Defining the 'Community' in Community College: A National Overview and Implications for Racial Imbalance in Texas

At least 38 states have created service areas or "districts" for each of their community colleges. However, little is known about the geographic boundaries of community college districts and the policymaking process that defines them. We studied state policy documents nationally and the actual district boundaries of Texas community colleges to investigate the larger policymaking processes of determining boundaries. We found significant variation across the United States, including in who determines the boundaries and whether the districts have associated tuition reductions. In our case study, we also found evidence that the majority of Texas's community college districts appear to reflect their larger local environments, although a small number may exhibit evidence of racial imbalance.

Keywords: district boundaries; enrollment; state policy; two-year institutions

Despite giving extensive attention to racial segregation within K–12 education (see [33] for a review), scholars have rarely examined segregation in postsecondary education ([6]). This is partially due to the fact that most theories of how students choose where to enroll in postsecondary education erroneously assume that students will choose from the entire population of institutions across the United States (e.g., [20]; [23]). This assumption ignores the ways geography plays a role in the college selection process, especially for the one sector of postsecondary education that has significant incentives for students to enroll at their "neighborhood" institution: community colleges.

Based on our investigation of state education codes and policies and our direct communication with government officials, we have learned that 38 states have created service areas or "districts" for each of their community colleges for academic year 2020–2021. These states hold 90% of the residents in the United States,[3] yet little is known about the geographic boundaries of community college districts, the policymaking process that defines them, and how they relate to institutional racial segregation.

This lack of evidence is important for many reasons. For example, economic resources matter for students and institutions (e.g., [4]; [28]; [31]). Still, the pricing and level of funding for community colleges can vary widely, even within states (e.g., [2]; [8]; [15]; [26]). This variation in pricing and funding has direct and indirect relationships with students' enrollment and academic outcomes, such as enrollment intensity, retention, and graduation (e.g., [2]; [4]; [15]). These funding decisions can be made via a political process that incorporates community college districts, which may privilege certain institutions (thereby privileging certain students within a state).

Given this dearth of knowledge, we studied state policy documents nationally and the actual district boundaries of Texas community colleges to investigate the larger policymaking process that influences how boundaries are drawn. We were particularly interested in how frequently district boundaries were racially imbalanced, which we define as showing a racial disparity between a geographic area and its larger local environment.[4] Texas has a large, racially diverse population that is spread across several different metropolitan areas with a robust community college sector, making it a useful state to study. This research is the first step in a larger project that analyzes the relationship between racial imbalance in community college districts and segregation. In the current study, we investigated the following questions:

• 1) What is the policymaking process underlying the creation of community college districts across the United States?
• 2) Do the district boundaries of community colleges provide evidence of racial imbalance in Texas?

Due to the lack of research on community college districts, it is critical to include a national and a state-specific investigation. This article provides the first national list of states with community college districts. No national repository tracks which states have these types of policies or the policymaking process underlying the creation of the physical boundaries.[5] Further, no prior research, to our knowledge, has focused on the policymaking involved in the creation of these district boundaries or whether these boundaries show evidence of racial imbalance. Therefore, our research provides a critical overview of the national policymaking landscape regarding community college districts and a case study of a single state's district design.

The historical inattention to community college district boundaries by researchers, policymakers, and the public does not indicate their unimportance or lack of impact on students' lives. In fact, the large number of students who enroll in community colleges suggests that how students are zoned for community colleges has significant impact on the nation's current and future workforce. Therefore, the process of drawing community college district boundaries, as well as its potential effect on the racial segregation of community college students, deserves the same consideration as K–12 attendance zones. The current study's findings regarding the varied state processes for creating community college district boundaries and the Texas case study, which indicates a potential for racial imbalance in a small number of districts, further justify the continued study of community college boundaries.

We begin this article with an overview of our theoretical framework and then summarize the prior research on K–12 school attendance zones and how geography relates to community college access and success. Next, we detail our research methods, including the measures of racial imbalance that we calculated for the current study. We then present evidence on the national policy context surrounding community college districts, based on close to 150 policy documents, and an in-depth analysis of Texas's community college districts.

Typically, researchers have used three different theories to study attendance zones. Several scholars have used [47] theory of public choice to guide their work (e.g., [7]; [10]; [17]; [19]; [24]; [29]; [42]; [55]). In the context of the public educational system, this theory asserts that attendance zone boundaries hold the potential to segregate "because individuals choose where to live in part on the basis of their neighbors—often opting to live near people more similar to them in terms of race/ethnicity—as well as on the basis of the school that their child will attend" ([36], p. 1121). [39], [40]) draws upon [48] first law of geography, the second theory researchers often use, which [22] paraphrases as "nearby things are more similar than distant things." Applying this theory to school attendance zones, Saporito finds that racially and economically segregated school attendance zones are largely the outcome of residential segregation. According to [39], [40]), most school districts draw compact attendance zones (perhaps to minimize transportation costs), which then reproduce residential segregation. Because people of the same race and income tend to live closer together than do people of different races and income, often due to deliberate public policy choices (e.g., [38]), [48] first law could also reasonably be applied to the racial segregation of schools.

[47] and [48] help contextualize the segregation of public K–12 schools, but their research does not wholly apply to the question of community college segregation. For example, postsecondary institutions are typically not concerned with student transportation costs, as these generally fall directly on students. Additionally, Tiebot's premise that people make residential decisions based on the quality of the schools to which neighborhoods are zoned has significantly less relevance in a community college context; it is not a common occurrence for homebuyers to consider the local community college in their purchasing plans. Although they do not perfectly translate to a college-level context, Tiebot and Tobler draw attention to the significant relationship between residential and school segregation, which is present at all levels of education.

When scholars have been interested in examining school zoning and segregation as phenomena distinct from residential segregation (e.g., [36]), they have often used the student exchange framework, a theory adapted from the political science "voter exchange" perspective on electoral gerrymandering. [37], p. 6) highlight that "[zone] compactness constitutes the single most important principle of gerrymandering, particularly for schools." Compactness can be measured in many ways ([9]) but is often divided into two constructs: indentation (how smooth is the perimeter of the boundary?) and dispersion (how dense is the area within the boundary?). Based on the prior literature, the fewer indentations and less dispersion in a boundary, the more compactness and the less evidence for imbalance ([37]).

The student exchange framework posits that indentations and dispersion occur when policymakers deliberately create attendance zone boundaries that encompass one racial group and exclude students from other racial groups ([36]). Therefore, gerrymandering would result from this systematic "exchange" of more desirable students for less desirable ones. This framework does not necessarily imply that gerrymandered attendance zones exacerbate segregation. As [36] argues, school districts could create student exchanges to diversify attendance zones. The potential for exacerbation or reduction in racial segregation is one of the reasons systematic research is needed to better understand the relationship between educational boundaries and segregation.

The student exchange framework is applicable to community college districts, even though the current work is studying geographic racial imbalances instead of gerrymandering. Districts are generally tied to an institution's county or are geographically determined by policymakers. Following student exchange, these boundary determinations allow for the potential of racial imbalance across community college districts, either directly through the creation of the boundaries or by linking institutions' boundaries to other geographic boundaries known to exhibit racial imbalance. For these reasons, we use the student exchange framework to guide our current research. Because students who live within a district boundary are more likely to attend their "zoned" community college ([2]; [53]), evidence of racial imbalance could be related to institutions' racial enrollment. Indeed, anecdotal evidence suggests that some Texas community college taxing district boundaries are created to ensure that the institutions have more advantaged students enroll (e.g., [34]). However, no systematic research has been conducted to determine whether this occurrence is rare or prevalent throughout the state. And no research, to our knowledge, provides a national understanding of how these districts are created.

The majority of research focused on policymaking about education boundaries—typically, attendance zones—has investigated K–12 schools ([56]). To our knowledge, little commensurate research has been done regarding higher education. This lack of research is concerning, as scholars posit that higher education can have segregation issues similar to those of K–12 ([6]), and community colleges in several states create district boundaries through policymaking processes similar to attendance zones in the K–12 sector ([2]; [12]; [15]). These institutions also educate a significant share of the higher-education students in the United States: approximately 8.4 million students, or 33% of all postsecondary students in 2017–2018 ([51]). Further, in Texas, community colleges educate more students than does any other sector ([44]), with approximately 24% of students enrolled full-time and around 19% in dual-credit enrollment ([46]). Therefore, it has become increasingly important to examine student enrollment in this unique context ([11]). To motivate the current study, we provide an overview of prior research on the potential for racial imbalance in education attendance zones and review the evidence on community colleges and geography.

The majority of research on geographic boundaries relating to education focuses on K–12 school districts and attendance zones (e.g., [30]; [43]).[6] The vast preponderance (92%) of K–12 school districts are independent districts whose boundaries are generally not forced to solely align with political jurisdictions (e.g., cities, counties). The creation and revision of these boundaries are inherently political. For example, [43], p. 748) note that "it stands to reason where [public schools] are located and how boundaries are drawn may also be a crucial piece of how local governments, particularly school boards, are especially responsive to how White and/or affluent residents preserve their relative advantage." The goal of fiscal and administrative independence from local government created a large number of districts nationwide; however, districts began to consolidate in the mid-20th century, with the aim of administrative efficiency and instructional specialization ([50]). Consolidation did not occur evenly across communities. [3] identify an inverse relationship between the racial diversity of districts and their decisions to consolidate.

Scholars provide mixed evidence on the current prevalence of imbalanced attendance zones. [37], p. 16) analyze data from the School Attendance Boundary Information System for the 2009–2010 school year and find that attendance zones were, on average, "substantially different than would be expected in the absence of gerrymandering." Using the student exchange framework, these results support the theory that the exclusion of nearby students in lieu of other students residing farther away is a common occurrence. In contrast, when analyzing School Attendance Boundary Information System data from the identical school year, [41] find little evidence of imbalanced school attendance zones.

One explanation for the diverging research conclusions could be the differences in approaches used to measure the presence and severity of imbalanced zones. Both sets of researchers use their selected measures of imbalance to compare school attendance zones to U.S. congressional districts. [37] include attendance zones for all traditional public schools in regular districts in their analyses, resulting in a sample of 23,945 public school attendance zones in 1,721 school districts. In contrast, [41] limit their research to districts that have first-grade attendance zones and are among the largest 350 districts in the country, yielding a sample of 13,169 attendance zones in 307 school districts. In each study, the evidence indicates that school attendance zones were less imbalanced than congressional districts, with larger differences in indentation than in dispersion. However, the difference in imbalance between attendance zones and congressional districts calculated by [41] is larger than the difference found by [37]. This may be why [37] argue that the compactness of the attendance zones raised concerns for gerrymandering. Part of the reason it is difficult to find agreement on whether attendance zones are imbalanced is that no accepted threshold indicates that a geographic area is or is not imbalanced, as we discuss further in the Methods section.

Therefore, it is not clear whether K–12 attendance zones are imbalanced. The imbalance of K–12 attendance zones matters, as this could be one mechanism through which between-school segregation is produced. Due to this lack of clarity at the K–12 level, our current research incorporates several different methods of assessing the evidence indicating that a district may be racially imbalanced. Based on prior research, it seems clear that design and sample selection can play a significant role in the interpretation of the evidence indicating imbalance.

As noted previously, community colleges can vary widely in their tuition costs, state and local funding, enrollment demographics, course offerings, and more (e.g., [4]). As one example of this variation, we present the average in-district tuition for full-time undergraduates at community colleges across the United States in Figure 1.[7] We show in-district rates because some states allow districts to provide an additional reduction for students who reside within their district beyond the in-state rate. The figure presents the average (dot) and the range (vertical bar) of the average in-district tuition. There is a difference of a little over $5,500 across the states (from California's $1,150 to Vermont's $6,720). The figure also shows wide variation in average tuition within states as well. For example, Tennessee reports a community college average tuition range of $210, while Pennsylvania has a within-state range of $7,816. Eight states report no within-state variation in average tuition (Connecticut, Delaware, Hawai'i, Kentucky, Maine, New Hampshire, Rhode Island, and Vermont).

Graph: Figure 1. Variation within and across states in average in-district tuition. Note. Dot represents the average in-district tuition at community colleges within each state. Vertical bar represents the range in average in-district tuition across community colleges within each state. States are ordered by their average in-district tuition. These estimates are based upon the authors' calculations of Integrated Postsecondary Education Data System (IPEDS) average in-district tuition (sticker price) for full-time undergraduates for 2020–2021. Because Alaska does not have community colleges, it is not included in the figure.

Geography matters when students decide whether and where to enroll in higher education (e.g., [13]; [16]; [21]; [23]; [49]). Scholars have found that students are more likely to attend institutions near where they live ([23]), can benefit from enrolling at in-state institutions that are farther from their home ([21]), and can live within the same county but have vastly different access to higher-education opportunities ([13]). Typically, prior research has not focused specifically on the community college sector. [35] explore a Houston community college system's institutional administrative data and find that students who graduated from high schools near college campuses were more likely to attend that campus. The authors also find that the local labor market context strongly related to students' enrollment patterns. Similarly, [1] finds that the local labor market directly affected the majors Michigan community college students chose. Districts do not solely affect a community college's price. As we show in the national results, districts are directly tied to the curriculum and program offerings in several states' community college systems. Because the local labor market can spur students to select different majors, and districts play a significant role in determining which majors are offered, living within a district may mean a tighter coupling between labor market needs and educational opportunities.

Therefore, while understudied, scholars have used geography and space to explore access and success within higher education. Scholars have less frequently studied the boundaries that drive access to higher education, such as community college districts. To our knowledge, only a small handful of studies engage with studying community college districts (e.g., [2]; [12]; [15]; [53]; [54]). Typically, these studies focus on states with a type of "local taxing district" that provides additional funding to the community college and allows residents to receive a reduction in tuition.[8] The benefits produced from living within taxing districts vary substantially across the United States. Figure 2 explores the differences between the in-district and in-state average tuition institutions reported to Integrated Postsecondary Education Data System (IPEDS).[9] The figure shows that across the 19 states that reported a difference, the range is approximately $5,000 (from California, with less than a dollar, to Illinois, with $4,970). Within states, the benefit for in-district students ranges from $0 to $9,230 (Illinois).

Graph: Figure 2. Difference in the average in-district and in-state tuition across and within states. Note. Dot represents the state average of each institution's average in-state tuition minus the average in-district tuition. Vertical bar represents the range in the average in-state tuition minus the in-district tuition across community colleges within each state. States are ordered by their average difference. These estimates are based upon the authors' calculations of Integrated Postsecondary Education Data System (IPEDS) average in-district and in-state tuition (sticker price) for full-time undergraduates for 2020–2021. Only states with at least one institution reporting a difference are included in the figure.

Prior research generally focuses on the relationship between districts and enrollment in higher education. [15] and [2] analyze Texas's and Michigan's taxing districts, respectively, and find that reductions in price at the local community college increased residents' enrollment at that institution and decreased their enrollment at other institutions. Similarly, [26] working paper identifies a negative correlation between tuition price and enrollment in Texas community colleges, with household income having a mediating effect. In 2020, the Center for American Progress published a policy report that investigates the relationship between Michigan's districts and racial segregation in enrollment at local community colleges ([12]). The policy report compares the share of White and Black adults in a district, respectively, to the share of White and Black students enrolled in the district's community college. [12] finds a significant underrepresentation of White students enrolled in community colleges across the state and an overrepresentation of Black students enrolled in community colleges near Detroit. Although this policy report does not use conventional measures of segregation, clear evidence shows that community college districts may play a role in racially segregating students enrolling in community colleges.

We build on this prior research focused on community college district boundaries by investigating the districts as actual artifacts of a political process. We explore the variation in policymaking structures across states with community college districts. We also dive deeply into a case study of Texas's community college districts, with statistical analysis of measures of imbalance. This type of research is essential for creating a nationwide understanding of the policymaking processes that shape community college districts and for building the evidence base on the actual design of the districts and their boundaries.

For the current study, we collected a novel set of policy documents related to community college districts across the United States (research question one) and used a geospatial analysis case study to investigate whether the boundaries exhibit evidence of racial imbalance in Texas (research question two).

For research question one, the descriptive analysis of community college districts across the entire country, we investigated all states' constitutions, education codes, and additional policy documents for evidence of community college district areas, as well as details about how those areas are defined. First, we used 2018 data from the U.S. Department of Education's IPEDS and compared the in-district tuition to the in-state tuition at public 2-year institutions in each state. Any state that had a different amount for in-district tuition was one we considered to potentially have community college district boundaries. This led to a pilot group of 16 states.[10]

One of the authors collected data on each of the 16 states to assess (a) whether they had community college districts and (b) whether their state education codes had a policy regarding these districts. To make these assessments, the research team member went to each state legislature's website and accessed its statutes.[11] Going to the education section in each set of statutes, the team member then found every state's chapter on community colleges or postsecondary education and began searching within the chapter for any information regarding community college districts. The team member searched for keywords within these sections, such as service areas, service districts, college districts, counties, and college boundaries. If the team member could not find any information specific to college districts in the education code from the state websites, she then conducted a Google search, using the state's name with the same aforementioned keywords. She categorized any states for which she could not find community college district-related information as not having a policy regarding college districts. Once the first 16 states were initially completed, all authors reviewed the documentation and provided additional documents for the team member to review in an attempt to find policy details. Finally, once the team member conducted the final search, the data collection for research question one expanded to the other 34 states (replicating the steps outlined in this paragraph). This led to a total of 37 potential states with community college districts.

Two of the authors, one of whom did the original search, then independently coded each state by answering the questions outlined in Appendix Table A1. The questions broadly focused on the presence of a district, which policy actors had a role in determining the boundaries, and whether institutions provided a tuition discount for students within the district. The two team members met frequently to resolve any differences between their independent coding. Coding was not considered complete until the two team members reached consensus. We contacted state higher education governing bodies with any questions relating to the items in Table A1 that could not be addressed by publicly available documents. We also contacted all the states for which we were unable to find any information about districts to verify that no such policy boundaries existed. These additional explorations led to a final total of 38 states with a community college district policy. We collected approximately 150 documents for the 38 states with some type of community college district policy.

For research question two, we collected data on the 50 district boundaries for the 82 community colleges in Texas in 2017–2018. Texas community college districts are outlined in Texas Education Code (chapter 130, generally §162 to §211), and information about their boundaries is publicly available. For example, Texas Education Code §130.162 outlines the district for Alamo Community College as the following:

• (1) Bexar, Bandera, Comal, Kendall, Kerr, and Wilson counties;
• (2) Atascosa County, except the territory within the Pleasanton Independent School District; and
• (3) Guadalupe County, except the territory within the San Marcos Consolidated Independent School District ([45].).

We converted the text descriptions of the 50 boundaries into electronic maps that could be used within geographic information systems analysis software using ArcGIS Pro software (2.5.2).

Texas community college districts are defined according to the intersections between city, county, taxing district, and independent school district (ISD) borders. We obtained Texas city shapefiles from the U.S. Census Bureau and included the city limits of the Colony, Corpus Christi, Frisco, Missouri City, Paris, Sugar Land, and Temple.[12] We obtained county shapefiles from the National Historical Geographic Information System website ([25]).[13] The Texas Legislative Council, which creates the official maps for the state of Texas, provided the ISD shapefiles. Taxing district boundaries affected only two community college district boundaries: Borger and Texarkana. The taxing district for the Borger Junior College District Service Area includes the portion of Spring Creek ISD that is also within the community college district's service area. Because Spring Creek ISD was located within Hutchinson County during 2017, the taxing district boundary was irrelevant. Likewise, the Texarkana College District Service Area includes a taxing district that encompasses all of Bowie County. Therefore, the county shapefile was used to identify the boundary for the Texarkana District.

We created the community college spatial boundaries by using polygons in ArcGIS Pro. We compared the boundaries of the polygons to the appropriate city, county, or ISD boundary, depending on the district. We accomplished this by using the "Edit Vertices" tool after enabling the map topology (which ensures that feature contiguity is maintained while editing polygons). Because community colleges adhere to the academic year (e.g., fall to summer), we compared the community college boundaries to city and county shapefiles from the fall in a given academic year. In other words, we created the 2017–2018 community college boundaries based on the shapefiles from 2017. This created a spatial data set of all Texas community colleges and their corresponding district spatial coordinates. For the final map, we used the projected coordinate system North American Albers Equal Area Conic. We shared these maps while they were in progress and, once completed, with experts in educational geospatial data and members of the Texas Legislative Council, who noted that our maps appeared to be accurate (although this is not a legally binding assessment).

Once we finalized the district maps, we merged the district shapefiles with American Community Survey (ACS) census block group data, the smallest geographical unit for survey data, on the race of residents from the National Historical Geographic Information System ([25]).[14] We used 5-year estimates, merging based on the final year from ACS to the fall of the academic year for the district (ACS 5-year estimates for 2013–2017 merged to 2017–2018 academic year district areas). When census block groups were located in multiple districts, we created a spatial weight for the area of the block group in each of the districts and apportioned residents based on that amount.[15] We applied that weight so that residents were spatially apportioned within districts.

For research question one, we conducted a descriptive analysis of the national data on community college districts with a focus on exploring who plays a role in this policymaking process. It is critical to understand the broader policy context of how these boundaries are created as political objects to understand the potential for them to exhibit signs of racial imbalance.

For research question two, based on the theoretical framework of student exchange, we assessed the racial dispersion of Texas residents within district boundaries to examine the potential for racial imbalance. Frequently, scholars use measures of indentation and dispersion of geographic boundaries to study gerrymandering, such as the Polsby-Popper, Reock, and Convex Hull indices (e.g., [37]). These measures typically create some "ideal" shape for a district, such as a circle for the Reock index, and then compare that ideal to the actual district. These indices have a number of limitations, such as when the "ideal" districts go beyond the state's borders (e.g., an "ideal" circular district near the Gulf of Mexico likely includes area that is solely water in which individuals cannot live). Also, there is no commonly agreed-upon threshold for when any of these indices indicate that a district is "officially gerrymandered." This challenge is compounded by the fact that legal assessments of district gerrymandering require some evidence of intent, which cannot be obtained solely through quantitative analysis. Finally, it can be more difficult to focus on the racial demographics of ideal districts when solely focusing on the shape of the geographic boundaries. For these reasons, it can be challenging to draw conclusions based on these more traditional measures of gerrymandering.

Due to our interest in focusing on the racial demographics of districts and concerns with those limitations, we therefore used a demographic approach to exploring racial imbalance by comparing the racial composition of a district to the racial composition of the region in which a district is located. This comparison builds upon prior research that relies upon the concept of "local environments" (see [32] and the appendix of [40] for more details). A local environment is the area surrounding each person. We crafted a local environment for each person in a district that consisted of their nearest N residents (where N equals the number of people in the district). Because we did not have access to the actual residential locations of individuals, we assumed that people live in the middle of their census block group.

For example, suppose that District X contains 300,000 residents who live in 150 block groups. For each block group in District X, we determined their nearest 300,000 residents. Once the nearest 300,000 residents were identified for all block groups, referred to as the local environment, we calculated the percentage of people in a district's region who are members of a particular racial group by using the following equation (with Latinx as an example):

$$\mathrm{Latinx}\mathrm{LE}(D_i)=\sum _{b=1}^n\frac{\mathrm{Number}\mathrm{of}\mathrm{Latinx}\mathrm{residents}\mathrm{in}L{E}_b}{\mathrm{Number}\mathrm{of}\mathrm{total}\mathrm{residents}\mathrm{in}L{E}_b}*\frac{\mathrm{Number}\mathrm{of}\mathrm{residents}\mathrm{in}\mathrm{block}\mathrm{grou}{p}_b}{\mathrm{Number}\mathrm{of}\mathrm{total}\mathrm{residents}\mathrm{in}\mathrm{distric}{t}_i}*100$$

Graph

We divided the total number of Latinx residents in each local environment of block group b by the total number of residents in the local environment of block group b and weighted that share by the proportion of residents in community college district i who live in the block group. In the above example, this denominator is 300,000. After weighting, we multiplied that ratio by 100, creating a percentage. We then summed all the weighted percentages of Latinx residents for each block group within community college district i. Continuing the prior example, this would involve summing the weighted percentage of Latinx residents for all 150 block groups. This created a district-level measure of the average percentage of Latinx people across all local environments. This result represents the racial composition of the region in which a district is located.

We then subtracted the percentage of residents in a district's region who belong to a given racial group (local environment) from the actual percentage of people in a community college district who are members of that racial group. For example, if the percentage of people in a district who are Latinx is 70, and the percentage of people in the district's region who are Latinx is 70, this difference of 0 suggests that the district generally reflects the racial composition of its local environment. This analysis method allowed us to explore how many districts in Texas had racial compositions different from that of their surrounding region. We calculated percentage-point differences for White, Black, Latinx, Asian, Hawaiian and other Pacific Islander, Native American/Alaska Native, two or more races, and other race residents. There is not a definitive threshold for when this type of analysis indicates that a district is "imbalanced."[16] The typical guidance is to explore any district with at least 5 percentage points of difference between the real district and the local environment. To ensure that we systematically investigated the differences between real districts and local environments, we iteratively explored all differences up to 10 percentage points (which is a fairly large difference between the real districts and local environments).

These two different levels of analysis for research questions one and two allowed us to investigate the policymaking process that drives district boundary creation across the United States and whether racial imbalance is evident in the creation of Texas community colleges' district boundaries. Throughout the entire study, the research team has consulted with expert peer debriefers, such as the Texas Association of Community Colleges (an intermediary organization that represents all Texas community colleges) and the Texas Legislative Council (a nonpartisan legislative agency that provides research and all official state maps to the Texas state legislature) to allow the team to create a deep understanding of the policymaking processes behind district formation in Texas, to ensure that maps were being created appropriately, and to incorporate field standards for design and sample decisions.

Our analysis of state constitutions, education codes, and policy documents, as well as direct communications with government and higher education officials, finds that 38 states have community college districts that dictate a specific area of the state that the community college should focus on serving. Table 1 includes a full list of all 50 states and whether each state had such a policy in academic year 2020–2021. For states with a policy, Table 1 also presents whether that state had a tuition reduction policy, whether the legislature or residents played a role in the boundary decision-making process, and the total number of districts.[17] No clear regional pattern emerged corresponding to which states have a policy. It does appear that states with smaller populations (e.g., Rhode Island) or with greater geographic dispersion (e.g., Hawai'i) were less likely to have a policy. Due in part to some states having multiple geographic regions (e.g., local taxing districts and service areas), the terms for these areas varied widely. Fifteen states used the term district, while sixteen used the term service area or service region.[18] Some states used a combination of these terms. Single states used such phrases as geographic areas of responsibility (New Mexico), merged areas (Iowa), and sponsor areas (Pennsylvania). The average number of community college districts in the 37 states with data was approximately 20 (ranging from 2 to 73). We also explored what underlying boundaries states used to create these districts. The majority of states created these districts by using a combination of county (76%), K–12 school (39%), and city (26%) boundaries.[19] A small share of states also used local streets, highways/interstates, and waterways to define boundaries for the districts.

Graph

Table 1 List of All States With Policy Status

1 Note. Data represent information collected on each state's community colleges. When a state has a community college and a technical college system, data represent only the general community college system. The first four columns include indicators of whether the respective state has evidence of a community college district, a reduction in tuition tied to the district or another geographic boundary, a role for the legislature in determining boundaries, and a role for residents in determining boundaries. For these columns, 1 means that we have found evidence for that column, and 0 means that we were not able to find evidence. CC = community college.

• 2 After several conversations with New York, we were unable to ascertain any additional details about their district policy beyond the fact that one exists.
• 3 Alabama historically had community college districts. In the 2015 state reorganization, a Board of Trustees was created that is now responsible for all community colleges and is currently reviewing the district policy that was adopted by the former governing body (as of July 2021).
• 4 Connecticut is currently pursuing a merger of its community colleges, which will likely make its districts obsolete (although that had not been determined as of July 2021).
• 5 New York has two different community college systems (the State University of New York System and the City University of New York). As both systems include community college districts for residents, we consider this state to have a district policy.
• 6 These states have a tuition reduction, but it does not apply to all residents located within the districts. These states have smaller geographic areas within their districts that frequently provide local taxes to receive the tuition reduction. These smaller geographic areas are not necessarily present in every district.
• 7 New Jersey's legislature created the community colleges to be county-based institutions. Therefore, if any institution wished to change its district, the boundaries of the county would have to be shifted, which would involve the legislature.
• 8 In Utah, the College of Eastern Utah was a community college that eventually became part of Utah State University. Given that residents of certain parts of Utah are still able to enroll at Utah State as if it was a community college, due to this historic relationship, we count Utah as having two districts (the other district being for Salt Lake Community College).

We found similar language detailing the districts and their purposes across the different states. In Arkansas, the legislative text states, "'Community college' means an institution of higher education established or to be established under the provisions of this chapter dedicated primarily to the educational needs of the service area and offering a comprehensive program, including, but without limitation, vocational, trade, and technical specialty courses and programs, college transfer courses, and courses in general adult education" (Arkansas Code 6-53-103). Colorado's Revised Statutes note, "The mission of the community colleges shall be to serve Colorado residents who reside in their service areas by offering a broad range of general, personal, career, and technical education programs" (Colorado Revised Statutes 23-60-201). In Texas's education code, community colleges are defined as "two-year institutions primarily serving their local taxing districts and service areas in Texas and offering vocational, technical, and academic courses for certification or associate degrees" (Texas Education Code 130.011).

The Texas legislative language also provides an example of another trend. Of the states with districts, at least six had two different types of geographic areas that community colleges were tasked with attending. When states had multiple areas, these geographic zones were typically a smaller taxing district and a larger service region. The institutions were tasked with crafting courses and majors based on the larger service regions, while the taxing area drove additional funding for the institutions and provided taxing-area residents with reduced tuition rates. When investigating the states that only had one geographic area, we found that 11 additional states also had a tuition reduction based on residency within the district (which means that 17 states total had some type of tuition reduction associated with residency within a region of the state). When multiple geographic zones were associated with community colleges, we focused our review on the larger service areas. We made this decision for several reasons, including the fact that these types of geographic areas were bigger, encompassed the smaller taxing districts generally, and dictated which residents institutions were supposed to target with their policies and curricula.

The policy actors who played a role in the formation of these districts varied across states. Of all states with districts, except New York, approximately 30% involved the state legislature, 24% involved residents, and 82% involved at least one other party (these numbers are not mutually exclusive). The residents were registered voters, often referred to as qualified electors, in the current district and proposed addition to the district. These residents would frequently need to petition a county or state board to incorporate a new territory and vote in a subsequent election. The other party was typically a governing or coordinating board for education, higher education, or community colleges. For example, Arkansas included the Arkansas Higher Education Coordinating Board, Illinois included the Illinois Community College Board, and New Mexico included the New Mexico Higher Education Department. Presidents of individual community colleges would sometimes be involved in states, including Missouri and West Virginia. We could not find evidence for a formal role for an education board for any level of education or community college presidents in at least eight states (Arizona, Florida, Iowa, Mississippi, Nebraska, New Jersey, Texas, and Washington). Several states also included the county commissioners responsible for arranging a local election or school district superintendents when the community college districts were based in large part on K–12 school districts.

Texas is a useful state for a case study for the second question for three reasons. First, the community college sector in Texas is robust and has significant racial variation. Texas has 82 community colleges that educate more than 700,000 students, 70% of whom are students of color ([46]). Second, because Texas public community colleges have tuition and fees below the national average ([46]), it is likely that a substantial share of prospective students consider community colleges to be part of their college search and enrollment process. Third, Texas has publicly available narrative information on each district's boundaries. As noted in the Methods section, Texas enshrined community colleges' districts into the Texas Education Code and lists each geographic area in text. A case study of Texas's community college district boundaries provides evidence that can be used to create a national study or similar case study of another state, while providing grounds for the creation of better state and local policies in and outside Texas.

In 1995, Senate Bill 397 created service area districts in Texas.[20] The state has 82 community colleges serving 50 districts. The legislature has ultimate control over the construction of the districts. Students may attend any community college in the state of Texas, but students who wish to receive a tuition reduction may only attend certain institutions. The district boundaries have been changed 18 times, with the last changes adopted in 2015. Figure 3 shows the districts in our most recent year of data, 2017–2018, with shading highlighting how many residents are served (from yellow to red, where districts that are redder serve more residents). It is clear that districts in the major metropolitan areas (Dallas-Fort Worth, Austin-San Antonio, Houston, and El Paso) serve the largest number of residents. To investigate this further, we turn to our explorations of the racial demographics of Texas districts and community colleges.

Graph: Figure 3. Texas community college districts, shaded by number of residents served. Note. District shading goes from yellow (fewer residents served) to dark red (more residents served). Areas with no shading within Texas are parts of the state that are not located within a district. Green triangles represent the major metropolitan areas.

Figure 4 shows the racial demographics for Texas community college students.[21] Each vertical bar represents a different community college district. Within each vertical bar, the different sections represent the enrollment share of the corresponding racial group (all other races includes Hawaiian and other Pacific Islander, American Indian and Alaskan Native, two or more races, and other race). The districts are ordered by their share of Latinx students. This means that districts on the left side of the figure had a smaller share of Latinx students enrolled compared to districts on the right side of the figure. It is clear from the figure that districts across Texas enroll varying shares of all racial groups.

Graph: Figure 4. Demographics of Texas community college students. Note. Each vertical bar represents a different Texas community college district. Within each vertical bar, the different sections represent the enrollment share of the corresponding racial group (all other races includes Hawaiian and Pacific Islander, American Indian and Alaskan Native, two or more races, and other race). The districts are ordered by their share of Latinx students. These estimates are based upon the authors' calculations of Integrated Postsecondary Education Data System (IPEDS) fall 2017 enrollment data for all undergraduates.

To investigate whether that variation relates to the Texas districts, we include Figure 5 and Appendix Figure A1, which present the relationship between each racial group's district share and respective community college share (Figure 5 includes White, Latinx, Black, and Asian shares, while Figure A1 includes Hawaiian and other Pacific Islander, American Indian and Alaskan Native, two or more races, and other race). The relationship between White and Latinx shares is incredibly high, with correlations of 0.97 and 0.96, respectively. The figures show a fairly strong relationship between district and community college shares of Black, Asian, American Indian/Alaskan Native, and Hawaiian and other Pacific Islander individuals (correlations are 0.87, 0.78, 0.73, and 0.54, respectively). There is little correlation between the two or more race and other race shares. These relationships may be due to how institutions and the government treat the two or more races category ([52]) and the incentives driving who reports the "other race" category, especially when enrolling in higher education ([18]). Therefore, we generally find strong relationships between the racial demographics of a district and their respective community colleges.

Graph: Figure 5. Relationship between Texas district and community college racial demographics. Note. Each panel represents a different scatter plot of the respective racial group's district share and respective community college share. The line represents exact parity, where the dots would be if the share of residents and students enrolling at their respective community colleges were the same. These estimates are based upon the authors' calculations of Integrated Postsecondary Education Data System (IPEDS) fall 2017 enrollment data for all undergraduates linked to 5-year U.S. Census block group estimates for 2013–2017 of Texas community college districts.

When comparing the racial composition of a district to that of the region in which the district is located, we found that the overwhelming majority of districts had small or no difference in racial composition compared to their local environment. Figure 6 shows a histogram of the differences for White, Latinx, Black, and Asian residents. Interpreting the top-left graph as an example, the x-axis represents the difference between the percentage of residents in the district who were White and the percentage of residents in a district's region who were White. Negative values for this difference mean that the district percentage was smaller than the local environment percentage. The y-axis shows the number of districts with the respective values.

Graph: Figure 6. Histogram of the frequency of differences in district residents and local environment residents. Note. Each panel shows the difference between the share of each racial group in the Texas community college district and the weighted share of each racial group in the local environment of the district. They were estimated separately for White, Latinx, Black, and Asian residents. Negative values mean that the district percentage was smaller than the local environment percentage (and the opposite for positive values).

Staying with the first panel, we found a slightly positively skewed distribution (0.781), where the majority of districts had small differences in the White resident share within districts and their region.[22] Turning to the next panel, we found that Latinx residents had a normal distribution, with the majority of districts having a small difference between real districts and local environments. However, when examining Black and Asian residents, we found a slightly different story. Although it is still true that the majority of districts had small differences with their local environments for Black and Asian residents, these distributions were negatively skewed (−2.408 and −4.389, respectively). A number of districts had fairly sizeable negative differences, which means that the districts' share of Black or Asian residents was smaller than that of the larger local environment, with no districts that had similarly sized positive differences. All other racial groups' differences were extremely small, with the overwhelming majority of districts having little to no difference between the real districts and local environments (see Appendix Figure A2 for the histogram for Hawaiian and other Pacific Islander, Native American/Alaskan Native, two or more races, or other race).

Because there was no clear threshold for determining that certain districts are racially imbalanced, we explored multiple minimum thresholds for "extreme" differences between districts and their local environments. Figure 7 presents the bar chart for White, Latinx, Black, and Asian residents, showing the number of districts with extreme differences based on varying thresholds. (None of the differences for the other racial groups was larger than 1 percentage point, so they were not included in this analysis.) We will use the first panel to explain the figure in more detail: The x-axis represents the threshold for consideration as an extreme difference. That means that the first bar represents the number of districts that had more than 1 percentage point difference in either direction. The purple portion of the bar is the number of districts with a negative difference (district has a smaller share than local environment), and the blue portion of the bar is for a positive difference (district has a larger share than local environment). Therefore, in the top-left panel, we show that 35 districts had a larger than 1 percentage point difference in either direction, with 16 districts in the negative direction and 19 districts in the positive. We also show that, as the thresholds increase in magnitude, the overall number of districts with extreme differences decreases, although there is a consistent trend that the majority of extreme differences are in the positive direction.

Graph: Figure 7. Number of Texas districts with extreme differences, based on varying thresholds. Note. Each panel represents 10 different comparisons of the real districts to the local environments for each racial group, with a threshold for an "extreme" measure varying based on the number on the x -axis. Therefore, bars at 1 on the x -axis show the number of districts with differences larger than 1 in either the positive or negative direction, while bars at the 10 show the number of districts with extreme values larger than 10 percentage points. The purple portion of the bar shows the number of districts with differences in the negative direction (district share is smaller than local environment share). The blue portion of the bar shows the number of districts with differences in the positive direction (district share is larger than local environment share).

Similarly, Figure 7 shows that the number of districts with extreme differences in the share of Latinx residents between the real district and local environment decreases as the thresholds become larger, although the majority of districts at each threshold size are positive (i.e., the district had a larger share than did the local environment). We also note that once the threshold size reaches eight, no districts had that extreme of a difference for Latinx residents. Turning to the bottom two panels, there was again a different pattern for Black and Asian residents. The majority of districts with extreme differences were overwhelmingly in the negative direction (i.e., the district had a smaller share of Black or Asian residents than the local environment). This shift from the top two panels to the bottom two panels shows that a handful of districts had fewer Black and Asian residents compared to their local environments, with no commensurate districts with a larger share.

To explore how much of these patterns were about the community college districts versus residential segregation, we also examined the same analysis by using Texas counties instead of districts. This additional exploration also had the benefit of allowing us to examine what would happen if Texas shifted to a county-based district system, which several other states have (as we detailed in the prior section). Although the patterns between counties and their local environments were generally similar to those of the district analysis, the trend of extreme negative values we found with districts and Black and Asian residents was not exactly replicated in the county analysis. Given the size difference between the number of Texas districts and counties, Figure A3 shows the percentage of districts and the separate percentage of counties with extreme negative differences (district or county share smaller than local environment share), using multiple minimum thresholds. For White and Latinx residents, we show that the share of districts and counties with negative differences is fairly similar. The bottom two panels show a consistently larger share of districts with extreme negative differences in Black and Asian residents compared to those of counties. This difference across the two analyses provides suggestive evidence that a small number of districts in Texas may exhibit signs of racial imbalance (although, again, we stress that there is no clear cutoff for when a district is racially imbalanced).

We found significant variation across states in their policymaking structures for the creation of community college districts. We also found evidence that, although the majority of districts in Texas have similar racial demographics when compared to those of their local environments, a small number of districts in Texas may create concern for racial imbalance. Although higher-education research more commonly examines policy impacts than the policymaking and political structures that undergird state higher-education policy ([5]; [27]), these structures themselves can have a lasting influence on the educational decisions and experiences of students. In Texas, it appears that some districts could be redrawn in a manner that would create more racially balanced geographic areas. This possibility warrants further study by scholars and Texas policymakers.

The national analysis shows that the majority of states created some type of community college district to determine whom the institutions would target. The majority of these districts' boundaries relied on other political geographic areas, such as counties, cities, and K–12 school zones. Therefore, it is likely that racial imbalance in these other political boundaries could also play a role in determining which community colleges students choose to attend. Further, around 45% of the states with districts had some sort of tuition reduction for students who resided within a certain area of the state. Frequently, within states that had a tuition reduction, all geographic areas of the state were not included in the tuition reduction areas (meaning that some residents within those states would still have to pay a higher "in-state" rate, even if they attended their in-district institution). Therefore, even states with community college districts can provide different incentive structures for residents to attend their in-district institutions.

There is still a significant amount of information the field and policymakers do not know about these policies. One of the central contributions of this study is Table 1, which provides an overview of each state, whether it has a community college district policy, and, if it does, key characteristics of the policy. We hope that our paper will help provide a basis for other scholars to explore the process stakeholders use to change these district boundaries, the motivations behind changes, how frequently changes are enacted across the country, and the causal impact of these policies on students' higher education enrollment and success. We examined documents from the 1960s to the 2020s and also found that, in select states, there was no written documentation of the districts' emergence (leading us to contact state policy actors). In some states, community college district policies are more of a historical fact or considered a "gentleman's agreement," as one state representative shared with the research team. In others, these boundaries are vigorously contested and being changed even as we collected the data for this research project. The current study provides an overview of the national landscape of community college districts while also shining a light on areas that are still understudied.

Based on our findings, two areas of future research are of particular importance. First, it would be useful to understand how the differences in who has ownership of this policy relate to political or educational characteristics of states. It could be that states that include the legislature in boundary decision-making drastically differ from states that give authority to the higher-education governing board. This exploration combined with document analysis of the district policy language could expand the field's understanding of the political dynamics relating to these policies. Second, a deeper understanding of how states decide to provide a tuition discount and how that is funded is necessary. Prior scholarship has focused on whether these tuition discounts incentivize students to attend their designated institution, which they appear to do (e.g., [2]; [15]). However, no systematic evidence shows how these tuition discounts are funded, nor is there a clear explanation of why some states choose to have multiple geographic areas (separating the target population for the institution from who receives a tuition discount) and others have a single area (defining who is the target population and who receives tuition discounts). Future research addressing these points would expand the field's understanding of community college districts and provide state policymakers and policy intermediary organizations with a stronger understanding of the policy practices and motivations of other states.

The case study of Texas provides evidence that a small number of districts could merit further consideration for the presence of intentional racial imbalance. We provide descriptive evidence that, even though racial demographics across Texas community colleges vary widely, the racial demographics of districts generally mirror those of their community college enrollments. We then show, similar to [37] regarding K–12 attendance zones, that evidence suggests that the boundaries of some of the Texas districts may have been created to concentrate groups of residents within certain districts. It appears that a "student exchange" may have occurred, with trade-offs in residents between districts. As noted previously, these types of exchanges are not necessarily undesirable; it could be that state policy actors have created districts with racial equity in mind to ensure stronger racial integration or more equitable exposure to higher resourced institutions.

Current analysis methods for racial imbalance are difficult to interpret, leading us to stress that although we have identified that some districts may warrant further exploration, we cannot speak to whether those districts are gerrymandered or "significantly" imbalanced. It can be difficult to interpret the implications of racial imbalance analysis. For example, the typical guidance when doing the demographic analysis we conducted is to examine districts with at least 5 percentage points of difference between districts and local environments. When we examine the differences for smaller racial groups (e.g., Native American students), we will likely never be able to find a 5 percentage point difference due to the population sizes in Texas as well as the dispersion of these individuals across the state. Therefore, we encourage researchers to work to create newer measures of racial imbalance or analysis techniques that could potentially work in situations like the ones we have just described.

Still, our finding that a handful of districts consistently had smaller shares of Black and Asian residents compared to those of the local environments does provide motivation for further exploration into how those districts were created and whether there is a better method of designing those districts. When we examined counties and their local environments, we found that the number of counties with extreme differences for Black and Asian residents have a different distribution from the same analysis for districts, with more balance between the county and its local environment. We recognize that our exploration of a county-based system has some limitations. If the state of Texas were to shift to a county-based system, it is probable that multiple counties would be grouped together (because the state would be unlikely to have more than 200 community colleges). And yet, even if new districts were created solely by combining counties, for equitable outcomes, that grouping would need to be done in a purposeful manner, given funding and other resource disparities across community colleges.

Future research could examine what happens to the racial demographics of community college enrollment when district boundaries are changed, potentially using this variation to produce causal estimates of their effects. Although prior researchers have examined district boundary changes (e.g., [15]), these analyses were typically done by using student-level data based on the K–12 district students attended. Examining these expansions by using a geospatial analysis could provide a unique contribution to the field's and policy actors' understanding of community college districts' boundaries and their potential role in racial segregation in enrollment. It would also be useful to better understand whether the racial patterns we identify relate to the age of residents. Due to data limitations, we were only able to examine the racial demographics of all residents. Additional exploration with age breakdowns could be helpful when attempting to examine such traits as whether community colleges with K–12 partnerships have more alignment in racial demographics with their district. Because the majority of community college students are generally adults 18 and older, though, we do believe that our research produces evidence that is still relevant to broader policy discussions.

This work is part of a larger research project that also analyzes the relationship between the policymaking decisions driving district boundary creation and racial segregation across community colleges within the state of Texas. Therefore, although this paper does not present evidence on whether district boundaries relate to segregation in Texas community college enrollment (we note the potential for it throughout the paper), we stress that future research must consider how these districts are crafted when including them in research. Community college district boundaries are not neutral; in fact, we found evidence that the drawing of some of these boundaries in Texas, or the boundaries they are based on, reflects a racial imbalance with the larger geographic region. Our findings emphasize that district boundaries are political artifacts; future research and policy must treat them as such.

Understanding how states determine who the "community" is for a community college is a vital undertaking. There is real potential that the manner in which these districts are crafted directly relates to racial segregation between community colleges. Given that community colleges educate a significant share of the U.S. postsecondary education population, it is profoundly important to understand whether the policies determining who is served by certain community colleges are politically driven to exacerbate racial segregation.

Graph: Supplemental material, sj-pdf-1-aer-10.3102_00028312231162347 for Defining the "Community" in Community College: A National Overview and Implications for Racial Imbalance in Texas by Dominique J. Baker, Bethany Edwards, Spencer F. X. Lambert and Grace Randall in American Educational Research Journal

The authors would like to thank Salvatore Saporito, Nicholas Hillman, Meredith Richards, and Christopher Bennett for their helpful comments. The authors would also like to thank seminar participants at Brown University, Harvard University, and the University of Arkansas for their feedback. The research reported in this article was made possible by a grant from the Spencer Foundation (#202000156). The views expressed are those of the authors and do not necessarily reflect the views of the Spencer Foundation. The authors bear sole responsibility for the content of this article.