Bureaucratic Discretion,Client Demographics,and Representative Bureaucracy

Client Demand

In the public education context, client demand refers to the idea that a bureaucracy’s minority students and their parents will lobby for minority teachers to have more discretion since this discretion will allow minority teachers to better serve minority students. Our expectation is that as the percentage of a school’s students who are minorities increases, lobbying for minority teacher discretion will increase. In other words, minority parent demand will be directly proportional to minority students’ (or parents’) prevalence in the school’s client population. There are a couple of assumptions built into this causal mechanism, each of which requires justification. The first is that a channel for minority parent demands exists. The justification for this assumption is mainly theoretical, and was noted above: Public schools are typically classified as open systems by organization theorists and education sociologists (Ballantine & Hammack, 2012; Hanson, 1996). The primary implication of public schools’ status as open systems organizations is that they are subject to forces in their external environments. Public schools are political institutions, and so these forces include the demands pressed on them by interested political actors, including parents (Chubb & Moe, 1988; Cutler, 2000; Meier et al., 2000). Since public schools are democratic political institutions, they are expected to attend to these demands, which are normatively legitimate.

While the preceding justification is theoretical, Chubb and Moe (1990) have shown empirically that public schools are subject to external demands from parents and other interested groups to a much greater degree than are private schools. Moreover, a great deal of research in the education literature indicates that parents are intensely involved in the governance of their children’s schools (e.g., Booth & Dunn, 1996; Cutler, 2000; Epstein & Salinas, 2004), sometimes to such a degree that teachers and principals perceive their involvement to be intrusive (Addi-Raccah & Arviv-Elyashiv, 2008). And so, it seems reasonable to conclude that parents are able to, and in fact do, press demands on their children’s schools. The key question is whether these include demands made by minority parents that minority teachers be given more discretion. ² Our assumption is that they do.

We are not aware of any research that would allow us to directly substantiate this assumption; that is, research showing that minority parents actually do make demands for more minority teacher discretion. Nevertheless, research does suggest that minority parents are actively engaged in their children’s schooling. More basically, this research indicates that minority parents are keenly concerned about what goes on at their children’s schools. Chavkin and Williams (1993) surveyed 1,188 African American and Hispanic parents about their level of interest in 14 areas of school decision making. At least 75% of surveyed parents were highly interested in the following areas: amount of homework assigned, classroom discipline, student evaluation, setting school behavior rules, and setting school rules for grading and passing children. (We note these specific decision-making areas because they are captured in our measure of teacher discretion). Moreover, 70% of surveyed parents were highly interested in textbook selection. Similarly, Ritter et al. (1993) showed that African American and Hispanic parents are highly involved in the areas of course selection and scheduling, and are well-attuned to the amount of homework assigned to their children. More recently, a number of studies have indicated that Hispanic parents are deeply concerned with the academic progress of their children, though it is important to note that their involvement with their children’s’ schools tends to be informal (e.g., helping with homework and talking about the school day’s events vs. attending PTA meetings; Behnke et al., 2004; De Gaetano, 2007; LeFevre & Shaw, 2012). While findings like these do not directly substantiate the assumption that minority parents use open channels to make demands about minority teacher discretion, they do suggest that minority parents are actively engaged in their children’s educations and are well-attuned to the happenings at their schools. It is plausible to think that engaged, attentive parents would be inclined to intervene on their children’s behalf by making demands about teacher discretion if they believed that doing so would benefit their children.

Studies in the representative bureaucracy literature lend some additional indirect support to our assumption that minority parents will make demands for more minority teacher discretion. These studies all suggest that a bureaucracy’s minority clients value minority bureaucratic representation. Thielemann and Stewart (1996), for instance, show that large majorities of Black and Hispanic persons living with AIDS (PLWAs) care whether the staffs of HIV service delivery agencies include members of their ethnic group. Moreover, they show that significantly larger percentages of Black and Hispanic PLWAs (relative to White PLWAs) are concerned about whether an agency’s director is from their ethnic group. Recent qualitative research by Watkins-Hayes (2011) provides additional support for the proposition that minority clients value minority bureaucratic representation. Watkins-Hayes (2011), who conducted interviews with 20 minority (Black and Latino) welfare clients in Massachusetts to explore the role of race in the client–caseworker relationship, notes that a

small but important group of clients hoped that a connection with caseworkers from their racial groups would translate into a sympathetic ear (and hand) when it came to service delivery and, in an often-used phrase, “an understanding of where I’m coming from.” (p. i242)

Bradbury and Kellough (2008) surveyed African American and White citizens in the city of Athens, GA, about the importance of African American representation in local administrative organizations. African American citizens, to a much greater degree than White citizens, believed that African Americans should be given access to positions in local administration and be encouraged to use those positions to advocate for African American citizens. While these findings do not directly show that a bureaucracy’s minority clients will make demands about minority bureaucrats’ discretion, they do indicate that minority clients place a premium on minority representation. That minority clients value minority bureaucratic representation suggests that it would be plausible for them to want minority bureaucrats to have the discretion necessary to better serve their interests.

Related to the studies described above, a small but growing body of representative bureaucracy research is concerned explicitly with symbolic representation, or the idea that minority bureaucratic representation legitimizes a bureaucracy’s actions in the eyes of minority clients. In other words,

members of different groups may feel better about public agencies simply because they believe that people who share their demographic characteristics will share their policy views, will be responsive to them, will not discriminate against them, and/or will act in their interest. (Keiser, 2010, p. 1206)

Research shows, for instance, that minority citizens perceive traffic stops to be more legitimate when the police officer making the stop is also a minority (Theobald & Haider-Markel, 2006). Similarly, female victims of sexual assault are more likely to file a formal report when their local police station has a high proportion of female officers on staff (Meier & Nicholson-Crotty, 2006). Again, these findings are not direct evidence that a bureaucracy’s minority (or female) clients explicitly demand that minority bureaucrats be given more discretion, but they do help build the case that minority clients want to be represented within the bureaucracy by individuals who share their race. As Theobald and Haider-Markel (2009) sum up in their own study of symbolic representation, “for population sub-groups, such as racial and ethnic minorities, real value is placed on having members of the group in elected or nonelected [emphasis added] positions in government” (p. 411).

Before moving to the second of our three causal mechanisms, we want to make two important points, both of which have implications for our empirical analysis. As noted at the outset of this section, our expectation is that minority parent demand for minority teacher discretion will be directly proportional to the prevalence of minority students in a school’s total student population. By extension, we expect to observe a linear relationship between our client demographic variables (e.g., % African American students) and minority teacher discretion. Our first point is that the expectation of a linear relationship should be considered a baseline, against which other expectations can be compared. A plausible alternative expectation is that the relationship between client demographics and minority teacher discretion hinges on the presence of a critical mass of minority students. In this view, client demographics have no effect on minority teacher discretion up to a certain point, but once that point is reached, an effect manifests.

The theoretical notion of a critical mass is a familiar one in the representative bureaucracy literature, where the expectation is that passive representation translates into active representation only when minority bureaucrats constitute a nontrivial percentage of a bureaucracy’s total workforce (see, for example, Keiser, 2010). Traditional representative bureaucracy studies have not yet identified the causal mechanisms that underpin this process, but a plausible argument is simply that minorities (whether they are elected or nonelected officials) become more assured of their actions and more likely to win over others once their numbers reach some threshold. Mansbridge (1999) makes this argument thus:

representatives of minority groups may need a critical mass for their own members to become willing to enunciate minority positions. They may also need a critical mass to convince others—particularly members of dominant groups—that the perspectives or insights they are advancing are widely shared, genuinely felt, and deeply held within their own group. (p. 636)

We believe that this reasoning applies in the context of our study. Once minority parents constitute a significant share of a school’s total parent population, it is reasonable to expect that they will be more likely to air demands—including demands about minority teacher discretion. Moreover, once these demands accumulate to a considerable threshold, it becomes much more likely that they will be heard.

Evidence suggests that the application of this reasoning in our study is warranted. Kerbow and Bernhardt (1993) find, for instance, that African American and Hispanic parent involvement in Parent–Teacher Organizations (PTOs) is greater in schools with high concentrations of African American and Hispanic students. Once members of these parent subgroups constitute a considerable share of a school’s total parent population, it appears that they become more comfortable formally participating in school governance. Weiher (2000) also offers evidence that speaks to our argument. Specifically, he shows that growth in the proportion of a school district’s total students who are Hispanic begins to have a positive impact on Hispanic student performance once Hispanic students reach critical mass. Conversely, decline in the proportion of a district’s total students who are African American has a negative impact on African American student performance only when African American students lose critical mass. These findings are relevant to our argument about parent demand for the underlying political dynamics they hint at. As Weiher (2000) notes,

Districts adjust their focus to address the needs of Hispanic students when Hispanics become a politically potent constituency. Furthermore, the fact that decreases in proportions of Black students in districts that are not heavily Black depress the performance of these students suggests that Blacks may be passing the same political threshold in the opposite direction. As proportions of African-American students in school districts decline, the districts become less attentive to them. (p. 894)

In the same way, it could be the case that minority parent demands for minority teacher discretion become politically efficacious once minority parents attain critical mass and cease to be efficacious once critical mass is lost.

The second point we want to make concerns how parents’ involvement in their children’s schools tends to change as their children get older. A good deal of evidence indicates that parent involvement is considerably higher at lower grade levels and that it is particularly high at the elementary school level (e.g., Greenwood & Hickman, 1991; Herrold & O’Donnell, 2008; Izzo et al. 1999). We call attention to this evidence because our client demand mechanism rests on the assumption that minority parents make demands about minority teacher discretion. While we cannot directly test this assumption in the current study, the relationship between parent involvement and school level noted above suggests that the parent demand causal mechanism is more likely to operate at lower grade levels than higher grade levels. If parents are more involved in their children’s schools at lower grade levels, it is reasonable to think that they will have a greater number of formal and informal interactions with school officials at these lower grade levels and that they will be more inclined—and have more opportunities—to make demands that are relevant to their children’s schooling. Consequently, we expect to observe a stronger relationship between client demographics and minority teacher discretion at lower grade levels than at higher grade levels. If such a relationship does exist in our data, it would make the client demand mechanism more plausible.

Managerial Deference to Expertise

In the context of our study, managerial deference to bureaucratic expertise refers to the idea that public school principals recognize minority teachers’ unique capacities for relating to and instructing minority students and, recognizing these capacities, grant minority teachers the latitude to act on them. For our purposes, then, “expertise” includes the special relational abilities that minority teachers possess vis-à-vis minority students, in addition to any unique technical skills they enjoy. ³ As with our client demand mechanism, a few key assumptions are built into the managerial deference mechanism. One assumption is that minority teachers actually do possess expertise vis-à-vis minority students. A second assumption is that principals will permit minority teachers the discretion to deploy that expertise.

The assumption that minority teachers are uniquely capable in the instruction of minority students is common in the representative bureaucracy literature (e.g., Meier et al., 1991, 1999; Meier & Bohte, 2001). Meier and Bohte (2001) note, for instance, that “minority teachers might well have additional insight into how to motivate and teach minority students” (p. 459). Similarly, Meier et al. (1999) point out that “minority teachers are more effective at teaching minority students” (p. 1030). Research bearing on this claim suggests that minority teachers are indeed specially equipped to instruct minority students. Much of this research, which focuses on the relational component of expertise, shows that minority teachers’ interactions with and perceptions of minority students are more positive than those of their White colleagues (Aaron & Powell, 1982; Cole, 1986; Dee, 2005; Irvine, 1990; Ladson-Billings, 1994; Lipman, 1995; Moore & Johnson, 1983). Recently, Dee (2004) has shown that Black elementary school students ⁴ who are randomly assigned to Black teachers tend to perform better on standardized math and reading tests than Black students who are randomly assigned to non-Black teachers. While the causal mechanisms underlying this finding are not clear—whether relational or technical expertise (or some other factor) is at work is not demonstrated—the key takeaway is that Black students do in fact do better under Black teachers than White teachers.

The justification for our second assumption—namely, that principals will grant minority teachers the leeway to capitalize on their expertise—is primarily theoretical. Public managers are held accountable for the performance of their organizations (Lipsky, 1980). Like any manager in any industry, public managers must rely on their employees to work diligently in pursuit of organizational goals (Brehm & Gates, 1997; Riccucci & Meyers, 2004). Oftentimes, a public manager’s employees may possess expertise that the manager lacks. When this is the case, it is in the manager’s interest to defer to the employees’ expertise (Lipsky, 1980, pp. 24-25). Consider, for instance, a manager at NASA, a federal agency whose work is of a highly technical nature. If this manager presides over a group of employees who have expertise in different areas of rocket design, it is unlikely that she—even if an expert herself—will be knowledgeable in all of these areas. Consequently, the manager will have to cede decision-making authority to her employees if she wants a well-designed rocket. As Romzek and Dubnick (1987) emphasize in their landmark case study of accountability at NASA, deference characterizes the relationship between managers and employees in highly technical agencies like NASA (p. 229).

Schools are not as highly technical as NASA, but their employees often do possess expertise that managers (principals, vice-principals, school district officials) lack. This is especially true at higher grade levels, where subject matter can be quite specialized (e.g., physics, calculus, chemistry; Busher & Harris, 1999). When a public school teacher has subject matter expertise that his principal does not, it is reasonable to expect that the principal will defer to the teacher’s judgment regarding matters like textbook selection, instructional pacing, and course content. At the same time, it is also reasonable to expect that teachers will use their expertise to resist top-down attempts at limiting their discretion. As Little (1995) notes in her case study of public school restructuring, “teachers routinely employ claims to subject expertise and department affiliation in shaping their response to new formal structures . . . and to shifts in the balance between individual and collective autonomy” (p. 50). More generally, studies in the public management literature suggest that frontline employees will resist managers’ efforts to control their actions when employees feel they have firsthand experiential knowledge that managers lack (Sandfort, 1999, 2000).

At this point, it is important to reemphasize that expertise in the context of our study has a relational component. As noted above, evidence suggests that minority teachers are better than their White counterparts at connecting with minority students. Consequently, we expect that public school principals (or vice-principals, etc.) will more readily defer to minority teachers than White teachers as the percentage of a school’s total students who are minorities increases. This is not to say that principals will rigidly oversee White teachers while employing a laissez-faire managerial style with minority teachers. Rather, it is to say that on the decision-making margins, minority teachers’ decisions are less likely to be contested than White teachers’ decisions.

Bureaucratic Appropriation

Our final causal mechanism—bureaucratic appropriation—refers to the idea that minority bureaucrats will feel empowered to claim more discretion for themselves when they serve client populations that are heavily minority. Street-level bureaucracy theory suggests that bureaucrats value discretion because it allows them to cope with the difficulties, ambiguities, and conflicting obligations of their jobs (Lipsky, 1980; Maynard-Moody & Musheno, 2003; Watkins-Hayes, 2011). Consequently, it is reasonable to expect that they will, in general, make claims for more of it. But we are interested in a more specific explanation for why minority bureaucrats will appropriate more discretion when the client population they serve is composed of a large percentage of minorities.

A simple explanation is that they will feel justified in appropriating it, and therefore will be more likely to do so. According to Saward’s (2010) theoretical notion of a “representative claim,” minority bureaucrats have special standing to demand more discretion when the bureaucracies in which they work have large minority client populations. Saward (2010) defines a representative claim as “a claim to represent or know what represents the interests of someone or something” (p. 38) and argues that these claims are legitimated by various kinds of “resources,” including descriptive similarity between the representer and the representee. In the context of the current analysis, Saward’s (2010) argument implies that minority bureaucrats will be able to legitimately claim more discretion when the client population they serve comprises a large percentage of minorities.

Research on minority teachers’ attitudes about teaching minority students suggests that they would indeed feel justified in claiming more discretion. In interviews with 22 minority public school teachers, Carr and Klassen (1997) found that nearly all believe that minority teachers “play a crucial role in connecting to racial minority students” (p. 75). Interviewees tended to emphasize their relational expertise vis-à-vis minority students. One interviewee noted, for instance, “The racial minority kids are more comfortable with me. They’re polite to me, let’s put it that way. They cooperate more” (p. 75). Another pointed out, “As a minority teacher I feel I have more patience and understanding of new students lacking some of the social aspects of schooling” (p. 75). Similar to Carr and Klassen (1997), Foster (1998) interviewed 20 Black public school teachers and found that many saw themselves as being uniquely able to connect with and instruct Black students. While these findings do not demonstrate that minority teachers actually appropriate discretion in heavily minority schools, they do imply that minority teachers would view doing so as legitimate.

Research in the representative bureaucracy literature provides additional support for the argument that minority bureaucrats will feel warranted in appropriating discretion to act on behalf of minority clients. Specifically, this research suggests that minority bureaucrats acknowledge, and often even embrace, a special responsibility to act as representatives for minority clients. Selden at al. (1998), for instance, use data on 184 bureaucrats working in the Department of Agriculture’s Farmers Home Administration (FmHA) to show that minority bureaucrats more readily adopt a “minority representative role”—an orientation toward acting on behalf of minority clients—than do nonminority bureaucrats. Similarly, Martinez (1991) and Murray et al. (1994) adduce evidence in support of the proposition that minority bureaucrats view advocating for minority clients as a duty. Martinez (1991), for instance, uses data on 93 public administrators from California to show that Hispanic administrators are significantly more likely than White administrators to view administrative representation on behalf of the Hispanic community as a valid role obligation. Likewise, Murray et al. (1994) produce survey evidence suggesting that many Black administrators view advocacy on behalf of Black clients to be a legitimate undertaking. More recently, Bradbury and Kellough (2008) demonstrate, via a survey of local government administrators in Georgia, that Black administrators believe much more so than White administrators that local government officials should make efforts to promote policies that benefit Black citizens. For instance, they show that 86% of surveyed Black administrators (as opposed to just 17.4% of surveyed White administrators) agree with the statement that government employees “should actively advocate in favor of policies that specifically address the interests of African American citizens” (p. 707) Moreover, 96% of surveyed Black administrators (as opposed to 28.9% of surveyed White administrators) agree that government employees “should actively advocate for organizational change to ensure responsiveness to African-American interests” (p. 707). These findings do not demonstrate the actual appropriation of discretion by minority bureaucrats; nevertheless, they constitute convincing evidence that minority bureaucrats are inclined to act on behalf of minority clients. In our view, this acting on behalf of minority clients would imply the claiming of discretion.

Importantly, the operation of our first causal mechanism (client demand) may reinforce the operation of our bureaucratic appropriation mechanism. The idea here is that minority clients’ demands induce minority bureaucrats to claim discretion to act on their behalf. Scott (1990), for one, finds that Black school district superintendents feel that Black citizens “impose greater demands on black superintendents” than on White superintendents (p. 167). Scott (1990) further shows that Black school district superintendents assimilate these demands by making the advancement of Black interests a priority: 56 out of 62 agree that “black superintendents should identify with black-directed endeavors to resolve the problems and needs of black Americans in a racist society” (p. 168). In a pair of earlier studies, Mann (1974, 1975) elucidates a similar dynamic, showing that minority superintendents are subject to increased demands by minority clients and therefore feel compelled to advance a minority-oriented agenda. In short, minority client demand spurs minority bureaucrats to appropriate discretion and become active representatives of minority clients’ interests.

Empirical Implications

The above reasoning has two primary empirical implications, which we call Hypotheses 1 and 1a. First, and as already emphasized, minority teachers’ discretion will increase as minority students constitute a larger and larger share of a school’s total student population. Second, this relationship will not hold for nonminority (White) teachers. White teachers’ discretion will either remain the same or decrease as minority students constitute larger and larger shares of a school’s total student population. The relationship between minority student share and White teachers’ discretion is expected to be zero or negative because the three potential causal mechanisms linking minority client share to minority bureaucrats’ discretion do not apply to White teachers. Minority students (or their parents) will not demand that White teachers be given more discretion; managers will not defer to White teachers when it comes to relating to minority students; and White teachers working in heavily minority schools will not make claims for more discretion. ⁵ Accordingly, our hypotheses are as follows:

We will test H1 for two minority groups: African Americans and Hispanics. Figure 2 helps to clarify our expectations.

OPEN IN VIEWER

Figure 2.

The relationship between minority student share and teacher discretion, by teacher race.

We also test two additional hypotheses, both of which were alluded to in our discussion of the client demand causal mechanism. The first of these hypotheses concerns the relationship between a minority student critical mass and minority teacher discretion. Our expectation is that the existence of a minority student critical mass will be positively related to discretion for minority teachers but not for White teachers. There is no consensus in the representative bureaucracy literature on what, precisely, constitutes a critical mass (Keiser, 2010). Nevertheless, as Hindera and Young (1998) point out, “two a priori points are suggested from normative theories of democracy” (p. 663). These, they note, are the point at which a minority group constitutes a plurality and the point at which a minority group constitutes a simple majority. Consequently, we will test our third hypothesis using both points.

As Figure 3 shows, we expect a threshold effect: Minority teachers in critical mass schools will have more discretion than minority teachers in noncritical mass schools. ⁶

OPEN IN VIEWER

Figure 3.

Minority student critical mass and teacher discretion, by teacher race.

The second of our two additional hypotheses (and the third overall) is that the relationship between minority student share and minority teachers’ discretion will be more pronounced at lower grade levels than higher grade levels. Recall the reasoning that underlies this hypothesis: Parents are more involved with their children’s schools when their children are young, and so the parent demand mechanism will be more operative at lower grade levels than higher grade levels. Consequently, we should observe a stronger relationship between the percentage of minority students and minority teachers’ discretion at lower grade levels. Again, we expect no relationship (or a negative relationship) for White teachers. Figure 4 illustrates these expectations.

OPEN IN VIEWER

Figure 4.

Minority student share and minority teacher discretion, by grade level.

Data

The analysis sample for this study comprises 31,640 public school teachers from 6,400 public schools. ⁷ Data for these units are drawn from the 2007-2008 wave of the Department of Education’s Schools and Staffing Survey (SASS). The SASS is a quadrennial survey that collects nationally representative data from public school teachers in all 50 states as well as data about these teachers’ schools, principals, and school districts. Thus, for any given teacher, the researcher can link to data about the teacher’s principal, school, and school district.

In brief, the SASS’s sample selection procedure is as follows. It first draws a nationally representative sample of public schools (schools are the SASS’s primary sampling units), ⁸ each of which receives the SASS “school questionnaire.” Once a school is selected for inclusion in the sample, that school’s district and principal are automatically sent a “district questionnaire” and “principal questionnaire,” respectively. Finally, from within each of the sampled schools, a sample of teachers is drawn, and each of these teachers receives the SASS “teacher questionnaire.” The separate questionnaires elicit a wealth of information from each set of respondents (teachers, principals, schools, and districts). Below, we present our model of bureaucratic discretion and describe the variables that go into it.

Model

Though we are chiefly interested in the interaction between teacher race and student demographics (recall Figure 1), we begin by estimating a simple model of teacher discretion, one that does not include any interaction terms:

d i s c r e t i o n t , s = β 1 b l a c k t c h t , s + β 2 H i s p t c h t , s + β 3 p c t b l a c k s t u s + β 4 p c t H i s p s t u s + c o n t r o l s + μ s + ε t , s

We then estimate a model that does include interaction terms:

d i s c r e t i o n t , s = β 1 b l a c k t c h t , s + β 2 H i s p t c h t , s + β 3 p c t b l a c k s t u s + β 4 p c t H i s p s t u s + β 5 b l a c k t c h t , s ⋅ p c t b l a c k s t u s + β 6 H i s p t c h t , s ⋅ p c t H i s p s t u s + c o n t r o l s + μ s + ε t , s

We begin with the simple model because doing so helps to clarify the theoretical importance and substantive interpretation of the interaction terms.

Above, discretion_t,s is the discretion of teacher t in school s; blacktch_t,s is a dummy variable that equals one if teacher t (in school s) is African American and zero otherwise; Hisptch_t,s is a dummy variable that equals one if teacher t is Hispanic and zero otherwise; pctblackstu_s is the percentage of school s’s students who are African American; pctHispstu_s is the percentage of school s’s students who are Hispanic; blacktch_t,s⋅pctblackstu_s and Hisptch_t,s⋅pctHispstu_s are interaction terms; controls is a collection of control variables; µ _s is a random school-level error term; and ε _t,s is a random teacher-level error term. It is important to note that teacher race is a six-category variable whose categories are the following: White, Black, Hispanic, Asian American, Native Hawaiian or other Pacific Islander, and American Indian or Alaska Native. We “dummy out” this six-category race variable, creating a set of five indicator variables—one for each race category except White. Consequently, White teachers constitute our reference group throughout the analysis. ⁹ Table 1, which contains descriptive statistics, shows that there are 1,770 Black teachers and 1,200 Hispanic teachers in our sample. As the table further shows, the student bodies of the 6,400 schools in our sample are about 14.6% Black and 12.0% Hispanic, with a good deal of variation around these means.

OPEN IN VIEWER

Table 1.

Descriptive Statistics: Focal variables.

	Mean	SD	Min	Max
Discretion	16.88	2.77	5	20
White teachers only	17.00	2.72	5	20
Black teachers only	16.17	3.01	5	20
Hispanic teachers only	16.20	3.02	5	20
%Black students	14.55	24.17	0	100
%Hispanic students	12.04	20.32	0	100
	Freq.	%
Teacher race
White	27,520	86.98
Black	1,770	5.61
Hispanic	1,200	3.79
Asian American	450	1.43
Native Hawaiian or other Pacific Islander	90	0.28
American Indian or Alaska Native	600	1.91

Note. N = 31,640.

The above model’s two focal interaction terms serve to test the hypothesis that minority teachers’ discretion will increase with the percentage of a school’s total students who are minorities. Positive signs on these interaction terms’ coefficients would support this hypothesis. ¹⁰ The pctblackstu_s and pctHispstu_s variables test H1a, which states that White teachers’ discretion will either remain the same or decrease as the percentage of a school’s total students who are minorities increases. Since White teachers constitute our omitted race category, these variables’ coefficients represent, respectively, the effects that increases in the percentage of a school’s students who are Black and Hispanic have on White teachers’ discretion.

To test our second hypothesis, we estimate the following model:

d i s c r e t i o n t , s = β 1 b l a c k t c h t , s + β 2 H i s p t c h t , s + β 3 b l a c k s t u P l u r a l i t y s + β 4 H i s p s t u P l u r a l i t y s + β 5 b l a c k t c h t , s ⋅ b l a c k s t u P l u r a l i t y s + β 6 H i s p t c h t , s ⋅ H i s p s t u P l u r a l i t y s + c o n t r o l s + μ s + ε t , s

Above, blackstuPlurality_s is a dummy variable that equals one if school s’s Black students constitute a plurality and zero if they do not; HispstuPlurality_s is a dummy variable that equals one if school s’s Hispanic students constitute a plurality and zero if they do not; and blacktch_t,sblackstuPlurality_s and Hisptch_t,sHispstuPlurality_s are interaction terms. A significant positive coefficient for the blacktch_t,sblackstuPlurality_s interactions term would indicate that the existence of a Black student critical mass is positively related to discretion for Black teachers. A null coefficient for the blackstuPlurality_s variable would indicate that a Black student critical mass has no effect on White teachers’ discretion. Similar logic would apply to the coefficients of the Hisptch_t,sHispstuPlurality_s and HispstuPlurality_s variables. Note that we estimate the same model as above with variables measuring minority student majorities.

To test our third hypothesis, we estimate the above two models while restricting the estimation sample to include only elementary schools. We then estimate the above two models for all other schools. Comparing our results across the different estimation samples allows us to informally assess our third hypothesis.

Method

As indicated above, we begin with a simple model and add in complexity. The simple model contains a group of control variables that we call exogenous. We use the term exogenous somewhat informally, applying it to any variable for which reverse causality can reasonably be assumed to be a nonissue. These are mostly demographic or structural variables like school size, the percentage of a school’s students who are eligible for free lunch, and organizational span of control (the ratio of vice-principals to teachers). Our more complex models include these exogenous controls as well as additional controls that we call endogenous, given that they do raise the issue of reverse causality. One such variable, for instance, is a measure of red tape: It is plausible to think that red tape affects teachers’ discretion, but it is also plausible to think that teachers with lots of discretion can choose to ignore red tape. Ultimately, whether the endogenous variables are included in our models does not impact our results in a substantively meaningful way. Nevertheless, we do want to acknowledge this potential problem. Our models include a long list of controls, measured at both the teacher level and the school level. To preserve space, and to keep the focus of our article on our variables of theoretical interest, we do not catalogue our control variables here. Instead, we list them in our control variables appendix (Table A1) along with the details of their measurement and brief comments regarding their purpose.

We estimate each of the models shown above with school district fixed effects, which control for any district to district differences that could be correlated with our independent variables of interest as well as our dependent variable, teacher discretion. The district fixed effects are particularly important because different school districts may have different formal policies in place that affect how much discretion teachers have. These policies might also limit or enable the operation of our three causal mechanisms. For instance, district policies could restrict formal parent involvement in school governance (stifling our parent demand mechanism). The inclusion of school district fixed effects in our models accounts for this kind of possibility. ¹¹ More generally, the district fixed effects give us more confidence in the validity of our results.

Measuring Discretion

Scott (1997) has defined discretion as “a range of choice within a set of parameters that circumscribes the behavior of the individual service provider” (p. 37). Formal policies will likely set the boundaries of teacher discretion; within these boundaries, different teachers will have more or less room to exercise their own judgment. As we noted above, since our models include school district fixed effects, any impact that formal, district-level policies have on teacher discretion is accounted for in our analysis.

We measure discretion using the sum of a teacher’s answers to five survey items. Teachers are asked to rate the amount of control they have over various areas of teaching and planning. These areas are as follows: (a) selecting textbooks and other instructional materials; (b) selecting content, topics, and skills to be taught; (c) selecting teaching techniques; (d) evaluating and grading students; and (e) determining the amount of homework to be assigned. Response options include (1) no control, (2) minor control, (3) moderate control, and (4) major control. Thus, if a teacher were to report major control for all five items, his score on our multi-item index of discretion would be 20. By contrast, if a teacher were to report no control for all five items, his score would be five. On average, teachers have a good deal of discretion—the mean value of the discretion index for our sample of 31,640 teachers is 16.88. The minimum sample value for the index is five and the maximum is 20; the standard deviation is 2.77. Black and Hispanic teachers tend to have less discretion than White teachers. For Black teachers, mean discretion is 16.17; for Hispanic teachers, mean discretion is 16.20; for White teachers, mean discretion is 17.00. These figures are shown in Table 1.^12.13