Abstract
This article explores how symbolic representation can increase behaviors associated with cooperation among immigrants in an educational setting. It posits that, due to a lack of trust and efficacy in public institutions, undocumented immigrants are less likely to engage in activities that are conducive to cooperation and compliance. However, this relationship is conditional on the presence of passive representation. In settings where immigrant interests are represented, even passively, immigrants are more likely to engage in cooperative behaviors. Using data from Texas school districts, the analysis finds some support that passive representation can enhance symbolic representation among this population. It finds that assessments of immigrants’ cooperative behaviors are likely to decrease as the size of the undocumented student population increases. However, this is only the case in schools with low levels of representation. This supports the expectation that symbolic representation can enhance assessments of cooperative behaviors among undocumented immigrants.
Keywords
Introduction
For more than 50 years, scholars of public administration have studied representative bureaucracy and how it can produce more efficient, effective, and equitable outcomes for citizens, particularly those who lack representation in political institutions. This article focuses on an increasingly salient subset of individuals who receive public services but who largely lack representation in our bureaucratic and political systems, namely, undocumented immigrants. As undocumented immigrants lack direct political representation (i.e., they cannot vote or hold many public offices), bureaucratic representation provides a particularly meaningful substitute for the benefits of representation within public institutions.
Specifically, I examine whether passive representation can produce symbolic representation for undocumented students and their families in a public education setting. I examine a relatively new concept within public administration—symbolic representation—which focuses on how bureaucratic passive representation can change not only the behavior of bureaucrats (i.e., active representation) but also the behavior of citizens. As discussed below, symbolic representation contends that passive representation within bureaucracies can lead to greater trust, efficacy, and cooperation among represented citizens. This then can result in cooperative behaviors in citizens that can lead to better public policy outcomes for these groups. This article proposes that such cooperative behaviors among undocumented immigrants will be lower when the schools they attend have lower levels of passive representation. Using data from Texas public schools, I find that the relationship between undocumented students and reported cooperative behaviors is conditional on the level of passive representation.
Representative Bureaucracy
Bureaucratic representation has traditionally been separated into two sorts: passive and active. Passive representation “concerns the source of origin of individuals and the degree to which, collectively, they mirror the society” (Mosher, 1968, p. 12). Active representation occurs when “an individual is expected to press for the interests and desires of those whom he is to represent, whether they be [sic] the whole people or some segment” (pp. 11–12). Studies of representative bureaucracy began with examinations of passive representation to determine whether the bureaucracy passively resembles the population on key demographic measures, such as gender, education, or race (Atkins & Wilkins, 2013; D. Brown, 1999; Dolan, 2000; Esman, 1999; Kelly & Newman, 2001; Meier & Nicholson-Crotty, 2006; Riccucci & Saidel, 2001; Theobald & Haider-Markel, 2009). Numerous studies have found that people of color and women are passively represented in the bureaucracy, but usually not at higher levels within the organization (see Selden, 1997).
A second line of research (although not mutually exclusive) is concerned with passive representation within the bureaucracy and its effect on organizational performance (G. Brown & Harris, 1993, 2001; Brudney et al., 2000; Dolan, 2000; Hindera, 1993b; Keiser et al., 2002; Meier, 1975; Meier et al., 1989; Meier & Nigro, 1976; Saltzstein, 1979; Selden, 1997). This research generally finds a positive relationship between passive representation, organizational performance, and policy outcomes targeted toward represented groups. Finally, a third branch is concerned with linking passive representation with the potential for active representation. Studies have attempted to explicitly assess the relationship between demographic characteristics and values or attitudes toward specific policies (Meier & Nigro, 1976; Wilkins & Keiser, 2006). This link between demographic characteristics and values is a central assumption that connects passive and active representation. While the findings have not been definitive, they tend to find that race/ethnicity is an important predictor of whether passive representation will result in active representation. As Guinier (1994) argues, “racial-group membership often serves as a proxy for shared experience and common interests” (p. 137). These common experiences and interests presumably result in shared values, which are then associated with decisions and behaviors.
Furthermore, a considerable body of research addresses the question of when passive representation will translate to active representation. Meier and Stewart (1992) found that the presence of Black schoolteachers resulted in positive policy outcomes for Black students. This finding also held for Latinos (Meier, 1993; Nicholson-Crotty et al., 2016) and women (Keiser et al., 2002; Selden, 1997; Wilkins & Keiser, 2006a). Hindera (1993a, 1993b) found similar results in Equal Employment Opportunity Commission (EEOC) district offices for both African Americans and Latinos. In fact, one study of Texas school districts found that both minority and nonminority students performed better in a more representative bureaucracy (i.e., more minority teachers; Meier et al., 1999). This may imply that representative bureaucracies are more efficient than nonrepresentative bureaucracies under similar circumstances.
Symbolic Representation
More recently, scholars have identified a third type of representation—symbolic representation (Gade & Wilkins, 2013; Hong, 2016; Meier & Nicholson-Crotty, 2006; Riccucci et al., 2014; Riccucci & Van Ryzin, 2016; Theobald, Haider-Markel, 2009). Symbolic representation argues that the presence of passive representation can actually produce changes in citizens and clients themselves, without bureaucrats even needing to take any representative actions (i.e., active representation). Citizens who interact with bureaucracies that are passively representative of their communities will be more likely to develop a greater level of trust and efficacy in these organizations and perceive them to be more legitimate than organizations that are not passively representative. This enhanced sense of trust will then result in greater levels of cooperation, engagement, and compliance of citizens. The reverse may also be true. Meier and Hawes (2009) go as far as to argue that the civil unrest and riots that France experienced in the mid- to late-2000s were due in part to the lack of passive representation among the French civil service and police force. This could be seen as an example of how the lack of passive representation can lead to a lack of trust and produce the opposite of cooperation and coproduction, even to the point of conflict and revolt.
Figure 1 presents a visual model of the dynamics between passive, active, and symbolic representation as developed by Riccucci and Van Ryzin (2016). As a substantial body of research demonstrates, social origins can shape values and actions of bureaucrats, which can then affect policy outcomes (i.e., active representation). However, Riccucci and Van Ryzin (2016) argue that a similar dynamic occurs for citizens where social origins—and the perceived match with bureaucrats—can lead to enhanced trust and perceived legitimacy, resulting in increased compliance and cooperation. This, in turn, can result in positive policy outcomes.
Passive representation, active representation, and symbolic representation.
For example, Theobald and Haider-Markel (2009) found that African Americans’ trust of police is higher when there are more African American police officers, thus increasing citizen cooperation and compliance. Meier and Nicholson-Crotty (2006) found that women were more likely to report sexual assault in cities with more women police officers. Riccucci et al. (2014) confirm that this result is due to women citizens’ increased perceptions of trust and fairness within the police force due to the presence of women police officers. Similarly, Gade and Wilkins (2013) find that veterans report higher levels of satisfaction with rehabilitation services when their counselors are veterans. These findings all suggest that citizen perceptions, and even behaviors, are altered by the composition of the bureaucracy.
Cooperation and Coproduction
Symbolic representation is closely linked to the concept of coproduction. Put simply, coproduction entails activities where citizens cooperate with public agencies in producing public goods and services (Brudney & England, 1983). That said, the literature on coproduction is extensive, varied, and has grown significantly over the past several decades. Traditional conceptualizations of coproduction primarily pertaining to citizen involvement in the delivery of public goods and services have given way to a substantially broader range of involvement, governance, management, and assessment (Brandsen & Pestoff, 2006). At its core, however, coproduction argues that the production and consumption of public services are inseparable (Bovaird & Loeffler, 2013). Interaction and cooperation between the provider and consumer are essential to the quality of the services rendered. When coproduction is absent, the delivery of public services is more likely to be inefficient, inappropriate to clients’ needs, and less likely to benefit from consumer assessment and feedback.
While there are many typologies for coproduction, Osborne and McLaughlin (2004) define three types of coproduction. Co-governance entails citizens or nonprofits engaged in the planning and delivery of public goods and services. Co-management involves a third party (e.g., nonprofit or community organization) producing services on behalf of, or in collaboration with, a public agency. Finally, coproduction, in a more limited definition, is where citizens produce their own services, if only in part. Similarly, Bovaird and Loeffler (2013) present at least four types of coproduction, including co-commissioning, co-designing, co-delivery, and co-assessment. Furthermore, they argue that the principles of coproduction include the idea that citizens can make public services more effective by cooperating with program requirements and engaging in collaborative relationships with public servants and other constituents.
This form of coproduction can include informal activities including parental involvement in public education, engaging in proactive activities to facilitate the delivery of public services, and participating in educational activities that are aimed at increasing the efficacy of public service delivery (Jacobsen & Andersen, 2013). Thus, coproduction in this sense does not require formal organizations engaged in management, planning, or assessment. Rather, it is ordinary citizens, either individually or collectively, engaging in activities that facilitate the delivery of public goods and services. As Bovaird and Loeffler (2013) explain, [I]n schools, outcomes not only depend on the quality of teaching delivered by schoolteachers but also on the attitudes and behaviour of students. If students are not willing even to listen, or [are] not prepared to carry out the follow-up work at home, the amount that they learn will be very limited. (p. 4)
Thus, coproduction can include citizen activities and behaviors at an independent or micro level, separate from formal organizations or roles. Furthermore, coproduction necessitates cooperative and compliant behaviors from citizens. While these cooperative attitudes and behaviors in and of themselves may not constitute coproduction, they are precursors, in that they can lead to behaviors, relationships, processes, and outcomes associated with coproduction. Thus, student and parental attitudes and behaviors that support cooperation and compliance with school administrators can be important precursors and predictors of coproduction.
Symbolic Representation and Citizen Cooperation
The connection between symbolic representation and coproduction is often implied but not always explicitly articulated. The intuition is clear, however, in that symbolic representation is purported to increase citizens’ trust, efficacy, and cooperation with public agencies that represent their interests, even passively. To the extent that these qualities lead to increased collaboration and engagement in activities that facilitate the delivery of public goods, coproduction will be enhanced. Thus, cooperation and compliance are important foundation blocks of coproduction.
Indeed, in an experimental study, Riccucci et al. (2016) found that women were more likely to report intentions to recycle when they believed the public recycling program was run by women. In this case, perceived gender representation within a government program changed the behavior of women citizens resulting in increased intentions of cooperation and, hence, an increased chance of coproduction. In the case of criminal justice, symbolic representation could result in greater cooperation and compliance with police investigations, as well as the creation and maintenance of informal social controls (e.g., Meier & Nicholson-Crotty, 2006). This results in better policy outcomes for members of these communities. Similarly, within an educational context, citizen cooperation may include increased parental involvement, volunteering, and participation in school initiatives and activities. It could also include changes in the behavior of students (increased attendance, fewer discipline problems, etc.). These behaviors should result in increased educational performance and better educational outcomes. To the extent that symbolic representation is at play, this process occurs due to passive representation, without the bureaucrats themselves necessarily taking direct action.
Hong (2016) builds on this work by articulating four conditions that must be met before we see the positive relationship between passive representation and coproduction. 1 First, he argues the policy area in question must be salient to the clientele being represented by the bureaucrats. Furthermore, the demographic trait (e.g., ethnicity or gender) must be sufficiently salient so that bureaucrats’ decisions are at least partially influenced by these characteristic traits. Second, a considerable level of discretion should be present in bureaucrats’ decision-making process. Third, the represented group in question should compose a significant percent of the clientele. Finally, bureaucrats must have direct contact and interaction with citizens. Hong argues that if these conditions are met, we are more likely to witness coproduction in public services as the result of passive representation and, in turn, policy outcomes that benefit the represented group.
Undocumented Students and Education
This research examines how symbolic representation can result in behavioral changes conducive to improved education goals among undocumented immigrant students and their families. The literature on Latino and immigrant education provides some guiding principles upon which to build expectations. Latinos (including immigrants of different legal statuses) tend to underperform in formal assessments compared with their White, non-Latino counterparts (Crosnoe et al., 2011; Gibson & Carrasco, 2009; Meier & Stewart, 1991). While the literature on Latino education and performance is extensive, there is far less research (particularly quantitative) on immigrants, especially undocumented immigrants.
The primary reason there is a lack of research on undocumented students’ performance is due to Plyler v. Doe (1982), a Supreme Court decision in which the Court ruled that all children – regardless of citizenship or immigration status – have a right to access public education. As a result of this court decision, most states and school districts do not collect the citizenship or immigration status of students. Hence, there is little work (see Hill & Hawes, 2011, for an exception) that addresses the actual performance of undocumented students because they are difficult to identify. There is, as noted above, a considerable amount of work on Latino educational performance (of which undocumented students are a subset) but little that focuses on undocumented students.
Meier and Stewart (1991) contend that Latino students are denied the same educational opportunities that their non-Latino counterparts enjoy. In fact, discrimination leads to, among other things, higher drop-out rates among Latinos. Others (E. E. Garcia, 2001; Ochoa, 2003) have argued that Latino students are treated as second-class citizens, and as such, they are denied many opportunities (such as access to core curriculum and adequate access to college). Relying on social stratification theory, E. E. Garcia (2001) posits that there is a connection between the cultural match of school and home, where home environments that prioritize education will see better results in school. However, E. E. Garcia (2001) also argues that social resources are key, and their absence can result in poor performance, which may account for educational inequalities among Latinos and non-Latinos.
Hill and Hawes (2011) draw several conclusions from this research. First, undocumented students have cultural biases against them—both tacit and explicit. For example, English is frequently not the primary language spoken in their homes or even their community; however, state exams are predominately administered in English, thus creating a disadvantage for Spanish-speaking children. Second, as a subset of Latino students, undocumented students presumably will have lower academic performance than their Anglo and African American counterparts. Latino students have a significantly higher risk of dropping out than their peers (J. A. Garcia, 2016; Leal et al., 2004). Research also finds that migrant students tend to perform less well academically than nonmigrant students (Bean et al., 2011; Menchaca, 2003). Arguably, this is due to a lack of culturally relevant or appropriate materials and teaching techniques (Menchaca, 2003). If culturally relevant teaching methods are key to test performance of migrant students (and perhaps undocumented students more generally), then it may be important to consider the composition and characteristics of teachers in schools. While most of this research examines academic performance (mostly measured in terms of state tests), it is reasonable to assume that we might see a similar pattern with other forms of academic life, including participation and engagement in school activities (which are linked to test performance).
Symbolic Representation, Undocumented Students, and Cooperation
Putting together the respective empirical and theoretical work examining symbolic representation and undocumented students, we may posit a relationship between undocumented students and increased cooperation/compliance. As this population can face significant educational and cultural barriers, participation in formal educational institutions may be more limited. This is especially the case in schools that have not created explicit programs designed to facilitate participation and involvement from this population. Thus, a baseline expectation from the education literature suggests that there will be fewer activities and behaviors associated with coproduction among undocumented immigrants, all else being equal. This, again, is largely due to systemic, cultural, and instructional barriers that exacerbate the disadvantages these students already face. From this, we can posit the following hypothesis:
However, the literature on representative bureaucracy suggests that passive representation can not only result in active representation but can also change the behavior of citizens and clients (i.e., symbolic representation). It may be that undocumented students are only likely to be disengaged (H1) when they feel disconnected from educational institutions and feel that these institutions do not represent their interests. When there is representation and a resulting perception of interest, we should see an increase in trust and efficacy, which will lead to increased cooperation and behaviors associated with coproduction. This suggests that H1 is conditional on the level of passive representation. Put formally,
Taken together, I posit that, in general, assessments of cooperative behavior are less likely in high undocumented immigrant settings; however, this relationship can be moderated by the presence of Latino teachers (i.e., passive representation). Figure 2 presents the proposed relationship between representation, symbolic representation, and cooperation. The relationship between passively represented citizens and levels of cooperation is not inherently clear. Rather what is important is the actual level of representation rather than the size of the represented group, per se. For example, in the case of undocumented students, schools with more undocumented students are expected to report fewer cooperative activities. Undocumented students and parents are less likely to become engaged in formal institutions and programs due to fears of deportation. They are more likely to “fly under the radar” than to become highly involved in public institutions, which could increase their risk of exposure. This is the basis for H1. However, passive representation can change this by working in at least two ways via symbolic representation. First, we expect it to have a direct effect akin to active representation. This is where Latino teachers’ decisions, behaviors, and direct actions lead to increased opportunities for cooperation for Latino students, including undocumented immigrants. These actions can include increased efforts to engage immigrant children, making and maintaining connections to parents, and adopting culturally appropriate teaching techniques to better relate to immigrant students. Thus, we expect a direct relationship between Latino teachers and cooperation among immigrants. As this form entails direct actions from bureaucrats, it is denoted as active symbolic representation. This is distinct from active representation because it still requires citizens to change their behavior. Thus, it is clearly under the category of symbolic representation, but distinct because it assumes direct action on the part of the bureaucrat.
A model of undocumented students, symbolic representation, and cooperation.
The second path is via symbolic representation in a more passive manner. Here, the mere presence of Latino teachers moderates the negative relationship between undocumented students and coproduction. This is due to increased trust and efficacy that immigrants have in the school system due to a recognition that their interests are represented, even passively. This is denoted as passive symbolic representation because there is no direct action on the part of the bureaucrat, but rather passive representation moderates the relationship between the target group and behaviors. The final element in the model is organizational support for cooperation and coproduction. Organizations that actively pursue strategies to build trust and connection with citizens will be more likely to witness cooperation (Bovaird & Loeffler, 2013). Finally, increased cooperation will increase the likelihood of coproduction, as denoted in the final step in Figure 2.
Thus, in this model, symbolic representation is manifested as both a direct and a moderating relationship between undocumented students and cooperation. If passive representation is absent or inadequate, undocumented immigrants are more likely to disengage, resulting in a negative relationship between the presence of undocumented students and assessments of cooperation/compliance. However, when passive representation is present in adequate quantities, undocumented immigrants are expected to become more engaged, resulting in a positive interactive relationship that captures the effect of symbolic representation.
Data and Method
This analysis utilizes data from Texas public schools. The data come from three sources: (a) the 2012 Texas Middle Management Survey (TMMS), (b) campus-level accountability data from the Texas Education Agency (TEA), and (c) a unique dataset measuring the number of undocumented students originally developed by Hill and Hawes (2011).
The TMMS (Thomas et al., 2011) is a survey conducted by the Project for Equity, Representation, and Governance (PERG) at Texas A&M University. The 2012 survey was administered between February and June in 2012 and was sent to over 5,000 Texas public school principals, of which approximately 1,034 responded (21% response rate). The survey asks school principals a series of questions related to school management, problems, resources, and policies.
The TEA campus-level data consist of publicly available measures of Texas public schools. These include student demographics, statistics on staff composition, financial data including expenditures, and other school characteristics. Finally, the Hill and Hawes (2011) dataset consists of data acquired from the TEA that provides counts of all students in Texas public schools who did not provide documentation upon registering for school. Hill and Hawes (2011) use these data to develop a school district measure of the undocumented student population in Texas schools. For the present analysis, I aquired campus-level data from the TEA to create a campus-level measure of undocumented students following the method developed by Hill and Hawes (2011).
Dependent Variables—Assessments of Cooperative Behavior
The analysis uses six variables to capture assessments of behaviors that are associated with cooperation/compliance—a precursor for coproduction. Coproduction can be defined as activities in which citizens cooperate with public agencies to produce a public service or good (Brundey & England, 1983). Within the context of education, a primary way citizens can coproduce is through parental involvement and participation in their children’s education (Bovaird & Loeffler, 2013). I use six dependent variables from the TMMS that measure principals’ assessments of cooperative behaviors of students and parents. Principals were asked to rate the quality of Parental Involvement and Community Support in their school on a 4-point scale: 1 = inadequate, 2 = below average, 3 = above average, and 4 = excellent. Principals were also asked how frequently they met with Parent Organizations, such as a Parent-Teacher Association (PTA). This variable is an ordinal scale and ranges from never (1) to very frequently (6). Schools that have a greater degree of cooperation from their parents and community members should, all else being equal, have higher assessments of parental involvement, community support from principals, and greater frequency of contact with school administration. This is true, in part, because, in an education setting, parental involvement and other forms of support are primary ways in which parents and community members can engage in education and coproduction.
Principals were also asked a series of questions that directly relate to immigrant students and families. Specifically, they were asked whether they considered a Lack of Parental Involvement, Attendance, or Disciplinary Problems to be among “the biggest issues in educating immigrant students.” Schools that experience low parental involvement from the parents of immigrants are arguably less engaged in coproduction among this group (i.e., immigrants). The rationale for this is the same as for the general parental involvement variable; however, this variable more directly measures parental involvement—a precursor for coproduction—within the immigrant population. The second measure captures, arguably, a low bar for cooperation/compliance: attendance. There is evidence that attendance is a key predictor of student success (e.g., test performance, completion rates; Balfanz & Byrnes, 2006; Corville-Smith, 1995; Epstein & Sheldon, 2002; Lamdin, 1996; Nichols, 2003). Attendance can be thought of as an important action by citizens and a necessary step in producing a quality education. This activity, arguably, more closely captures parental effort than student effort, particularly in lower grade levels. Attendance problems, therefore, can be thought of, in part, as less effort on the part of parents to cooperate with and coproduce in the education system. Finally, students with disciplinary problems could indicate a lack of student cooperation/compliance with staff, which would make coproduction less likely in the long run. The responses in the survey for all three items were measured as dichotomous variables where a value of 1 indicates that the principal identified a problem (“Yes”). For the purposes of clarity within the analysis, these items are reversed so that a value of 1 indicates greater assessments of cooperative behavior.
Finally, it is important to note, these variables are not direct measures of citizen behavior; rather, they rely on a professional assessment of citizens’ behaviors that should be correlated with actual behavior. 2 At a minimum, these items capture principals’ perceptions of cooperative behaviors of students and parents, including immigrants. If the hypothesized relationships are supported, there are at least two possibilities. First, the effect of passive representation results in symbolic representation in the form of changed citizen behavior that is conducive to coproduction. Alternatively, it could be that principals’ assessments do not accurately reflect citizen behavior. However, if passive representation increases the likelihood that principals will see citizen behavior—particularly immigrant behavior—as being more fostering of coproduction, this is still an important finding. Symbolic representation may not only change the behavior of citizens, it may also change the behaviors, values, and perceptions of the organization including coworkers and supervisors toward groups who are passively represented (Lim, 2006).
Independent Variables
Passive Representation
To measure Latino passive representation, I use a variable measuring the percent of teachers within each district who are Latino. 3 The median value for Latino teacher representation is 8% (M = 22), but the variable ranges from 0% to 100%. About 18% of the usable cases have at least 50% Latino teachers, while about 15% have no Latino representation.
Undocumented Students
The analysis uses data and a method developed by Hill and Hawes (2011) to estimate the undocumented student population. Students who register in Texas public schools are not required to report their immigration status. However, the state of Texas has implemented a tracking system for students called the Person Identification Database (PID). This database assigns a unique identifier for each student based on several identifiers, including social security numbers. When students register, they must provide a social security number; however, if they do not provide one, they are assigned an alternative number in its place. These identifying data are available by race and ethnicity, which can be used to identify all Latino students within each district who did not provide a social security number during school registration.
Hawes and Hill use these data to create a measure that approximates the percentage of undocumented students in each school district and campus. Their adjusted measure is the percent of Latino students with alternative PID numbers minus the percent of non-Latino students who had alternative PID numbers, that is,
Adjusted undocumented measure = (Count of alternative PID Latino students/Total Latino students) – (Count of alternative PID non-Latino students/Total non-Latino students). 4
While not all students in this measure are undocumented, all undocumented students enrolled in public schools (provided they do not provide a falsified social security card or documentation) should be included in these figures. At the very least, these measures should be highly correlated with true undocumented student enrollments. 5 Given these considerations, this measure should serve as an adequate proxy for undocumented students in Texas public schools and is arguably the best measure presented to date. Figure 3 presents the district aggregates of an adjusted measure of Latino undocumented students as a percentage of total enrollments. We see here that, as a percentage of total student enrollments, the figures tend to be highest in more heavily populated areas, along the border, and in the panhandle—an area with high agriculture and livestock production.
Adjusted measure of percent Latino undocumented students.
Organizational Facilitation of Coproduction
I also use five variables from the TMMS that capture the level of organizational support a school has created that facilitates coproduction. The survey asks principals if their school has any of the following are true regarding programs or services offered by the school: A staff member (a) is assigned to work on parent involvement; (b) has a service that allows parents to retrieve homework assignments; (c) has workshops or courses for parents or guardians; (d) has services to support parent participation, such as providing child care or transportation; and (e) has a parent drop-in center or lounge. Each variable is dichotomous (Yes/No). As Figure 2 illustrates, I expect schools that offer more services that facilitate parental involvement will have higher levels of coproduction, all else being equal.
Control Variables
A number of additional control variables are also included. These fall into three broad categories: financial resources, student characteristics, and teacher characteristics. Financial resources are captured using the percent of total expenditures that are spent on instruction. Schools with more resources devoted to instruction may experience higher levels of parental and community involvement. Student characteristics include the following: (a) the percent who have Limited English Proficiency (LEP), (b) the percent of students who are economically disadvantaged, and (c) the size of the school measured as total student enrollment.
Teacher characteristic controls include variables that measure teacher experience and class size. Teacher experience is operationalized as the average number of years of professional experience for teachers in the school. Class size is measured as the student–teacher ratio. Teachers with large student class sizes are likely not able to spend as much time with each student, which hampers their ability to form relationships with students and parents that are conducive to coproduction. Table 1 presents summary statistics for all of the variables in the model. As the dependent variables are either ordinal or dichotomous, logistic regression is used (ordered logit for the ordinal variables). 6 The models use robust standard errors clustered by school district because there can be multiple school campuses within a single district. 7
Summary Statistics.
Note. TEA = Texas Education Agency; LEP = Limited English Proficiency.
Findings
Table 2 presents the results from the baseline logistic regression models. The expectations are that (a) undocumented students will be negatively associated with assessments of cooperation and (b) programs that facilitate coproduction will be positively associated with assessments of cooperation. Passive representation (percent Latino representation) is not necessarily expected to have a direct effect on principals’ assessments for the general measures of cooperation (Models 1–3) because symbolic representation should only occur for targeted groups (Latinos, immigrants). Therefore, the expectation is that passive representation will only improve cooperation for Models 4 to 6 (the immigrant-specific dependent variables).
Baseline Models.
Note. Robust standard errors in parentheses, clustered by district. LEP = Limited English Proficiency.
p < .1. **p < .05. ***p < .01.
The results in Table 2 do not provide strong support for H1; that is, with the exception of the frequency of contact with parent organizations, an increased presence of undocumented students is not generally associated with significantly lower assessments of cooperation, including on the immigrant-specific models (Models 4–6). While the relationship is in the hypothesized direction, it is only significant in Model 3. Also surprising is that most of the measures of organizational support for facilitating coproduction are statistically insignificant. Parent drop-in centers are associated with better assessments of parental involvement, community support, and contact with parent organizations (Models 1–3), but not the immigrant-specific models. Interestingly, Latino teacher representation is positively and statistically significant for the general quality of parental involvement (Model 1). This suggests that, on average, schools with more Latino teachers have higher quality of overall parental involvement. However, it is negative and statistically significant for principals’ frequency of contact with parent organizations. This could indicate that Latino teachers act as a buffer between principals and parents, thus reducing the need for more principal–parent interaction.
Table 3 presents interactive models that include an interaction between the percent of undocumented students and passive representation (Latino teacher representation). As discussed above, the expectation here is that passive representation will enhance symbolic representation among immigrant communities and manifest itself via increased assessments of behaviors that are conducive to cooperation. The implication, then, is that schools without representation will witness less coproducing behavior among Latino and immigrant groups, including undocumented immigrants.
Interactive Models.
Note. Robust standard errors in parentheses, clustered by district. LEP = Limited English Proficiency.
p < .1. **p < .05. ***p < .01.
The models in Table 3 test this proposition. Here, the relationship between undocumented students and assessments of cooperation is modeled as conditional on the level of passive representation. The interactive term in the models suggests that this is the case for four of the six models (Models 3–6). In three of these models, the size of the undocumented student body is negatively related to the assessment of cooperative behaviors (supporting H1) when passive representation is absent; however, this effect diminishes as passive representation increases (as noted by the positive interaction term). Importantly, this interactive relationship is present and statistically significant in all of the immigrant-specific models (Models 4–6).
This can be seen clearest in Figure 4, which presents the predicted probabilities from the interaction for immigrant parental involvement, perhaps the most direct measure of cooperative behavior for immigrants. Since the variable is reversed, the figure represents the probability a principal will not identify lack of parental involvement as a problem for immigrant students. Here, we see that the predicted probability of adequate parental involvement is about 25% (between 18% and 40%) in a school with a high undocumented immigrant population with no passive representation (compared with about 50% in a school with no undocumented immigrants). However, as passive representation increases, the probability in a high immigrant school significantly increases and becomes statistically indistinguishable from a school with no undocumented immigrants once passive representation is about 18%. At high levels of passive representation (above 80%), schools with many undocumented immigrants are actually more likely to earn a favorable assessment of parental involvement from the principal.
Effect of undocumented students on quality of parental involvement conditional on Latino teacher representation.
These findings suggest that passive representation can translate into symbolic representation for immigrants in an education setting. When passive representation is absent or low, problems associated with low parental engagement and involvement are more likely to occur in schools with large undocumented student population. Theoretically, this is because students and parents in the undocumented community feel unconnected and unrepresented in schools without passive representation. This leads them to withdraw from activities related to educational coproduction. However, it appears that as passive representation increases, these cooperative behaviors—or at least principals’ assessments of them—are also more likely to occur. Arguably, this is because trust, efficacy, and perceived legitimacy of educational institutions may increase among immigrants leading to increased cooperation, compliance, and potentially coproduction.
Conclusion
This research examines how passive representation may increase symbolic representation and cooperation among undocumented immigrants in Texas public schools. It proposes that assessments of cooperative behaviors of undocumented immigrants will be lower when the schools they attend have lower levels of passive representation. It argues that the causal mechanism is symbolic representation, which will enhance cooperation and trust resulting in behavioral changes in immigrant parents and students. As a result, cooperation and compliance should improve as passive representation increases. Using data from Texas public schools, it finds that the relationship between the presence of undocumented students and assessments of educational cooperative behavior is conditional on the level of passive representation. More specifically, principals in schools with many undocumented students are less likely to give favorable assessments to immigrants than in schools with smaller immigrant student bodies. However, this relationship is contingent on the level of passive representation. In schools with adequate Latino teacher representation, principals’ assessments of immigrants are predicted to be positive in immigrant-heavy schools. This is particularly true for assessments of parental involvement of immigrants.
There are several caveats to this research, some of which have already been noted. The measure of “undocumented” students has limitations because it may include those who are not actually undocumented students but simply did not provide proper documentation when registering. That said, it is the currently the best large-N measure available and should, in theory, be highly correlated with the true undocumented figures. In addition, due to data limitations, the research design is cross-sectional. Future research would ideally examine longitudinal data to explore how representation and coproduction change over time. The data are also aggregate (campus level) rather than individual level. This does not allow us to observe individual-level behaviors that are linked to coproduction. However, the vast majority of research on representative bureaucracy has employed aggregate-level data (often at higher levels of aggregation than used here). The theory of representative bureaucracy is both an individual-level theory and an organizational one. That is, the theory attempts to explain individual behavior and decisions of bureaucrats that are connected to their representational roles. However, it is also a theory of organizations in that the aggregate composition of an organization has macro-level impacts on outcomes. Thus, the use of aggregate data is less problematic.
Perhaps more significantly, the measures of cooperative behavior are not direct measures of coproduction. Rather, they are based on principals’ perceptions of the levels of parental involvement rather than measures of actual coproduction. The theoretical argument presented has implications for the effect of passive representation on coproduction and, ideally, we would have measures that directly capture immigrant behaviors that are related to coproduction (e.g., membership in parent organizations, volunteer activities, student participation). Unfortunately, these data are not available for all school campuses. Rather, this analysis examines precursors of coproduction, namely, assessments of behaviors that are linked to cooperation and compliance. It finds support for the hypothesis that passive representation can lead to greater cooperation. Future research should extend the analysis to explore how this may increase coproduction.
In addition, we cannot be certain whether the principals’ assessments accurately reflect citizen behavior. However, given the size of the sample (1,000+ schools), it is reasonable that, on average, schools with higher levels of these actual behaviors should have higher assessments from principals than those with lower levels. Furthermore, even if these assessments reflect nothing more than the principals’ attitudes, this is still an important finding. Research suggests that passive representation can not only promote active and symbolic representation, but it can change organizational culture, socialization, and the behavior of other bureaucrats. Thus, even if it were the case that passive representation only changes principals’ perception of immigrants, this is still an important finding. Future research should examine this more directly to disentangle to what extent symbolic representation affects citizens’ versus fellow bureaucrats’ behaviors.
This research adds to our empirical and theoretical understanding of the theory of representative bureaucracy in several important ways. First, it presents a new theoretical model for understanding symbolic representation. It argues that symbolic representation can take multiple forms. Active symbolic representation entails direct actions taken by bureaucrats (akin to active representation) that encourage behavioral changes in citizens, including those that are conducive to cooperation. Passive symbolic representation results from the mere presence of representation in the bureaucracy and does not require any action on the part of the bureaucrat but results in behavioral changes in citizens. While the article is unable to fully test these theoretical arguments, it does provide evidence that symbolic representation is at play via a moderating relationship. Future work should more fully explore the multiple paths symbolic representation can take in changing citizen behavior.
Finally, this research also presents new empirical evidence from an understudied population—undocumented immigrants. There have been few empirical, large-N studies on this population, in large part due to data limitations (see Hill & Hawes, 2011). Immigrants, particularly undocumented immigrants, are often overlooked in political and bureaucratic institutions because they possess little political power. However, their success is critical to the nation’s economic future because immigration policy is eventually inextricably tied to labor and economic policy. Thus, educational attainment and success of immigrants, including undocumented students, is crucial for our future.
Footnotes
Author Note
The author received a small research award from my university to purchase the data used in the analysis.
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.
