' Behavior Indirect: Party Cues F I G U R E 1 Political control of the bureaucracy: Direct and indirect pathways Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13600,Wiley Online Library on [31/01/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Counties with non-Republican sheriffs will experience fewer disparities in arrests among Black youth than counties with Republican sheriffs.Hypothesis 1A 2 .Counties with non-Republican sheriffs will experience fewer disparities in arrests among Hispanic youth than counties with Republican sheriffs.Hypothesis 1C 2 .Counties with Democrat state attorneys will experience fewer disparities in prosecutions among Hispanic youth than counties with Republican state attorneys.
Counties with Black sheriffs will experience fewer disparities in arrests among Black youth than other counties.
As elected officials, sheriffs and state attorneys are also constrained by the need to win elections, often from a conservative electorate voting in local races.Voters expect sheriffs to enforce laws against all violators and prosecutors to win convictions.•Organizational and socialization pressures.Law enforcement agencies rely heavily on organizational socialization for uniformity in policy implementation.
Sheriff/ State Attorney Direct: Active Representation Street Level Bureaucrats' Behavior Indirect: Symbolic Representation F I G U R E 2 Representative bureaucracy: Direct and indirect pathways T A B L E 1 Descriptive statistics • Electoral pressures.
Predicting justice outcomes for Black youth Predicting justice outcomes for Hispanic youth T A B L E 2 Note: The table contains the coefficients from a robust regression estimation, standard errors in parentheses.*p < .1.**p < .05. ***p < .01.T A B L E 3 Note: The table contains the coefficients from a robust regression estimation, standard errors in parentheses.*p < .1.**p < .05. ***p < .01.
F I G U R E 4 Predicted Black youth prosecutions at the county level Predicting prosecutions while controlling for arrests 15406210, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13600,Wiley Online Library on [31/01/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License T A B L E 4 15406210, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13600,Wiley Online Library on [31/01/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License T A B L E 4 (Continued) The table contains the coefficients from a robust regression estimation, standard errors in parentheses. Note: