Representative Bureaucracy and School Discipline

Abstract

This article examines whether the racial context within local communities influences the assignment of disciplinary policies in public schools. First, we consider whether different policies may be assigned to similar target groups across varying racial contexts. Then, we consider whether the racial context moderates the transition from passive representation to active representation among bureaucrats. We draw from two theories of intergroup relations—group contact theory and group threat theory—to help explain the passive-to-active representation link. Using a sample of Georgia public schools, we find that schools rely more on more punitive disciplinary measures in school districts characterized by greater segregation and that this occurs especially among schools with sizable African American student populations. We also find that active representation appears to occur more often in segregated environments, perhaps because of the greater salience of race within these communities.

Keywords

representative bureaucracy school discipline segregation racial context

A representative bureaucracy may support practices that lead to more effective outcomes for clients (Hindera, 2004; Meier, Eller, Wrinkle, & Polinard, 2001; Meier, Wrinkle, & Polinard, 1999; Pitts, 2005; Selden, 1997a). This appears most likely to occur when we see the transition from passive representation, where bureaucrats and targets simply share characteristics linked to attributes such as race or gender, to active representation, the case in which bureaucrats choose to act in favor of those individuals who share similar attributes (Mosher, 1982). The benefits of such representation do not necessarily reflect a zero sum game; they may extend to represented minority groups as well as potentially toward other service recipients (Keiser, Wilkins, Meier, & Holland, 2002; Rocha & Hawes, 2009; see also Nicholson-Crotty, Grissom, & Nicholson-Crotty, 2011).

Past research on representative bureaucracy has identified a number of conditions under which this transition from passive representation to active representation is most likely to occur, such as when individual characteristics shared with target groups are salient to bureaucrats and when they have discretion within their jobs while engaging in policy design or implementation (Meier, 1993b). We, however, have only limited information about how the racial context may affect the occurrence of active representation (see Grissom, Nicholson-Crotty, & Nicholson-Crotty, 2009). Previous research has demonstrated the role of contextual factors, such as the degree of racial segregation or the presence of sizable minority groups, in shaping attitudes regarding race (e.g., Key, 1949; Oliver & Wong, 2003; Taylor, 1998). In this research, we are interested in how the racial context may interact with the bureaucrat’s own discretion to represent minorities’ interests, affecting the likelihood of the passive to active transition.

Previous research has emphasized how race-salient policies, such as education policy, provide arenas in which bureaucrats may more actively represent their clients (Meier, 1993b). Grissom and his colleagues, however, recently showed that the salience of race may also vary across regions, affecting the likelihood of active representation. According to these researchers, active representation appears more likely among bureaucrats in the southern part of the United States, in part due to the greater salience of race in the South that occurs because of the history of racial conflict within this region (Grissom et al., 2009). In this article, we argue that the salience of race will also vary across local communities in part because of their changing racial contexts, which help determine how race is experienced within these communities.

We focus on segregation, a contextual factor that has been the subject of a significant volume of previous research exploring its effects on economic, social, and health-related outcomes (for a review see Carr & Kutty, 2008). Drawing on previous research on racial contact and racial threat theory, we expect that race should be more salient to individuals in segregated communities, because there is likely to be more limited contact among racial groups causing other groups to appear threatening and prejudicial attitudes to increase (Oliver, 2010; Rocha & Espino, 2009). In these cases, the personal beliefs and values held by bureaucrats should also become more relevant, facilitating the transition from passive representation to active representation.

We investigate the relationship between segregation and the likelihood of active representation while focusing on discipline in public schools. Discipline policy in schools works to limit the extent to which students are able to disrupt the schools’ learning environments and generally involves more and less punitive policies. For instance, some discipline policies may support the use of strict sanctions, whereas others will involve the use of policy tools, or ways of disciplining students, that are more capacity building and learning oriented (Schneider & Ingram, 1997).¹ As discipline policies vary in their design, they may involve different types of “rules, tools, rationales, causal logic, and messages [for students]” (Ingram, Schneider, & DeLeon, 2007, p. 96). These policies may also vary in whom they are most likely to target and in their outcomes for students. For example, the use of some disciplinary practices, such as expulsions, can have a negative effect on student achievement (Brown, 2007), and evidence suggests that these types of actions tend to be employed disproportionately in cases of racial and ethnic minorities (Skiba & Peterson, 1999; Verdugo, 2000; Verdugo & Glenn, 2002). Other discipline policies, such as in-school suspensions, are less disruptive to students and are more likely to keep students from falling behind in their academic studies (Adams, 1992; Shah, 2012; Short, Short, & Blanton, 1994). We believe that this is a particularly useful policy area in which to test our research questions as the selection of these policy tools is determined by local public officials who should be influenced by local conditions.

We see this research as making several contributions. First, we contribute to the literature on policy design, by examining whether different policies may be assigned to similar target groups across different contexts. Second, we also contribute to the literature on representative bureaucracy by demonstrating the role that contextual factors have on the transition from passive representation to active representation.

We begin by examining racial contact and racial threat theory. Next, we consider the role of representative bureaucracy and past research that has explored the role of contextual effects in the transition from passive to active representation. We then turn to consider some past research on discipline within public schools and present a number of hypotheses that we address empirically in the following section, relying on data from Georgia public schools. We consider the implications of our findings in the article’s conclusion.

Racial Contact and Threat

The group contact thesis argues that isolation from other racial and ethnic groups leads to prejudice and ignorance and increases an individual’s own-group bias. Intergroup contact, on the other hand, reduces racial “fears of the imagination” (Allport, 1954, p. xv), breaks down negative stereotypes, “weaving together diverse people and groups” (Emerson, Kimbro, & Yancey, 2002, p. 758) and creates more accepting attitudes (Pettigrew, 1997). A meta-analysis of more than 500 group contact studies revealed that 95% of them found that intergroup contact reduced prejudice (Pettigrew & Tropp, 2006), even when that contact is just casual (Dixon & Rosenbaum, 2004).

One way that researchers have examined group contact is by linking it to residential segregation (Rocha & Espino, 2009). Research has shown that segregation reduces the likelihood of frequent, varied, and regular intergroup interaction—conditions that are generally considered to improve intergroup relations (Welch, Sigelman, Bledsoe, & Combs, 2001). For instance, Oliver and Wong (2003) showed that in integrated neighborhoods individuals had less negative impressions of other racial groups than in segregated neighborhoods. As stated by Oliver (2010) “ . . . most folks who live in segregated neighborhoods or suburbs have few opportunities, at home, at church, in their civic lives, or even at work, for positive interracial contact” (p. 105). Even on accounting for the possible causality problem that more prejudiced individuals elect to move to more segregated areas (Pettigrew, 1998; Zubrinsky & Bobo, 1996), most evidence suggests that individuals, regardless of how prejudiced they were initially, become less prejudiced over time when living in integrated areas (Christ et al., 2014; Pettigrew & Tropp, 2006).

Research suggests that these varied attitudes present in more and less segregated communities may translate into different policy beliefs. Authors (2008) showed that voters in racially segregated Alabama counties in 2003 expressed lower support for two redistributive policies—additional public education revenues and a more progressive tax system—than voters in less segregated counties. Residing in more segregated neighborhoods has also been shown to influence Whites’ opinions about policies they see as targeting African Americans—such as aid to big cities or open housing (Kinder & Mendelberg, 1995).

The presence of segregation in a community may also heighten the likelihood of intergroup threat. Intergroup threat considers the threat posed to the majority group by sizable minority groups (Blalock, 1967; Giles & Evans, 1985; Key, 1949; King & Wheelock, 2007; Quillian, 1995). Threat to the majority group’s dominant status from minority groups has been linked to greater prejudice and own-group bias (Blumer, 1958; Campbell, 1965). Some researchers suggest that it is the threat of Black crime, as opposed to an economic or political threat, on the majority group that has the most significant effect on the degree of social control imposed on Blacks within a community (Eitle, D’Alessio, & Stolzenberg, 2002). The lack of intergroup contact in segregated environments can make sizable minority groups appear more threatening (Schlueter & Scheepers, 2010). For instance, Rocha and Espino (2009) showed that Whites who lived in more segregated areas were less likely to hold Latino-friendly attitudes, especially in areas of high Latino concentration, than Whites who lived in more integrated areas.

Recent research has also applied the group threat perspective to understanding school discipline, particularly as it relates to the disproportionate use of punitive discipline among African American students (e.g., Kupchik, 2009; Payne & Welch, 2010; Welch & Payne, 2010). Welch and Payne (2010) identified discipline as a form of social control that is mediated by administrators’ perceptions of racial threat, which they linked to the percentage of African American students present in a school. According to these authors, the

underlying threat of black criminality perceived at the local level by policy makers, school district board members, school administrators, and teachers may lead to the enactment, implementation, and enforcement of harsher disciplinary policies and practices in schools with greater number of black students because of their increased proportions. (p. 30)

Thus, Welch and Payne (2010) argued that as the percentage of Black students increase, schools increasingly rely on more punitive disciplinary measures, independent of actual student misbehavior, due to the increased presence of racial threat.

Thus, we expect the degree of segregation will influence how positively or negatively minority students are viewed in their communities, where higher degrees of segregation are likely associated with lower levels of intergroup contact, increased interracial threat, and more prejudicial attitudes. We also expect the heightened racial salience present within more segregated communities to influence the actions of teachers and administrators. We develop our expectations about the direction of this influence in the following discussion.

Representative Bureaucracy

Representative bureaucracy research examines whether the employees of public organizations match the general population according to individual characteristics such as race, ethnicity, or gender (for a review, see Krislov, 2012). As we have already mentioned, one important distinction in this research is between passive representation and active representation (Mosher, 1968, 1982). The passive-to-active transition is based on the notion that values and beliefs are likely to be shared by public officials and members of the public who have similar characteristics, which leads to similar policy preferences and increases the likelihood of more beneficial policy outcomes when active representation occurs (Bradbury & Kellough, 2008; Meier, 1975; Park, 2014; Selden, 1997b). Research in public administration has found positive effects of representation in areas such as education (e.g., Meier, O’Toole, & Nicholson-Crotty, 2004) and law enforcement (e.g., Meier & Nicholson-Crotty, 2006). Yet, some scholars question whether bureaucratic representation may be in fact democratic and argue instead that bureaucrats should maintain impartiality in their implementation of public policies or they may pose a “threat to orderly democratic government” (Mosher, 1968, p. 12; see also Lim, 2006). Some research has found negative cases of redistribution as a result of representation—where benefits provided to one group appear to detract away from those enjoyed by others, providing what opponents would claim is a means of reverse discrimination (Lim, 2006; Pitts, 2007). Others, however, have shown that these zero sum redistributions do not always occur and that threats to democratic values through active representation may be less consequential given that representation may help to rectify inequities, such as in the underrepresentation of minorities in Gifted and Talented programs in public schools (Meier et al., 1999; Nicholson-Crotty et al., 2011).

Overall, research within public administration shows that passive representation leads to active representation in a number of policy areas and provides benefits for disadvantaged groups (see Selden, 2006). Research on representation from areas outside of public administration, such as sociology and education, has also demonstrated how minority representation produces benefits for minority clients and reduces perceptions of organizational discrimination (Goldsmith, 2004; Hirsh & Kornrich, 2008).

Research suggests, however, that the transition from passive to active representation may not always occur. According to Meier (1993b), scholars generally have agreed on several criteria that need to be present for the transition from passive representation to active representation to happen, such as when shared demographic traits are salient to bureaucrats and to the policy area in which bureaucrats work and when bureaucrats have discretion within their jobs (see also Keiser et al., 2002; Selden, 1997b). Research has also identified other factors that may affect this transition. Meier and Stewart (1992) and Meier (1993a) found greater evidence of active representation among street-level bureaucrats than among managerial-level administrators. From their studies of Florida schools, they concluded that socialization to professional norms may inhibit the transition from passive to active representation among administrators. Lim (2006) suggested that active representation is likely when bureaucrats perceive the benefits to the represented group as outweighing the costs. The costs of active representation may be high, for instance, when the only avenue of active representation bureaucrats see as available involves penalizing other groups as bureaucrats actively representing other groups may also adopt such behaviors. Other research has demonstrated that the transition from passive to active representation is more likely when officials see themselves as advocates of minority rights or needs, assuming a “minority representative role” (Selden, 1997a; Selden, Brudney, & Kellough, 1998). While minority status and assuming a minority representative role are closely correlated, non-minorities could also adopt such a role because of their background or socialization. Selden’s research suggests that it is the adoption of this role, rather than race itself, that leads to the active representation of minorities’ interests (Bradbury & Kellough, 2008; Selden et al., 1998; Sowa & Selden, 2003).

Some past research has also examined the influence of the contextual environment on the transition from passive to active representation. Some of this work has focused on how the structure of the organizational environments in which bureaucrats work affects this transition (see Grissom et al., 2009). For instance, Meier and Bohte (2001) found that active representation is more likely in organizations that provide their employees with greater discretion from administrative authority, and Keiser et al. (2002) showed that active representation for women was more likely to occur in schools with less stratification by gender and less hierarchy. Other research has shown how factors such as a state’s economy or partisanship influence the behavior of bureaucrats and the degree of discretion available to them, though not specifically the change from passive to active representation (Keiser, 1999).

A few studies have also explored the relationship between the racial context and the transition from passive to active representation.² As mentioned previously, Grissom and colleagues (2009) found a higher degree of active representation among Black school teachers in the South. Thus, they argued that the region of residence moderates the transition from passive to active representation and that this occurs because of the higher degree of racial tension in the South, causing race to be more salient to Southern bureaucrats. Preuhs (2006), with findings in the opposing direction, found that Black passive representation in state legislatures is less likely to promote active representation in polarized political contexts (e.g., South vs. non-South) in which racial cleavages are emphasized. Other research has accounted for socioeconomic contextual factors at a more local level. Rocha and Hawes (2009) explored the effects of representation among minority teachers on outcomes for minority students, while controlling for contextual factors within school districts. These authors found positive evidence for representation on minority student outcomes, and they found significant contextual effects—greater diversity within school districts and increased per capita income among minorities relative to Whites improved outcomes for these students. Unlike Grissom and colleagues and Preuhs, these authors do not consider whether representation is more or less likely to occur within these varying socioeconomic contexts.

We build on this previous research by examining whether contextual differences across communities moderate the transition from passive to active representation within schools. We expect that in less segregated environments, race will play a more diminished role and that a range of values, including personal values as well as those drawn from the school and local environment, will influence the behavior of bureaucrats (see Schneider & Ingram, 1997). In these cases, active representation may occur according to a number of possible factors, such as gender, class, or race, or bureaucratic behavior may simply be passively representative, constrained by school and community-wide norms. In more highly segregated environments, race will likely become more salient to bureaucrats and race-oriented disparities present in these areas may cause the potential benefits of active representation to outweigh the costs, increasing the likelihood of active representation by race or by those individuals taking on a minority representative role. We now turn to place our research questions more firmly within the disciplinary context.

Schools and Student Discipline

Public schools provide an environment in which there is likely to be a transition from passive representation to active representation. Teachers and administrators generally should have enough discretion within their work environment to initiate policies that may be partial to students that have like ethnic identities. In addition, race and ethnicity are salient to many of the policies enacted within an education environment. For instance, laws, such as No Child Left Behind (2001), require that schools analyze student performance according to racial subgroups, and research also often examines whether the performance of minority students is on par with that of non-minority students. Findings from student assessments such as National Assessment of Educational Progress (NAEP) also often point to the existence of a Black–White achievement gap, which has increasingly drawn public attention (Rothstein, 2004). As a result, we expect that race should be salient to many bureaucrats within schools.

Past research has already shown benefits to students that occur in representative environments within academic settings. For example, representation in public schools has been shown to relate to higher test scores for students of color (Meier et al., 2001; Meier et al., 1999), higher academic achievement for female students (Keiser et al., 2002), and fewer students dropping out of school (Pitts, 2005). Past research has also demonstrated the influence of representation on the types of discipline used within schools (Authors, 2010).

When we look at student discipline in our work, we follow this past research in distinguishing between those policy tools designed to support learning and capacity building, such as in-school suspensions, and those more closely associated with potentially stigmatizing sanctions (Schneider & Ingram, 1993). In-school suspensions generally keep students in the school’s environment and may provide rehabilitation through counseling, resulting in a more complex and costly disciplinary technique for schools (Lawrence & Olvey, 1994; Short et al., 1994). In contrast, students are removed from the educational environment once they receive an out-of-school suspension. The use of these more punitive punishments in recent years has increased, as more schools have initiated “zero-tolerance” policies, leading to automatic suspensions or expulsion in response to certain behaviors (Kaufman et al., 2001). As we have mentioned, research suggests that more punitive disciplinary policies disproportionately target some groups of students, particularly minorities. For instance, African American students in 2011 received out-of-school suspensions at roughly 3 and 2 times the rates of Whites and Hispanics, respectively, nationwide (U.S. Department of Education, 2014).

One possible explanation for this difference lies in the potential cultural barriers between administrators, teachers, and students. Past research suggests that the actions of minority students may be misinterpreted by school staff due to racial stereotypes or unfamiliarity with interactions patterns among students of color (Goff, Steele, & Davies, 2008; Skiba, Michael, & Peterson, 2002). These differences may influence teachers’ behaviors within the classroom. Research has shown that African American students are more likely than other students to receive disciplinary referrals from teachers (Skiba et al., 2011) and that these referrals often occur for more subjective reasons (Vavrus & Cole, 2002). Administrators, who often determine disciplinary actions after referrals, also assign African American students suspensions more often for subjective reasons such as being disrespectful or appearing threatening (Skiba et al., 2011).

When we look at Georgia, our chosen research site, we expect that teachers and administrators will have flexibility in their assignments of disciplinary referrals and disciplinary actions to students. Discipline in Georgia is guided by state statutory provisions that generally prescribe a “progressive discipline process” that reserves out-of-school suspensions for the most disruptive students (Georgia Department of Education, 2013). Aside from automatically applying zero tolerance to certain types of incidents (e.g., firearm, bullying), much of the development and implementation of student discipline policy falls to local control, affording teachers and administrators the ability to handle day-to-day disciplinary challenges (Georgia Appleseed Center for Law and Justice, 2011).

We expect that negative assessments of students will be more likely generally when schools are in more segregated local environments, where limited interracial contact may lead to the increased presence of prejudicial attitudes. Larger groups of minority students may also appear more threatening in more segregated environments. In these cases, teachers and administrators may increasingly view larger groups of minority students as threatening the social order within schools. As a result, we hypothesize as well that more punitive punishments will be used more frequently within schools in segregated environments with larger groups of minority students.

As we have already discussed, we also expect the increased salience of race to influence the likelihood of active representation. Thus, we also hypothesize that we will be more likely to see a transition from passive representation to active representation in local communities where there is greater segregation. In these cases, the racial match between teachers or administrators and students should help to mute the negative outcomes driven in part by the environment and provide more beneficial outcomes for minority students through the enactment of more rehabilitative policy tools.

Data

We use data from public schools in the state of Georgia for the 2008-2009, 2009-2010, and 2010-2011 school years provided by the Georgia Department of Education (GDOE). Given the low rate of disciplinary incidents among elementary schools, we restrict our data set to the 778 traditional middle schools and high schools in the state that lie in 170 school districts (150 county districts and 20 city districts) and for which data are available for all three school years.³ We also use data from the 2010 U.S. Census to create contextual variables at the school district level that we also include in our models.⁴

These data allow us to construct representation variables at the school level for both teachers and administrators and to account for important control variables. Perhaps most importantly, these data allow us to control for student infraction rates that might affect the use of various types of discipline. Georgia public schools are required to report these infractions, and the GDOE classifies and archives these disciplinary incidents. Table 1 shows the average number of incidents that occur in the schools in our sample.

Table 1.

Summary of Disciplinary Incidents for School Years 2008-2009, 2009-2010, and 2010-2011.

Category	Incident	M	SD
Personal	Battery	2.65	5.74
	Homicide	0.00	0.00
	Kidnapping	0.00	0.00
	Sexual battery	0.11	0.57
	Sexual harassment	2.27	3.79
Property	Arson	0.06	0.33
	Burglary	0.11	0.57
	Larceny/theft	3.79	5.40
	Motor vehicle theft	0.03	0.24
	Robbery	0.08	0.46
Weapon	Knife	1.69	2.00
	Handgun	0.12	0.47
	Rifle	0.02	0.17
	Other	1.62	3.11
	Other firearm	0.02	0.18
Misdemeanor	Alcohol	1.37	2.64
	Computer/trespass	1.73	6.98
	Drugs (not alcohol)	6.49	10.30
	Fighting	42.97	42.57
	Threat/intimidation	15.36	27.07
	Sex offense	2.77	6.43
	Tobacco	10.69	16.81
	Trespassing	1.13	14.37
	Vandalism	3.22	4.76
	Serious bodily injury	0.03	0.26
Other minor	Disorderly conduct	41.39	147.92
	Other discipline offense	743.60	652.00
Total incidents		883.32	709.49

Source. Georgia Department of Education.

Dependent Variable

In Table 2, we summarize the different types of disciplinary actions used in response to the infractions shown in Table 1. We see that teachers and administrators are most likely to respond to those infractions by assigning either in-school or out-of-school suspensions. On average, we find that 59% of disciplinary actions are in-school suspensions; 31% of disciplinary actions are out-of-school suspensions. Given the strong inverse correlation between out-of-school suspension use and in-school suspension use (r = –.86), we chose to use only one of these as a dependent variable and focus on the use of out-of-school suspensions in our models. We choose out-of-school suspension rather than in-school suspensions as our dependent variable because we are interested in the use of more punitive disciplinary measures, and this measure is commonly used in studies of student discipline (e.g., Meier & Stewart, 1992; Mendez, Knoff, & Ferron, 2002). The dependent variable that we use in our models is the percentage of all disciplinary actions enacted within the schools that are out-of-school suspensions.⁵

Table 2.

Summary of Disciplinary Actions for School Years 2008-2009, 2009-2010, and 2010-2011.

Disciplinary action	M	SD
In-school suspension	535.73	459.16
Out-of-school suspension	279.16	303.56
Permanent expulsion	0.73	3.13
Non-permanent expulsion	3.36	8.11
Juvenile or court referral	2.03	7.27
Assigned to alternative school	10.25	18.33
Bus suspension—10 or fewer days	34.93	48.60
Bus suspension—more than 10 days	3.06	7.24
Returned to class review panel	0.01	0.14
Returned to class admin	0.02	0.49
Alternative setting review panel	0.03	0.54
Alternative setting admin	0.32	4.58
Other discipline action	19.62	51.00
Total disciplinary actions	890.24	721.75
Percent in-school suspensions	59.18
Percent out-of-school suspensions	30.55

Source. Georgia Department of Education.

Independent Variables

Community context

We measure racial context within the community using segregation, which we calculate using the index of dissimilarity (Duncan & Duncan, 1955; White, 1983) as follows:

S_{i} = \frac{1}{2} x \sum_{N} | X_{k} - Y_{k} |

where S_i is the level of segregation in school district i, X_k is the proportion of school district i’s non-Whites living in census tract k, Y_k is the proportion of school district i’s Whites living in census tract k, and N is the number of census tracts. S_i is interpreted as the proportion of a school district’s White population that would need to move to a different tract within the school district to reduce segregation to zero.⁶ In Georgia, the mean segregation index value for the school districts in our sample is 0.24 (SD = .13) and the maximum segregation value is 0.67.

Even though high and low segregated districts are distributed fairly evenly across the state, we still include a measure of urban density as an additional control to account for any association between segregation and areas of greater population density.⁷ Furthermore, we distinguish between city and county school districts using a dummy variable that equals 1 when a district is a city district and a 0 when the district is a county district.⁸ We also experimented with adding a fourth community-level variable to our models, the percentage of African Americans living within school districts, which would provide an additional contextual variable measuring the degree of racial threat present in communities. We, however, were not able to include this variable as it was highly correlated with the percentage of African American students within each school (r = .85), and it was more critical for our analyses to include this latter variable.

Representation variables

We also construct variables indicating the extent to which the school’s staff represents the ethnic make-up of their students. To measure the degree of ethnic representation, we created two separate representation indices: one for teachers (R_t) and one for administrators (R_a). We model these two types of representation separately because research suggests that teachers and administrators may play different roles in a wide array of student outcomes (e.g., Hallinger & Heck, 1996). The teacher representation index uses the following formula:

R_{t} = (1 - \sqrt{{(H_{s} - H_{t})}^{2} + {(W_{s} - W_{t})}^{2} + {(B_{s} - B_{t})}^{2} + {(A_{s} - A_{t})}^{2} + {(O_{s} - O_{t})}^{2}}) \times 100

where H_s = Proportion of Hispanic students in the school;

H_t = Proportion of Hispanic teachers;

W_s = Proportion of White students;

W_t = Proportion of White teachers;

B_s = Proportion of African American students;

B_t = Proportion of African American teachers;

A_s = Proportion of Asian students;

A_t = Proportion of Asian teachers;

O_s = Proportion of other students;

O_t = Proportion of other teachers.

The result is an overall measure of the ethnic match between teachers and students. Note that we inverted the measure that we include in our analyses so that lower numbers indicate less representation and higher numbers indicate more representation. In those cases where teachers largely do not represent the school’s students, the index may fall below 0. When constructing the administrative representation index, we relied on the same formula but use the proportion of administrators in the place of the proportion of teachers.⁹ Representative bureaucracy researchers have formulated a variety of methods for measuring representation, but this approach permits us to use a single measure to examine the extent to which an organization’s personnel are representative of all client ethnic groups (see Pitts, 2005).

Across these 3 years of data, the teacher representation variable ranges from −28.6 to a nearly perfectly represented school with a score of 98.8., with a mean score of 66.8. Of those schools in the top 5% of the teacher representation index distribution, 24% had a majority of African American students and 76% a majority of White students. On the other hand, the schools in the 5th percentile of the teacher representation index distribution were schools with high proportions of Hispanic students.

For the administrator representation index, the values ranged from −38.3 to 100, with 19 schools having a score of less than 0 during at least one of the school years under study. The mean score for the administrator representation index was 67.7.

Other school-level variables

A vast literature links resources and constraints within a school’s environment to student outcomes (Burtless, 1996; Fuller, Eggers-Pierola, Holloway, Liang, & Rambaud, 1996). Following this research, we included three series of control variables to tap various factors that might have an impact on disciplinary actions. First, we controlled for factors specific to the teachers and administrators for whom the representation indices were constructed. We include average salary and experience variables for both administrators and teachers, as well as the percentage of each group that holds graduate degrees. Second, we controlled for the attributes of the students within schools. For example, we included variables representing the percentage of students who qualified for federally funded reduced-price lunch, the percentage of students with disabilities, and the percentage of students who were language impaired. We also include a measure of the percentages of students in the school that are African American, which provides a measure of the degree of racial threat within the school (see Welch & Payne, 2010). To control for the severity of student misbehavior, we included variables for the number of disciplinary incidents per student in a series of categories, including those of a personal nature, those involving property, those involving weapons, misdemeanors, and other minor incidents, as well as a count of the total number of incidents that occur within a school. Third, we also included a series of control variables to account for other particularities about each school: the student–teacher ratio, the student–administrator ratio, whether it was a high school or middle school (dichotomous), and school size.

Method

Given that we consider data both at the school level and at the district level within our analysis, we estimate multi-level models using hierarchical linear modeling software (HLM) to predict the use of out-of-school suspensions. Public administration scholars have advocated for more advanced multilevel models (e.g., Heinrich & Lynn, 2000). Multilevel analysis offers several advantages, including accounting for the dependence between units at lower levels within the hierarchical data structure and better exploring causal heterogeneity—the likelihood that causal effects of lower level predictors are conditional on higher level predictors (Steenbergen & Jones, 2002). Hofmann and Gavin (1998) emphasized that in cross-level investigations, individuals within the same group are all exposed to similar group stimuli and are likely to be more similar to one another than individuals in other groups. Therefore, ordinary least squares (OLS) regression may produce standard errors that are biased downward, which is not the case with HLM. We also use the first-order auto-regressive function of HLM to account for the likelihood that the correlation of outcomes decreases with increasing distance between school years. This function automatically generates time point indicators within the model, avoiding the need to include a time variable in the model.¹⁰

Our models have three different levels of data. We treat our representation and student and school characteristic variables as time-varying variables across the examination years at Level 1 of our models. We treat one school-level control variable, middle or high school, as a time-invariant dummy variable at Level 2. Finally, we treat the contextual variables common to schools within the same district (e.g., segregation) as time-invariant variables at Level 3. Furthermore, we center our interval-level variables at the grand mean.¹¹

To make certain that we see variability at both the individual and group levels within our data, we estimated an unconditional model without predictors that allows us to dissect that variance among schools and districts. We do not report the results for this model, but we used them to calculate the proportion of the variance in our dependent variable that occurs between groups. We calculated an intra-class correlation at Level 3 of .37 and at Level 2 of .42; thus, the proportion of variance among school districts is 37% and among schools within school districts is 42%, telling us that grouping the schools by school district is useful and there is variance to explain at the school level. We next discuss the results for our models, focusing on the possible effects of these school-level factors and contextual environments in which schools are embedded on the use of more punitive disciplinary measures.

Results

Do Schools With Higher Percentages of African American Students Assign More Punitive Punishments?

As we have discussed, research has shown that African American students are often disproportionately assigned more punitive punishments. Previous work also suggests that higher percentages of African American students are sometimes viewed as a racial threat, leading to an increase in prejudicial attitudes (Welch & Payne, 2010). Thus, we expect to see a higher rate of out-of-school suspensions assigned more often to students in schools where more of the students are African American and particularly so in areas with higher levels of segregation. We examine this relationship empirically by estimating two models of out-of-school suspensions, one including the percentage of African American students and a second model that includes this variable as well as an interaction between the percentage of African American students and segregation. Both models also include the control variables that we have identified. We present the results of these calculations in the Table 3.

Table 3.

Coefficients for Hierarchical Multivariate Linear Model (HMLM2) Regression of Out-of-School Suspensions on African American Student Share.

	Out-of-school suspensions
	(1)	(2)
Cross-level interaction
% African American student × Segregation		0.66*** (0.12)
District-level
Segregation	14.62* (7.29)	14.00 (7.18)
Population density	0.00 (0.00)	−0.00 (0.00)
City district	1.52 (2.90)	−0.03 (2.87)
School-level
Student characteristics
% African American student	0.27*** (0.03)	0.21*** (0.03)
Economically disadvantaged students	2.36 (3.33)	−1.53 (3.38)
Students with disabilities	27.91* (12.05)	26.25* (11.95)
English language learners	−1.54 (12.84)	24.43 (13.51)
School characteristics
Number of students	−0.00 (0.00)	−0.00 (0.00)
Middle school	−6.19*** (0.99)	−6.25*** (0.97)
Incidents
Personal	−57.88* (26.42)	−57.39* (26.36)
Property	−51.54 (45.16)	−45.62 (45.06)
Weapons	16.53 (54.26)	7.60 (54.19)
Misdemeanor	7.21 (3.97)	5.28 (3.96)
Other	−5.63*** (.91)	−4.70*** (0.92)
Total incidents	0.00 (0.00)	0.00 (0.00)
Intercept	30.48*** (2.37)	31.94*** (2.35)
School-level / district-level units	778 / 170	778 / 170
Deviance	17,573.63	17,543.72
Variance (within schools, sigma squared)	70.21	69.70
Variance (between schools, tau pi)	92.34	87.73

Note. The standard errors are in parentheses.

p < .05. **p < .01. ***p < .001.

In Model 1, we see that schools with higher percentages of African American students have higher rates of out-of-school suspensions. Note that we are also controlling for the disciplinary incidents that have occurred among students within the schools. We also see a significant effect among our variables tapping the conditions within our communities. Those communities that have greater residential segregation are more likely to see an increase in the rate of out-of-school suspensions, perhaps because of more negative racial climates in these communities. Thus, it appears that in these types of contextual environments teachers and/or administrators are more likely to assign more punitive disciplinary measures.

When we turn to look at Model 2, we see that the interaction between the percentage of African American students in a school and the degree of segregation within the school district is also positive and significant. The school staff in schools with larger percentages of African American students and in more segregated districts appear to assign more out-of-school suspensions. The higher percentage of African Americans within schools may indicate a racial threat to the typical majority group, Whites, or the social order of the school generally and this may play a more significant role in these more segregated communities.

It is important to note that our dependent variable reflects disciplinary actions assigned to all students within the school, only some of whom are African Americans. We believe that our conclusions are likely, however, because of the strong association we see in our models between the percentage of African American students and the use out-of-school suspensions and because of past research that demonstrates a strong link between African American students and the use of more punitive disciplinary actions (McFadden, Marsh, Price, & Hwang, 1992).

We turn to consider the influence of the representation of teachers and administrators on these outcomes and whether active representation appears more likely to occur within more or less segregated communities.

Does Teacher or Administrator Representation Decrease the Use of Punitive Disciplinary Actions?

We next estimate four models in which we examine whether representation decreases the rate of more punitive sanctions within the schools. In the first model, we examine the effects of teacher and administrator representation, and in the second and third models we separately examine the extent to which the effect of each representation variable is conditional on community segregation. Finally, in the fourth model, we include both representation variables as well as their interactions with the segregation variable. We control for the same variables as in the previous models.

We present our results in Table 4. In Model 1, we see that there appears to be no significant effect of teacher or administrator representation on the percentage of out-of-school suspensions used within schools. Note that after controlling for representation, we continue to see effects for the percentage of African American students and our contextual variables that are similar to those that we saw in our previous models.

Table 4.

Coefficients for Hierarchical Multivariate Linear Model (HMLM2) Regression of Out-of-School Suspensions on Teacher and Administrator Representation and Student and School Characteristics.

	Out-of-school suspensions
	(1)	(2)	(3)	(4)
Cross-level interactions
Teacher representation × Segregation		−0.37** (0.13)		−0.36* (0.15)
Administrator representation × Segregation			−0.10 (0.08)	−0.00 (0.09)
District-level
Segregation	14.61* (7.33)	14.54* (7.31)	14.66* (7.34)	14.55* (7.31)
Population density	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
City district	1.80 (2.93)	1.93 (2.92)	1.83 (2.94)	1.93 (2.92)
School-level
Teachers
Representation	0.04 (0.03)	0.08** (0.03)	0.04 (0.03)	0.08* (0.03)
Salary	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Experience	0.18 (0.21)	0.20 (0.21)	0.19 (0.21)	0.20 (0.21)
% with master’s	−0.12** (0.04)	−0.12** (0.04)	−0.12** (0.04)	−0.12** (0.04)
Administrator’s characteristics
Representation	−0.02 (0.01)	−0.02 (0.01)	−0.01 (0.02)	−0.02 (0.02)
Salary	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Experience	0.03 (0.07)	0.03 (0.07)	0.03 (0.07)	0.03 (0.07)
% with master’s	0.00 (0.03)	0.01 (0.03)	0.01 (0.03)	0.01 (0.03)
Student shares
% African American	0.27*** (0.03)	0.29*** (0.03)	0.27*** (0.03)	0.29*** (0.03)
Economically disadvantaged	3.45 (3.46)	3.50 (3.45)	3.54 (3.46)	3.50 (3.45)
With disabilities	30.83* (12.25)	30.36* (12.22)	30.75* (12.25)	30.36* (12.22)
English language Learners	0.20 (14.02)	−5.96 (14.15)	−2.16 (14.14)	−6.01 (14.19)
Schools
No. of students	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)
Student–teacher ratio	0.46** (0.14)	0.46** (0.14)	0.45** (0.14)	0.46** (0.14)
Student–administrator ratio	−0.01 (0.00)	−0.01 (0.00)	−0.01 (0.00)	−0.01 (0.00)
Middle school	−5.77*** (1.04)	−5.66*** (1.04)	−5.79*** (1.04)	−5.66*** (1.04)
Incidents
Personal	−58.32* (26.34)	−61.82* (26.36)	−58.59* (26.33)	−61.80* (26.36)
Property	−54.23 (26.34)	−49.94 (45.04)	−53.19 (45.03)	−49.43 (45.04)
Weapons	29.56 (54.29)	38.44 (54.37)	30.12 (54.29)	38.39 (54.37)
Misdemeanor	7.70 (3.98)	8.18* (3.97)	7.72 (3.97)	8.18* (3.97)
Other	−5.73*** (0.91)	−5.77*** (0.90)	−5.72*** (0.91)	−5.77*** (0.90)
Total incidents	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Intercept	26.10*** (2.41)	26.16*** (2.40)	26.14*** (2.41)	26.16*** (2.40)
School-level units	778	778	778	778
District-level units	170	170	170	170
Deviance	17552.53	17544.77	17550.96	17544.77
Variance (within schools, sigma squared)	68.80	68.62	68.40	68.60
Variance (between schools, tau pi)	91.64	90.29	91.75	90.30

Note. The standard errors are in parentheses.

p < .05. **p < .01. ***p < .001.

In Model 2, we extend our model to include an interaction between teacher representation and the degree of segregation in the school district. We can see that its effect is negative and significant, whereas the coefficients for teacher representation and segregation are positive and significant.

Thus, the results indicate a decrease in the use of out-of-school suspensions among schools in districts with higher levels of segregation. One finding that is unexpected, however, is that at low levels of segregation, representation appears to increase the use of out-of-school suspensions. One possible explanation is that race is less salient to teachers in low segregation districts, and as a result they may be influenced to a greater extent by schoolwide norms that lead to pursuit of traditional “get tough” disciplinary practices. Increasing levels of teacher representation increase the out-of-school suspension rate until the segregation index equals 0.45 (e.g., 45% of whites or minorities would need to transfer census tracts to eliminate segregation within the district). Above a segregation value of 0.45, however, increasing teacher representation decreases the out-of-school suspension rate. For instance, within a school district with a segregation index of 0.55, the use of out-of-school suspension decreases by about 4 percentage points as the value of teacher representation changes from its lowest value to its highest value.¹² In the most highly segregated districts, the use of out-of-school suspensions decreases by about 9 percentage points as representation changes from its lowest to highest level.

In these highly segregated districts, race should be more salient to bureaucrats, and thus active representation should be more likely to occur, muting the negative effects of the more prejudicial attitudes that may be present within these environments. Thus, teachers within these more segregated communities are more likely to be more cognizant of their own and their students’ racial characteristics or may be more likely to take on the role of a minority representative, regardless of their own minority status (see Sowa & Selden, 2003). As a result, they are more likely to act to moderate the high levels of out-of-school suspensions present within these schools.

In Model 3, we see that this transition from passive representation to active representation does not appear more likely to occur among administrators within more segregated environments. Overall, we see little influence of administrator representation on the types of disciplinary tools used within the schools. Our final model for out-of-schools suspensions (Model 4) shows that once we also account for the interaction between administrator representation and segregation, the variable for administrator representation remains insignificant, and we continue to see that active representation appears more likely among teachers in more segregated environments. Among administrators, professional norms may inhibit this transition from passive representation to active representation. In sum, it appears that the transition from passive to active representation is more likely to occur in districts where there is a higher degree of segregation, leading to a decreased use of more punitive policies such as out-of-school suspensions.

The Racial Context Matters

This research has explored the role of racial context in the implementation of a racialized school policy—student discipline. Our findings demonstrate a pervasive role for racial context; the environments in which schools are embedded appear to influence the types of disciplinary actions assigned to students and the extent to which we see active representation within the schools. Overall, the discipline policies in schools appear conditional on the racial environments in which they are embedded.

We find that when schools are embedded in environments with greater segregation, students are more often assigned out-of-school suspensions. Thus, these environments support harsher disciplinary outcomes for students, potentially because of the more negative racial climates present in these areas that may be the result of the lower levels of contact among racial groups within the community. Thus, our results are consistent with the notion that the social constructions of students may be more negative in these environments and support other research suggesting that certain target populations are less likely to experience beneficial policies when they are negatively constructed within society (e.g., Nicholson-Crotty & Nicholson-Crotty, 2004; Schneider & Ingram, 1993, 1997). We also find, mirroring the results of past research, that teachers and staff in schools with more African American students are increasingly likely to assign to their students punitive disciplinary measures such as out-of-school suspensions (e.g., Skiba et al., 2002; Welch & Payne, 2010). This result is consistent with previous research that has shown that African American students are generally overassigned more punitive disciplinary actions. Larger percentages of African American students within the student body may create an atmosphere of racial threat within the school, leading to the increased use of disciplinary tools such as out-of-school suspensions.¹³ These levels of racial threat appear, however, most evident in schools embedded in more segregated school districts, where prejudicial attitudes may be more likely because of more limited intergroup contact. Thus, our results support the idea that in more integrated environments, the effects of racial threat are less apparent. We emphasize that we see all of these results after controlling for the disciplinary incidents that have taken place in the schools and the socioeconomic makeup of the schools’ student populations.

When we turn to examine the influence of teacher representation of students on disciplinary outcomes, we see significant positive effects for this representation variable only in more segregated environments. Thus, it appears that the contextual environment moderates the transition from passive representation to active representation, and we suggest that this occurs because of the greater salience of race in more segregated areas. In these more segregated environments, teachers are more likely to act to promote more positive outcomes for students of their own race or students of another race for which teachers hold a minority representative role. It is likely that in these environments educators see the benefits of active representation as outweighing the potential costs. In these areas, the rates of disciplinary suspensions appear high and racial justice may be suspect, suggesting greater benefits for active representation. Teachers may also be able to produce benefits for represented groups without relying on potentially costly strategies of active representation such as reverse discrimination. As they engage in active representation, teachers may be able to rely on shared values or their more empathetic understanding of the behavior of minority students, allowing them to view students’ behaviors within the classroom less harshly and potentially limiting their use of disciplinary referrals. Our empirical results do not indicate the presence of a zero sum game or a prevalent use of reverse discrimination when active representation occurs.¹⁴ If this were to happen, we would fail to see a link between our measure of representation and the use of fewer punitive disciplinary actions. Thus, our results suggest a more positive outcome than that feared by researchers who may question the democratic implications of representative bureaucracy and are wary of the possible negative, redistributional impacts resulting from active representation (Lim, 2006).¹⁵ Instead, we find that disciplinary actions appear less punitive for students generally as representation increases among students in segregated districts.

We see no significant effects of administrator representation—either on its own or as a base variable with our interaction terms. These results are similar to those that have been reported in some previous research (e.g., Authors, 2010). Our results may be explained in part by other external organizational and environmental forces that may also confront the administrator. In his classic work, Herbert (1974) argued that administrators are often subject to a number of forces, such as organizational sanctions and rewards, as well as pressure from colleagues to conform to informal or professional norms, which may influence the extent to which administrators take on a minority representative role. While Herbert was writing about minority administrators, these conclusions may be applied to administrators generally and may help explain the results of our research (see also Henderson, 1988; Meier & Stewart, 1992).

It is important to note that in this research we are assuming that teachers and administrators are influenced by the attributes of the local communities in which they teach. Research suggests that schools often reflect the socioeconomic environments in which they are embedded (Kozol, 1991). Teachers and administrators who live close to the schools in which they work should also be even further immersed in these environments, and even if they live at a distance, they should learn about many of these cultural factors present in their districts through their interactions with parents, other teachers and administrators, and also from local media. Our empirical study also relies on data drawn from the southern United States, where race may be more salient than elsewhere. The possibility exists that the effects that we report are heightened by this saliency and are greater than they would be if a similar study were conducted in another region or country.

We also note that in this research we lack the ethnographic data that allow us to fully document the motivations underlying the behavior of teachers and administrators, which would fully allow us to verify the presence of active representation. Our models, which are based on aggregate data, only allow us to suggest that educators are behaving in a way that is consistent with active representation. We encourage researchers to work whenever possible to identify these missing causal linkages in future research on representative bureaucracy through the use of ethnography and individual level data.

This research should help educators and policy makers understand how the social environment may shape disciplinary decision making. Teachers may benefit from training, especially in districts with greater degrees of segregation, about the connection between race and more severe disciplinary actions. States such as Georgia are paying increasing attention to discipline policy, and schools across the country are implementing programs, such as Positive Behavioral Intervention and Supports (PBIS), seeking to help create more positive social environments within schools and reduce disruptive behavior problems. Our findings suggest that it may be particularly important to target these programs to more segregated school districts and possibly other areas in which schools rely on more punitive disciplinary actions.

We believe that this research makes an important contribution by showing the role of social context in determining the types of disciplinary policy tools assigned to students in middle and high schools. Overall, we see a greater use of out-of-school suspensions in districts with greater segregation. We also see in the schools in these districts, however, that the racial match between school staff and students appears to mute the use of these more punitive tools. Representation appears to play a greater role in more segregated districts. Thus, the social context appears to play a key role in moderating this transition from passive representation to active representation. The challenge for future research is to test these ideas in more varied contexts and whenever possible with individual-level data.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Christine H. Roch is an Associate Professor of public management and policy at Georgia State University. Her research interests include education policy, representation, and state and local governance.

Jason Edwards is a recent graduate of the doctoral program in Public Policy from Georgia State University. His research interests include education policy, social contexts, local politics and the policy process.

References

Adams

A. T.

(1992). Public high schools: The use of rehabilitative and punitive forms of discipline: A final report. Washington, DC: Office of Educational Research and Improvement.

Allport

(1954). The nature of prejudice. Reading, MA: Addison Wesley.

Blalock

H. M.

(1967). Toward a theory of minority-group relations. New York, NY: John Wiley.

Blumer

(1958). Race prejudice as a sense of group position. The Pacific Sociological Review, 1, 3-7.

Bradbury

M. D.

Kellough

J. E.

(2008). Representative bureaucracy: Exploring the potential for active representation in local government. Journal of Public Administration Research and Theory, 18, 697-714.

Brown

T. M.

(2007). Lost and turned out: Academic, social, and emotional experiences of students excluded from school. Urban Education, 42, 432-455.

Burtless

(1996). Does money matter? The effect of school resources on student achievement and adult success. Washington, DC: Brookings Institution.

Campbell

(1965). Ethnocentric and other altruistic motives. In Levine

(Ed.), Nebraska symposium on motivation (Vol. 13, pp. 283-311). Lincoln: University of Nebraska Press.

Carr

Kutty

(2008). Segregation: The rising costs for America. New York, NY: Routledge.

10.

Christ

Schmid

Lolliot

Swart

Stolle

Tausch

Hewstone

(2014). Contextual effect of positive intergroup contact on outgroup prejudice. Proceedings of the National Academy of Science, 111, 3996-4000. doi:10.1073/pnas.1320901111

11.

Dixon

Rosenbaum

(2004). Nice to know you? Testing contact, cultural, and group threat theories of anti-black and anti-Hispanic stereotypes. Social Science Quarterly, 85, 257-280.

12.

Duncan

(1955). A methodological analysis of segregation indexes. American Sociological Review, 20, 210-217.

13.

Eitle

D. S.

D’Alessio

Stolzenberg

(2002). Racial threat and social control: A test of the political, economic, and threat of black crime hypotheses. Social Forces, 81, 557-576.

14.

Emerson

Kimbro

Yancey

(2002). Contact theory extended: The effects of prior racial contact on current social ties. Social Science Quarterly, 83, 745-761.

15.

Fuller

Eggers-Pierola

Holloway

S. D.

Liang

Rambaud

M. F.

(1996). Rich culture, poor markets: Why do Latino parents forego preschooling? Teachers College Record, 97, 400-418.

16.

Georgia Appleseed Center for Law & Justice (2011). Georgia Appleseed Center for Law & Justice. Retrieved from www.gaappleseed.org.

17.

Georgia Department of Education (2013). Addressing Climate, Safety, and Discipline in Georgia Schools: School-wide Behavioral Interventions and Supports. Atlanta, GA: Georgia Department of Education,. Retrieved from www.gadoe.org/Curriculum-Instruction-and-Assessment/Special-Education-Services/Documents/PBIS/PBIS%20Final%20white%20paper_%20Sept%204.pdf

18.

Giles

Evans

(1985). External threat, perceived threat, and group identity. Social Science Quarterly, 66, 50-66.

19.

Goff

P. A.

Steele

C. M.

Davies

P. G.

(2008). The space between us: Stereotype threat and distance in interracial contexts. Journal of Personality and Social Psychology, 94, 91-107.

20.

Goldsmith

(2004). Schools’ racial mix, students’ optimism, and the black-white and Latino-white achievement gaps. Sociology of Education, 77, 121-147.

21.

Grissom

Nicholson-Crotty

(2009). Race, region, and representative bureaucracy. Public Administration Review, 69, 911-919.

22.

Hallinger

Heck

(1996). Reassessing the principal’s role in school effectiveness: A review of empirical research, 1980-1995. The Journal of Leadership for Effective & Equitable Organizations, 32, 5-44.

23.

Heinrich

Lynn

(2000). Governance and performance: The influence of program structure and management on Job Training Partnership Act (JTPA) program outcomes. In Heinrich

Lynn

(Eds.), Governance and performance: New perspectives (pp. 68-101). Washington, DC: Georgetown University Press.

24.

Henderson

(1988). Urban administrators: The politics of role elasticity. In Karnig

McClain

(Eds.), Urban minority administrators: Politics, policy, and style (pp. 15-31). New York, NY: Greenwood.

25.

Herbert

(1974). The minority administrator: Problems, prospects, and challenges. Public Administration Review, 34, 556-563.

26.

Hindera

J. J.

(2004). Representative bureaucracy: Further evidence of active representation in the EEOC district offices. Journal of Public Administration Research and Theory, 3, 415-429.

27.

Hirsh

Kornrich

(2008). The context of discrimination: Workplace conditions, institutional environments, and sex and race discrimination charges. American Journal of Sociology, 113, 1394-1432.

28.

Hofmann

Gavin

(1998). Centering decisions in hierarchical linear models: Implications for research in organizations. Journal of Management, 24, 623-641.

29.

Ingram

Schneider

DeLeon

(2007). Social construction and policy design. In Sabatier

(Ed.), Theories of the policy process (pp. 93-126). Bellevue, TN: Westview Press.

30.

Kaufman

Chen

Choy

Peter

Ruddy

Miller

. . . Rand

(2001). Indicators of school crime and safety: 2001 (NCES 2002-113/NCJ-190075). Washington, DC: U.S. Departments of Education and Justice.

31.

Keiser

L. R.

(1999). State bureaucratic discretion and the administration of social welfare programs: The case of social security disability. Journal of Public Administration Research and Theory, 9, 87-106.

32.

Keiser

L. R.

Wilkins

V. M.

Meier

K. J.

Holland

C. A.

(2002). Lipstick and logarithms: Gender, institutional context, and representative bureaucracy. American Political Science Review, 96, 553-564.

33.

Kellough

(1990). Integration in the public workplace: Determinants of minority and female employment in federal agencies. Public Administration Review, 50, 557-566.

34.

Key

(1949). Southern politics. New York, NY: Random House.

35.

Kinder

Mendelberg

(1995). Cracks in American apartheid: The political impact of prejudice among desegregated whites. The Journal of Politics, 57, 402-424.

36.

King

Wheelock

(2007). Group threat and social control: Race, perceptions of minorities and the desire to punish. Social Forces, 85(3), 1255-1280.

37.

Kozol

(1991). Savage inequalities: Children in America’s schools. New York, NY: HarperCollins.

38.

Kreft

de Leeuw

(1998). Introduction to multilevel modeling. London, England:Sage.

39.

Krislov

(2012). Representative bureaucracy. New Orleans, LA: Quid Pro Books.

40.

Kupchik

(2009). Things are tough all over: Race, ethnicity, class and school discipline. Punishment & Society, 11, 291-317.

41.

Lawrence

P. A.

Olvey

S. K.

(1994). Discipline: A skill not a punishment. American School Board Journal, 181, 31-32.

42.

Lim

(2006). Representative bureaucracy: Rethinking substantive effects and active representation. Public Administration Review, 66, 193-204.

43.

McFadden

Marsh

II Price

Hwang

(1992). A study of race and gender bias in the punishment of handicapped school children. The Urban Review, 24, 239-251.

44.

Meier

K. J.

(1975). Representative bureaucracy: An empirical assessment. American Political Science Review, 69, 526-542.

45.

Meier

K. J.

(1993a). Latinos and representative bureaucracy: Testing the Thompson and Henderson hypotheses. Journal of Public Administration Research and Theory, 3, 393-414.

46.

Meier

K. J.

(1993b). Representative bureaucracy: A theoretical and empirical exposition. In Perry

J. L.

(Ed.), Research in public administration (pp. 1-35). Greenwich, CT: JAI Press.

47.

Meier

K. J.

Bohte

(2001). Structure and discretion: Missing links in representative bureaucracy. Journal of Public Administration Research and Theory, 11, 455-470.

48.

Meier

K. J.

Nicholson-Crotty

(2006). Gender, representative bureaucracy, and law enforcement: The case of sexual assault. Public Administration Review, 66(6), 850-860.

49.

Meier

K. J.

Eller

Wrinkle

Polinard

(2001). Zen and the art of policy analysis: A response to Nielsen and Wolf. The Journal of Politics, 63, 616-629.

50.

Meier

K. J.

O’Toole

Nicholson-Crotty

(2004). Multilevel governance and organizational performance: Investigating the political-bureaucratic labyrinth. Journal of Policy Analysis and Management, 23, 31-47.

51.

Meier

K. J.

Stewart

J. S.

(1991). The politics of Hispanic education. Albany: State University of New York Press.

52.

Meier

K. J.

Stewart

J. S.

(1992). The impact of representative bureaucracies: Educational systems and public policies. American Review of Public Administration, 22, 157-171.

53.

Meier

K. J.

Wrinkle

R. D.

Polinard

J. L.

(1999). Representative bureaucracy and distributional equity: Addressing the hard question. The Journal of Politics, 61, 1025-1039.

54.

Mendez

Knoff

Ferron

(2002). School demographic variables and out-of-school suspension rates: A quantitative and qualitative analysis of a large, ethnically diverse school district. Psychology in the Schools, 39, 259-277.

55.

Mosher

(1968). Democracy and the public service. New York, NY: Oxford University Press.

56.

Mosher

(1982). Democracy and the public service (2nd ed.). New York, NY: Oxford University Press.

57.

Nicholson-Crotty

(2004) Interest group influence on managerial priorities in public organizations. Journal of Public Administration Research and Theory, 14(4), 571-583.

58.

Nicholson-Crotty

Grissom

Nicholson-Crotty

(2011). Bureaucratic representation, distributional equity, and democratic values in the administration of public programs. The Journal of Politics, 73, 582-596.

59.

No Child Left Behind Act of 2001, 20 U.S.C. §6301 (2002).

60.

Oliver

(2010). The paradoxes of integration: Race, neighborhood, and civic life in multiethnic America. Chicago, IL: University of Chicago Press.

61.

Oliver

Wong

(2003). Intergroup prejudice in multiethnic settings. American Journal of Political Science, 47, 567-582.

62.

Park

(2014). The politics of redistribution in local governments: The effect of gender representation on welfare spending in California counties. Journal of Public Policy, 34, 269-301.

63.

Payne

Welch

(2010). Modeling the effects of racial treat on punitive and restorative school discipline practices. American Society of Criminology, 48, 1019-1062.

64.

Pettigrew

(1997). Generalized intergroup contact effects on prejudice. Personality and Social Psychology Bulletin, 23, 173-185.

65.

Pettigrew

(1998). Intergroup contact theory. Annual Review of Psychology, 49, 65-85.

66.

Pettigrew

Tropp

(2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90, 751-783.

67.

Pitts

D. W.

(2005). Diversity, representation, and performance: Evidence about ethnicity in public organizations. Journal of Public Administration Research and Theory, 15, 323-339.

68.

Pitts

D. W.

(2007). Representative bureaucracy, ethnicity, and public schools: Examining the link between representation and performance. Administration Society, 39, 497-526.

69.

Preuhs

(2006). The conditional effects of minority descriptive representation: Black legislators and policy influence in the American states. The Journal of Politics, 68, 585-599.

70.

Quillian

(1995). Prejudice as a response to perceived group threat: Population composition and anti-immigrant and racial prejudice in Europe. American Sociological Review, 60, 586-611.

71.

Rigby

(2007). Same policy area, different politics: How characteristics of policy tools alter the determinants of early childhood education policy. The Policy Studies Journal, 35(4), 653-669.

72.

Roch

Rushton

(2008). Racial context and voting over taxes: Evidence from a referendum in Alabama. Public Finance Review, 36(5), 614-634.

73.

Rocha

Espino

(2009). Racial threat, residential segregation, and the policy attitudes of anglos. Political Research Quarterly, 62, 415-426.

74.

Rocha

Hawes

(2009). Racial diversity, representative bureaucracy, and equity in multiracial school districts. Social Science Quarterly, 90, 326-344.

75.

Roch

Pitts

Navarro

(2010) Representative bureaucracy and policy tools: Ethnicity, student discipline, and representation in public schools, Administration & Society, 42(1), 38-65.

76.

Rothstein

(2004). A wider lens on the black-white achievement gap. Phi Delta Kappan, 86, 104-110.

77.

Schneider

(2012). Punishment policy in the American states from 1890 to 2008: Convergence, divergence, synchronous change, and feed-forward effects. Policy Studies Journal, 40(2), 193-210.

78.

Schneider

Ingram

(1993). Social construction of target populations: Implications for politics and policy. American Political Science Review, 87, 334-347.

79.

Schneider

Ingram

(1997). Policy design for democracy. Lawrence: University Press of Kansas.

80.

Schlueter

Scheepers

(2010). The relationship between outgroup size and anti-outgroup attitudes: A theoretical synthesis and empirical test of group threat- and intergroup contact theory. Social Science Research, 39, 285-295.

81.

Selden

S. C.

(1997a). The promise of representative bureaucracy: Diversity and responsiveness in a government agency. Armonk, NY: M.E. Sharpe.

82.

Selden

S. C.

(1997b). Representative bureaucracy: Examining the linkage between passive and active representation in the Farmers Home Administration. American Review of Public Administration, 27, 22-42.

83.

Selden

S. C.

(2006). A solution in search of a problem? Discrimination, affirmative action, and the new public service. Public Administration Review, 66, 911-923.

84.

Selden

S. C.

Brudney

J. L.

Kellough

J. E.

(1998). Bureaucracy as a representative institution: Toward a reconciliation of bureaucratic government and democratic theory. American Journal of Political Science, 42, 717-744.

85.

Shah

(2012, November). Suspended in school but still learning. Education Week, 32(11), 1-18.

86.

Short

P. M.

Short

R. J.

Blanton

(1994). Rethinking student discipline: Alternatives that work (ERIC Document Reproduction Service No. 378-658).Thousand Oaks, CA: Corwin Press.

87.

Skiba

R. J.

Horner

Chung

C.-G.

Karega

May

Tobin

(2011). Race is not neutral: A national investigation of African American and Latino disproportionality in school discipline. School Psychology Review, 40, 85-107.

88.

Skiba

R. J.

Michael

R. S.

Peterson

R. L.

(2002). The color of discipline: Sources of racial and gender disproportionality in school punishment. The Urban Review, 34, 317-342.

89.

Skiba

R. J.

Peterson

(1999). The dark side of zero-tolerance: Can punishment lead to safe schools? Phi Delta Kappan, 80, 372-382.

90.

Sowa

J. E.

Selden

S. C.

(2003). Administrative discretion and active representation: An expansion of the theory of representative bureaucracy. Public Administration Review, 63, 700-710.

91.

Steenbergen

Jones

(2002). Modeling multilevel data structures. American Journal of Political Science, 46, 218-237.

92.

Taylor

(1998). How White attitudes vary with the racial composition of local populations: Numbers count. American Sociological Review, 63, 512-535.

93.

U.S. Department of Education. (2014). The civil rights data collection. Washington, DC: Office for Civil Rights.

94.

Vavrus

Cole

(2002). “I didn’t do nothing”: The discursive construction of school suspension. The Urban Review, 34, 87-111.

95.

Verdugo

R. R.

(2000). Zero tolerance policies: A critical review. Washington, DC: National Education Association.

96.

Verdugo

R. R.

Glenn

B. C.

(2002, April 1-5). Race-ethnicity, class, and zero tolerance policies: A policy discussion. Paper presented at the Annual Meeting of the American Educational Research Association, New Orleans, LA.

97.

Welch

Payne

(2010). Racial threat and punitive school discipline. Social Problems, 57, 25-48.

98.

Welch

Sigelman

Bledsoe

Combs

(2001). Race and place: Race relations in an American city. New York, NY: Cambridge University Press.

99.

White

(1983). The measurement of spatial segregation. American Journal of Sociology, 88, 1008-1018.

100.

Zubrinsky

Bobo

(1996). Prismatic metropolis: Race and residential segregation in the city of the angels. Social Science Review, 25, 335-374.