Abstract
This chapter examines how studies focused on the same topic—disproportionality in special education—can generate vastly different conclusions about its sources and causes. By analyzing existing disagreements in the field, we explore essential questions about what constitutes high-quality and relevant evidence when seeking to understand how, when, for whom, and why disproportionality occurs. Using a holistic review of the empirical literature on disproportionality, we illustrate how differing epistemological and ontological views inform research around the topic of disability in schools and argue that to develop high-quality evidence around disproportionality, researchers need a shared framework that describes how school-based disabilities and classification processes intersect. A shared framework will enable researchers to evaluate whether their findings are expected or unexpected, connect to other related research, and build and rebuild paradigms around issues of equity in special education, rather than disregard one set of findings over another.
Nationally, students of color 1 are numerically disproportionately represented in special education classifications (U.S. Department of Education, 2020). Artiles and Trent (1994) describe this disproportionality as the unequal representation of minoritized students in special education programs, with which two patterns are associated—underrepresentation and overrepresentation—both representing unique equity issues. American Indian or Alaskan Native students and African American students are numerically overrepresented in special education, particularly in disability classifications associated with practitioner judgments (e.g., emotional disturbance, intellectual disability, specific learning disability), while Asian American students are underrepresented. 2 Researchers had identified equity issues related to disproportionality before the enactment of the earliest federal special education laws and policies (e.g., Dunn, 1968; Education for All Handicapped Children Act, 1975), but research on special education disproportionality is still relevant within the scholarly community, as the field continues to amass a large number of studies seeking to explicate the root causes of overrepresentation, particularly for African American, Latinx, and American Indian and Alaskan Native students (e.g., Morrier & Gallagher, 2011; Shifrer, 2018; Waitoller et al., 2010; see Figure 1).
Frequency of Studies by Year of Publication
Despite an extensive history of research demonstrating the presence of overrepresentation in the above groups, recent studies have challenged this long-held finding, arguing that minoritized students are, in fact, underrepresented in special education. These studies show that although bivariate risk indices indicate overrepresentation, when controlling for academic and behavioral factors, particularly with preschool academic readiness variables, African American, Latinx, and American Indian and Alaskan Native students are underidentified 3 for special education services, given their need for academic and/or behavioral supports compared with similar White peers (e.g., Morgan et al., 2012; Morgan et al., 2015). Moreover, several of these studies challenge the existing notion that systemic racial bias in special education (as framed by Coutinho & Oswald, 1998; 2000) contributes to disproportionality, amplifying the rift among scholars studying disproportionality (see Blanchett & Shealey, 2016; Collins et al., 2016; Connor et al., 2019; Morgan & Farkas, 2016; Skiba et al., 2016) and compelling the field to rethink what it looks like to examine issues of equity in special education related to disproportionality. While new and emerging scholarship in the field highlights the role that racial bias plays in special education, these studies continue to argue that, when controlling for academic, behavioral, and socioeconomic predictors, students of color are less likely to be classified as disabled than their peers. The rift lays bare essential questions about what constitutes high-quality and relevant evidence when seeking to understand disproportionality.
Scope of the Chapter
Through a holistic review of the literature, we explore the disjointed findings among disproportionality studies. We describe the current state of research and identify how research focused on the same topic—disproportionality—generates vastly different conclusions. We aim to understand and, where possible, reconcile meaningful gaps between conflicting claims about the topic and offer a framework to assess the veracity and relevance of research in this field. In doing so, we show that high-quality research evidence extends beyond generally accepted methodological and discipline-specific standards for empirical investigation and reporting for any single study or type of study. High-quality research evidence necessitates a framework constructed from a combination of research and theory (see Ravitch & Riggan, 2016) that engages with scholarship across disciplines and methodological traditions working within the field. Having a framework helps scholars (a) identify and understand the gaps in research and (b) bring disparate findings into conversation with one another, thus, moving the field closer to comprehensive and actionable understandings of disproportionality that can meaningfully inform policy and practice at multiple levels (e.g., national, state, and local).
The chapter is based, in part, on a review and critical analysis of 73 recent empirical studies on disproportionality, published between January 2006 and December 2019. 4 The authors of these studies examined processes, practices, and outcomes related to understanding student ability and disability and special education placements in K–12 school spaces and published their findings in peer-reviewed academic outlets. 5 Additionally, given the large and comprehensive body of research on disproportionality, we draw on existing research syntheses that focus on specific time frames (e.g., literature published before 2006; Waitoller et al., 2010) as well as focused reviews grouping studies based on methodology (e.g., Cruz & Rodl, 2018; Harry & Fenton, 2016; Morgan et al., 2015) and specific demographic categories (e.g., Cooc & Kiru, 2018; Kulkarni, 2017).
As scholars continue to employ varied disciplinary lenses and approaches to the research, a clear consensus in the field is that disproportionality represents a highly complex phenomenon resulting from multiple factors and features (Artiles et al., 2016; Coutinho & Oswald, 2000; Cruz & Rodl, 2018; Heller et al., 1982; Skiba et al., 2008; Waitoller et al., 2010). This review’s underlying structure enabled us to consider a wide range of research from multiple fields (e.g., psychology, sociology, public health) to uncover lessons for the field of education research on the topic.
We recognize that through the intricacy of examining a broad field and grouping studies into categories, we run the risk of creating, maintaining, or re-informing rigid binaries. Thus, it is essential to acknowledge as we categorize studies that scholars often hold nuanced and thoughtful views of the field that are not easily categorized. We see their work as reflective of rigor and grace and indicative of a desire to improve educational outcomes. In our efforts to discuss what constitutes high-quality research, it is not our intention to minimize or cast aspersions on any study or scholar, as we have come to see each piece of research as part of a larger conversation. We bring perspectives on special education, disability, and disproportionality that are informed by our own experiences as practitioners and researchers. We have all worked as practitioners serving diverse students with a range of dis/abilities and in varied classroom settings. 6 We have also served as technical assistance providers, supporting schools and districts wrestling with disproportionality. As researchers, we have engaged with issues related to disproportionality from multiple perspectives, which informs our belief that we can learn from the vast array of scholarship in this field. We are open to multiple approaches to disproportionality study. At the same time, our experiences and interpretation of the research literature lead us to advocate for a reconceptualist view of disability (Andrews et al., 2000).
Learning From a Disjointed Field
The majority of research on disproportionality falls into one of two dominant frames—professional practices and sociodemographic predictors (Waitoller et al., 2010; see Table 1). Professional practice studies examine the more technical dimensions of classifications as they occur within schools (e.g., referral practices, assessments, decision-making processes), using a combination of in situ approaches (e.g., ethnographies and case studies) or collecting data within a bounded set of schools or districts through means such as observations, interviews, surveys, and focused reviews of local administrative data. Sociodemographic studies use quantitative designs, drawing on demographic, economic, and academic performance data obtained from large national databases (e.g., Early Childhood Longitudinal Study, Education Longitudinal Study), local school districts, or state administrative data, estimating the relationship between individual students’ race and their likelihood of being classified as disabled. While the studies using these frames share a focus on understanding disproportionality, their conclusions can seem disjointed (i.e., different methodologies, theories, and frameworks can produce different interpretations of results; see Cruz & Rodl, 2018).
Descriptive Characteristics of Selected Studies
Note. Although Waitoller et al. (2010) used framework categories that were mutually exclusive, our analysis did not, and therefore, in the “Key frameworks” row, counts do not add up to 73, and percentages do not add up to 100.
Research on professional practices generally agrees that minoritized students experience the special education classification process differently than their peers, with several studies identifying various ways in which educators and other stakeholders interpret and assess the behaviors and abilities of minoritized youth, leading to their increased likelihood of being referred for special education within local contexts (Escamilla, 2006; Fish, 2017) and, ultimately, their likelihood of being classified as disabled (Artiles & Trent, 1994; Hosp & Reschly, 2003; Linton, 2015; Ysseldyke et al., 1997).
Scholars have found evidence of subtle differences in teachers’ perceptions of student academic ability and social and behavioral skills based on race, gender, socioeconomic status, and other cultural standards (Covay Minor, 2014; Ready & Wright, 2011). Fish (2017) suggested that practitioners were more likely to refer African American students to be assessed for special education when they exhibited behavioral challenges, as opposed to White students, who were referred when exhibiting academic challenges. While teachers’ assessments of behavior relate in part to their perceptions of achievement (Bennet et al., 1993; Espinosa & Laffey, 2003; Mehan, 1980; Tach & Farkas, 2006; Tyler et al., 2006), this relationship is stronger for African American students than for White students (Covay Minor, 2014). For example, Cooc (2018b) examined teachers’ perceptions about whether or not students in their classes had a disability and found that teachers disagreed about the presence of disability more often when students were African American, male, and from lower socioeconomic strata. While academic achievement differences did not predict disagreement, disruptive classroom behavior did. Cooc theorized that teachers make assumptions about student abilities based on behavior rather than the level of engagement a student displays, suggesting a need to understand why “student behavior differs across classroom[s] to the extent that teachers have different opinions about disability status” (p. 76). Cooc’s study aligns with research showing that observations and assumptions about students’ socioeconomic status, culturally specific behaviors, and other nonacademic factors (i.e., race and language background) may moderate teachers’ subjective evaluations of students’ academic ability and behavior (e.g., Andrews et al., 1997; Dhuey & Lipscomb, 2010; Escamilla, 2006; Lambert, 2015; Neal et al., 2003; Saft & Pianta, 2001; Rist, 1970; Shippen et al., 2009). Moreover, research focusing on the special education classification process points to potential biases resulting from how teams of educational experts and service providers interact with each other (Hart et al., 2010; Knotek, 2003; Kramarczuk Voulgarides, 2018), how the available data on students inform disability classification risk (Cruz & Rodl, 2018; Kincaid & Sullivan, 2017; Klingner & Harry, 2006; Knotek, 2003; Overton et al., 2004; Wilkinson et al., 2006), and how policy guidance may be inadequate and ill defined (DeMatthews et al., 2014; Kramarczuk Voulgarides, 2018). The totality of this research highlights the ways in which minoritized students face differential risk of being referred for evaluation and ultimately classified as disabled.
While professional practice studies appear to agree that minoritized students face bias in the process of being classified as disabled, sociodemographic research has uncovered a larger story about the processes surrounding disproportionality. Initially, aggregate-level studies—those examining school- or district-level proportions of students identified as having disabilities—sought to empirically estimate how a range of school- and district-level factors influence a student’s likelihood of being placed in special education. These studies found that schools’ and districts’ sociodemographic characteristics were associated with the proportion of students identified with disabilities (see Coutinho et al., 2002; Eitle, 2002; Hosp & Reschly, 2004; Oswald et al., 1999; Skiba et al., 2005). While these studies consistently documented racial/ethnic overrepresentation in special education, aggregate-level studies are unable to account for differences in individual student achievement and behavior that may be related to variations in special education outcomes (National Research Council, 2002a). Aggregate-level studies also cannot account for individual characteristics that might be related to the likelihood that a students is classified as disabled (i.e., any other confounding variables), and they are prone to the ecological fallacy (Freedman, 2001; see Morgan, Farkas, Cook, et al., 2017).
To correct for this, there has been an increase in studies that utilize solely individual student-level data (without aggregate-level data) to estimate the extent to which an individual student’s race is associated with the likelihood of receiving a special education classification, while also controlling for factors such as student-level academic achievement, behavior, and socioeconomic status (e.g., Morgan et al., 2012; Morgan et al., 2015). 7 These studies concluded that the socioeconomic and academic variables confound the effect of race as a significant predictor of whether or not students are classified as disabled. For example, Morgan and colleagues concluded in several studies that, when including these variables, minoritized students are underrepresented in special education when compared with White students, indicating that students of color are less likely to be receiving special education services than otherwise similar White peers (Morgan et al., 2012; Morgan et al., 2015; Morgan, Farkas, Hillemeier, & Maczuga, 2017). These studies have been the basis for calls to modify current federal educational legislation and policymaking designed to address disproportionality, which remain controversial and contested (see Blanchett & Shealey, 2016; Cavendish et al., 2020; Collins et al., 2016; Connor et al., 2019; Skiba et al., 2016; Welner & Skiba, 2015).
While this work has sparked criticism, sociodemographic studies of disproportionality have increasingly drawn on different data sources and more refined statistical models to show that seemingly clinically relevant student-level predictors of special education mitigate individual markers of racial and ethnic differences (e.g., Elder et al., 2019; Fish, 2019a; Galster et al., 2016; Hibel et al., 2010; Shifrer, 2018) and particularly that there are significant academic and behavioral predictors that reduce the effect of race (see Morgan, Farkas, Hillemeier, & Maczuga, 2017).
In addition to student-level factors, sociodemographic studies have also identified school characteristics, such as the demographics of both teachers and students within schools (Elder et al., 2019; Fish, 2019a; Galster et al. 2016; Hibel et al., 2010; Shifrer, 2018), the level of student mobility in and out of schools (Bal, Sullivan, & Harper, 2014), and schools’ average level of student engagement (Hibel et al., 2010), as significant factors. Studies by Fish (2019a), Shifrer and Fish (2019), and Elder et al. (2019) provide support for the idea that the racial composition of a school and/or district and a student’s racial distinctiveness within a school shape racialized outcomes in special education. For example, Elder et al. (2019) found that while African American and Latinx students were classified as disabled at lower rates than their White peers, these groups were overrepresented in special education in schools with relatively small shares of African American and/or Latinx students and substantially underrepresented in schools with large minority shares, what Hibel et al. (2010) described as the “frog pond effect” (p. 312). Fish’s (2019a) analysis showed that as the proportion of White students increased in a school, the risk of being classified as having a lower-status disability (e.g., intellectual disability) increased for minoritized students. However, Fish also noted that as the proportion of White students decreased, White students’ risk of being classified as having a higher-status disability (e.g., speech and language impairment) increased, indicating the need for a nuanced understanding of how students are pathologized based on school context. Thus, the landscape of sociodemographic studies evolved from claiming overrepresentation to general claims that all things being equal, students of color are less likely to be classified as disabled than their peers when controlling for academic and socioeconomic variables.
As sociodemographic research shifted from aggregate-level studies to individual- and mixed-level studies, key divisions between sociodemographic and professional practice studies have emerged. While both qualitative and quantitative studies on professional practices continue to find credible evidence that minoritized students are likely to face differences in educational processes, especially in special education referral and classification, the recent sociodemographic research described seems to challenge this general assumption, noting that when controlling for clinically relevant predictors, specifically academic achievement, students of color are as likely or less likely to be classified as disabled compared with their peers, and indicating potential biases that contribute to underrepresentation and potential underidentification (e.g., Morgan, Farkas, Hillemeier, & Maczuga, 2017). It is only in hypersegregated settings—where minoritized students represent only a small portion of student enrollment—that they are more likely to be classified as disabled than similar White peers (Elder et al., 2019; Fish, 2019a).
Relatedly, there is a disconnect concerning where the two types of studies situate racial bias as a contributor to racialized special education outcomes. Sociodemographic research describes racial inequity as related to racialized social structures (Sullivan & Artiles, 2011), which create the conditions for disproportionality to occur (e.g., Elder et al. 2019; Fish, 2019a, 2019b). At the same time, professional practice research identifies sources and sites of racialized bias at the individual level (e.g., how students are perceived, treated, or discussed in schools) and how they contribute to racialized special education outcomes (e.g., Cooc, 2017; Escamilla, 2006; Lambert, 2015; Skiba et al., 2006; Yoon, 2019). These findings’ disjointed nature provides an interesting chance to explore essential questions about what constitutes high-quality and relevant evidence when seeking to understand how, when, for whom, and why disproportionality occurs.
It may be incomplete to consider high-quality and relevant evidence as a property or a result of any single study—as any single study or line of research, no matter how rigorously sourced or executed, can often only give a partial answer to a complex question. While not detracting from the importance and high-quality evidence of any single study or group of studies, the disjointedness in research on special education disproportionality necessitates a view of high-quality and relevant evidence that resonates across a larger, more complicated, and nuanced field. In what follows, we explore the different views that inform research around the topic of disability in schools and show how disconnects in approaches to studying disproportionality and in views on disability itself mirror the disjointedness in the field. We then argue that to develop high-quality evidence around disproportionality, researchers need a shared framework that helps reconcile and give space for differing and disjointed views so the field can advance and provide new insights on this complex issue.
Agency and Structure
The disjointedness in our understanding of disproportionality is related, at least in part, to the differing approaches to research (i.e., research designs, methodologies, data selection and analysis) embodied in the sociodemographic and professional practice frames. As noted above, professional practice studies focus attention on the processes within schools that lead to or contribute to disproportionality, using primarily qualitative approaches that critically examine the technical dimensions of classifications (e.g., referral practices, assessments, decision-making processes). In contrast, sociodemographic studies rely primarily on quantitative approaches and large administrative data sets or surveys to model the characteristics of individuals and contexts believed to be related to special education classifications and disproportionality. Without privileging one approach over another, we observe that research on disproportionality is marked by a gulf between research on the social structures associated with special education classifications and research on the role of human agency in the special education classification process. This is similar to an idea that Mehan (1992) employs when discussing “micro-macro, agency-structure dualism” (p. 17) and is based on a long line of sociological research and theory (see Alexander et al. 1987; Collins, 1981; Giddens, 1984).
As a group, sociodemographic studies effectively elucidate the role that macrosocial structures play in shaping special education outcomes, positioning individual characteristics and outcomes in relation to structural elements of schools and communities (Cooc, 2018a; Kincaid & Sullivan, 2017; Sullivan & Artiles, 2011). By engaging in localized data collection and focusing on the local policies, practices, and beliefs of local stakeholders involved in the special education classification process, professional practice studies often place at the center of their analysis the role that individuals’ agency plays in the special education classification process, either in the furtherance of classifications or in their efforts to resist or minimize them. These studies provide insights into the mechanisms that lead to disproportionality—the spaces where subjectivity and educators’ agency give way to bias (e.g., Escamilla, 2006). Still, in general, these agency-centered studies do not sufficiently attend to how structural elements around social actors bound or influence their choices—unless the studies are comparative, theoretically situated, or include discussion of broader quantitative data to frame the analysis.
Structural studies, conversely, are more likely to frame individuals in the special education classification process as passive actors and at the whim of larger social forces, as agentic actions are constrained by variable choice and statistical modeling strategies. What we see when we put sociodemographic studies in conversation with one another is that the methodological choices they embody underscore the importance of understanding context in special education classifications, and they contribute to dramatic changes in how we view the ways in which racial bias functions in the special education classification process. In effect, the sociodemographic studies point to a more structural understanding of disproportionality.
Aggregate-level studies within the sociodemographic field, drawing mainly from district-level data, often find that structural racial bias contributes to overrepresentation (e.g., Coutinho et al., 2002; Oswald et al., 1999; Skiba et al., 2005). Studies have found that “nationally representative data from very large samples of schoolchildren characterized as otherwise similar in their clinical need for services as indicated by their academic achievement” (Morgan, Farkas, Hillemeier, & Maczuga, 2017, p. 307) show an underrepresentation of minoritized students classified as disabled and have led to assertions that barriers and inequitable access to services may contribute to inequities (Morgan et al., 2015). Where sociodemographic studies seem to converge is around the finding that minoritized youth are largely underrepresented in special education (when controlling for presumed confounders). Still, as researchers consider context variables alongside student-level predictors in their models, the magnitude of this underrepresentation in special education reduces (Hibel et al., 2010; Mann et al., 2007; Shifrer et al., 2011; Talbott et al., 2011), underscoring the importance of local contextual factors and variation in special education classification outcomes. In recent years, multilevel analyses have highlighted the relationship between racialized special education outcomes and context (Elder et al., 2019; Fish, 2019a; Shifrer & Fish, 2019)—a point not dissimilar from the finding of the original aggregate-level research (e.g., Coutinho et al., 2002; Coutinho & Oswald, 2000; Oswald et al.,1999). But structural views of disproportionality from sociodemographic research require sufficient understanding of the special education classification process to draw valid inferences about racial bias (National Research Council, 2004). As sociodemographic scholarship expands, researchers are still limited in their ability to model a coherent understanding of the process that they believe leads to racialized special education outcomes.
Studies on professional practices, when pieced together, are arguably better able to elucidate possible pathways to racialized special education outcomes. However, their more narrow or bounded scope, as compared with sociodemographic studies, can obfuscate the role larger social structures and contexts may play in moderating the influence of particular practices that produce disproportionate outcomes, unless comparisons and theoretical framings explicitly draw these connections. Compared with sociodemographic studies, research evidence focused on professional practices comes from relatively confined contexts and a limited number of informants. Researchers purposely source their evidence from contexts where disproportionality is apparent, enabling them to ask questions about how and why disproportionality occurs. By delving into agency and constitutive actions of and within schools, these studies make claims about how key stakeholders in the special education classification process engage with students, how professionals shape student trajectories toward or away from disability classification, and how parents engage with and make sense of special education and the classification process (Dávila, 2015; Harry et al., 2005; Kozleski et al., 2008; Lambert, 2015). As such, the studies that center individual agency tend to link racialized outcomes to the treatment of individual students that results from interpersonal interactions. For example, studies show how teachers’ interpretations of student ability (Banks 2017; Escamilla, 2006; Lambert, 2015), inconsistencies in the special education classification process (e.g., Cooc, 2018b; DeMatthews et al., 2014), or a combination of both (e.g., Harry & Klingner, 2014) disadvantage minoritized students and contribute to the persistence of disproportionate outcomes in local contexts.
The literature shows that special education classifications and variations in outcomes are born from the relationship between individual actors within schools and districts as well as local policies and practices (i.e., agency), which are surrounded by contextual, sociohistorical, and political factors (i.e., structure). And while we found that agency-centered studies, in general, are well equipped to identify how racial bias or racially biased policies or practices are related to disproportionate special educational outcomes, they do not sufficiently account for the way racialized structural elements bound or influence choices. Conversely, structural studies frame individuals in the special education classification process as passive and following larger, scripted social forces. Neither perspective on its own can fully tell the story of disproportionality. But in holding both views in relationship with each other, we are better positioned to understand the intersections between professional practice and sociodemographic literature in order to advance the field and resolve the disjointed findings.
By foregrounding the relationship between agency and structure, we can both understand the limitations of individual studies and create connections across fields and disciplines. The agency–structure relationship highlights how high-quality sociodemographic research sheds light on when, where, and for whom disproportionality occurs while professional practice research provides insights into how and why it occurs. Both must be accounted for in relation to each other, providing a more complete and productive understanding of disproportionality. More broadly, the agency–structure continuum in disproportionality research is an example of an epistemological divide that many fields contend with. In fields where research spans multiple approaches, scholars must bridge gaps and make connections across approaches so that findings from individual studies speak and relate to each other in meaningful ways.
Defining Disability
In addition to contending with the tensions between agency and structure, and the aforementioned ways in which studies situate racial bias given their approach to the agency–structure continuum, we found that different studies employ different definitions of disability. The idea that different researchers and practitioners hold conflicting conceptions of disability is not new, as scholars have noted how these differences shape both research and practice and have sought to bridge the divide (see Andrews et al., 1997; Baglieri et al., 2011). While a diversity of research approaches can shed light on different aspects of disproportionality and create a more complete picture of the phenomenon, studies that utilize different definitions of disability may be exploring different phenomena altogether.
There are two prominent understandings of disability that permeate the field: a medical model of disability and a social model, each often implicitly structured into research studies on disproportionality. Medical models view disability as an observable trait in a student that is uncovered through medical diagnoses and identified throughout the classification process (Valle & Connor, 2019). In contrast, social models view disability as deriving from the interactions between individuals and society and as a “social and context-dependent construct” (Cooc, 2018b, p. 76). These models describe policy and practice developments that have evolved to define and interpret human differences as disabling (Artiles et al., 2016; Baglieri & Lalvani, 2020; McDermott & Varenne, 1995; Sleeter, 2010; Valle & Connor, 2019). The different perspectives shape the constructs and variables under consideration in research studies and how evidence related to disproportionality is analyzed and interpreted.
The medical view of disability is present in both sociodemographic studies as well as a subset of studies on professional practices (e.g., studies of assessment and intelligence measures or studies that measure the academic or social performance of students in an experimental or quasi-experimental design; e.g., Beaujean & McGlaughlin, 2014; Nagliery & Rojahn, 2001; Nakano & Watkins, 2013). Collectively, under the medical model, disability diagnoses or classifications are tied to measures of the individual’s capacity and ability to learn (Dudley-Marling, 2004).
Social models of disability are more commonly found within agency-centered approaches that describe the interactions between educators or between educators and students and implicitly define disability as the result of a social process. For example, studies that examine how teachers make referrals or how Individualized Education Program teams classify students as disabled often do not assume that they are disabled at the start (e.g., Bardon et al., 2008; Cooc, 2018b, 2019; Escamilla, 2006). Instead, researchers observe a process unfold related to classification and the administration of special services in schools. Research on professional practices continues to demonstrate the ways in which individual and group actions are related to special education classification, either in the labeling process or through the early intervention process, and, relatedly, the enactment of policies and practices that might prevent the need for eventual classification and special education service receipt within local contexts (e.g., Bardon et al., 2008; Muschkin et al., 2015). As such, an individual student’s special education classification is the result of a series of professional practices, from teachers’ referrals to assessments, to meetings, leading to classification and receipt of services (Gottlieb et al., 1994), but as discussed above, it is also subject to broader structural and social influences.
A critical difference between the social and medical models of disability in research studies on disproportionality is the way researchers treat and understand academic achievement and behavior as they relate to a disability diagnosis. While both models implicitly or explicitly link achievement and behavior to disability classifications, they do so in different ways. The medical model implies that disabilities reside within individuals, whereas the social models situate disability within structures that operate within culturally constructed and often problematic assumptions of normalcy (e.g., Lambert, 2015; Lewis-McCoy, 2016). Using the medical model, sociodemographic studies control for academic achievement with the assumption that it is possible to determine a student’s (dis)ability through academic assessments and other “clinically relevant factors” (Shifrer, 2018, p. 387), such as poverty, that may be indicative of neurological or developmental differences (e.g., Beaujean & McGlaughlin, 2014; Scarborough & McCrae, 2010; Simos et al., 2005; Temple et al., 2010). This approach is in line with the idea that “deviation” is related to limited “biological function resulting from a physical, cognitive, or sensory impairment,” which distinguishes two states of being “normal and pathological” (Baglieri & Lalvani, 2020, p. 19). However, even the diagnostic criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders for many school-based disabilities are socially rooted and subjective (Pickersgill, 2012; Shifrer, 2018).
Though differences in neurocognitive development may also be related to specific abilities and disabilities (e.g., Mussolin et al., 2010; Paulesu et al., 2001; Simos et al., 2005), biological and neurological markers of difference and special education classifications in schools are not necessarily related, as the social model indicates. Muschkin et al. (2015), for example, describe three etiologically distinct groups of children who are at risk for disability, which touch on the intersection of the medical and social models of disability in research: one group that comprises children “born with chronic disabilities that require life-long attention, such as physical handicaps and genetic abnormalities”; a second group, “who are at risk for a later chronic disability that could be alleviated by early identification and educational treatment”; and a third group, for whom, “with high-quality early environments, child care, and educational experiences, the need for special education placement could be prevented altogether” (p. 479). As such, the connection between students’ abilities, behaviors, and support needs are not necessarily fixed but instead are often socially constructed within schools. Opportunities provided to students and decisions made about students within school contexts can affect both achievement and classification (Cooc, 2018b; Lambert, 2015). Moreover, while achievement and behavior data are a part of special education decision making, they are not used to classify students in systematic and consistent ways (Fish, 2017; Hart et al., 2010; Knotek, 2003), and as discussed above, teachers’ perceptions of students within their local school context influence referral and assessment decisions (Mehan et al., 1986). As such, studies employing only medical models of disability may fail to map well onto the processes associated with special education classification (starting in general education through to receipt of an Individualized Education Program and special education services).
The political and legal definitions of disability categories defined by the Individuals with Disabilities Education Improvement Act (IDEA, 2004) are centered within a medicalized model of disability (Artiles, 2011; Sleeter, 1986), and these categories underpin special education practice and research (Kozleski, 2016). Researchers employing medical models use common definitions of disability categories to compare student outcomes across large amounts of data and varied contexts to estimate the influence that non-disability-related factors play in disability classification (e.g., poverty, race, gender, school composition) and to examine the structural factors associated with special education placement (see Sullivan & Artiles, 2011).
However, even though categorizations, definitional terms, and regulations span contexts, in the special education classification process, what constitutes school-based disabilities is weakly structured in everyday use across school districts and within schools. Operational and conceptual definitions of ability and disability vary across states, districts, and schools (Kidder-Ashley et al., 2000; Reschly & Hosp, 2004; Singer et al., 1989), as do the formal mechanisms that constitute the special education classification and labeling process (Bahr et al., 1999; Buck et al., 2003; Truscott et al., 2005). Administrators and practitioners interpret federal and state policy regulations differently across state and local jurisdictions (Harry & Klingner, 2014; Kramarczuk Voulgarides, 2018; Tefera & Kramarczuk Voulgarides, 2016). Furthermore, these definitions have changed considerably (Artiles, 2011; Sleeter, 1986). Coupled with variations in educational resources (often based on variations in local contexts), special education classification policies and practices thus differ across schools (Bradley et al., 2007) and differentially affect the students who are classified as disabled. These differences create conditions in which disabilities are socially constructed while still loosely coupled to technical and medical definitions.
In this sense, special education classifications operate as boundary objects (Akkerman & Bakker, 2011; Star & Griesemer, 1989), whereby the process for determining what constitutes a valid special education classification is locally contested but is also tied to broader social and structural conditions and medical and technical definitions (i.e., the aforementioned agency–structure dualism; also see Artiles et al., 2016). 8 Local needs, resources, and constraints of schools, districts, and related stakeholders shape special education classifications; yet because disability labels are universally defined in policies such as IDEA, they also share a collective identity across schools and districts. This creates an interconnected network or “virtual community” of school officials, professionals, parents, students with disability labels, researchers, and policymakers (Artiles et al., 2016) that forms understandings of disability and difference in schools. Both agency and structure are accounted for in this virtual community as special education classification processes are created, re-created, and reproduced within and across contexts.
Understanding disability as a boundary object asks sociodemographic studies to delve deeper into the classification process and to recognize the contextual, locally constructed nature of disability; similarly, disability as a boundary object compels professional practice studies to attend to broader political, historical, legal, and technical definitions that may inform local processes. A shared definition of disability that acknowledges the social and medical definitions of disability will enable the field to engage in cross-cutting research that allows for simultaneous conclusions from multiple perspectives. While multiple approaches within and across different disciplines and perspectives (social science, medical science, critical theory) can help build paradigms and theory, a single field must develop shared concepts to engage in field building that contributes to a body of high-quality research evidence.
A Framework for Understanding Disproportionality
Our knowledge base about disproportionality is expanding. However, “our explanations of the problem are not explicitly grounded in theoretical frameworks” (Sullivan & Artiles, 2011, p. 1528). Theory plays a vital role in education research—it can help guide research and the choices made in a field (Kuhn, 1962; National Research Council, 2002b) and guide interpretation of findings and conclusions (Anyon, 2009). Both the growth of and shift in sociodemographic research and the misalignment of sociodemographic and professional practice studies demonstrate how important it is for disparate fields to come together under a shared framework that defines critical theoretical assumptions and explains key relationships.
The disjointedness outlined in the chapter speaks to the need for a more holistic view of research evidence—one that “keeps up with the latest developments in the social sciences” and situates the professional practice and sociodemographic frames in conversation with each other (Waitoller et al., 2010, p. 44), and by extension acknowledges an intricate understanding of disability that spans disciplinary and methodological boundaries. Moreover, to sustain a focus on high-quality evidence, and to stave off balkanization, while still maintaining the unique contributions provided by different approaches and disciplines (Bridges, 2014), it is important for the field to have a shared framework for understanding disproportionality, which will enable researchers to evaluate whether their findings are expected or unexpected, connect to other related research, and build and rebuild paradigms around issues of equity in special education (Cruz & Rodl, 2018; Muthukrishna & Henrich, 2019; Sullivan & Artiles, 2011; Waitoller et al., 2010), rather than disregard one set of findings over another.
In this regard, we draw from Mehan’s (1992) call to collapse the agency–structure dualism that separates different approaches to understanding disproportionality and, in collapsing this dualism, to support a shared understanding of what a disability is within the educational context. Put simply, special education classifications and variations in outcomes are borne through the various contexts and ideologies that surround and constitute a school or district (structural spaces) and the relationships between individual actors within schools and districts (interactional spaces). The interplay of structural spaces and interactional spaces defines both the boundary of special education classification and the process that moves particular students through the threshold to classification.
Structural spaces consist of the ideologies and broader social contexts in which students, schools, and districts are situated. Sociodemographic studies provide clear evidence for the need to consider context. Building off the work of scholars such as Mehan (1992), Bonilla-Silva (1997), and Sullivan and Artiles (2011), these structural spaces shape the policies and practices of educational systems, including the processes that produce special education classifications. The interactional space comprises individuals’ practices, dispositions, and policy interpretations as they implement special education classification processes.
By analyzing the structural space that surrounds schools, researchers can describe how the contours and impacts of racial and class stratification that exist across schools and within society (e.g., racial segregation, discriminatory housing policies, equal access to health care; see Desmond & Emirbayer, 2015) affect how schools and districts operate. The structural space also accounts for how social contexts are shaped, and it is reflective of racial relations, which, in turn, influence institutional practices within a locality and ultimately shape educational opportunities and outcomes (Carter & Welner, 2013). Within this space, researchers can also account for a systems perspective, whereby contexts represent the multifaceted social, economic, political, and environmental patterns that shape students’ development, both academically and behaviorally (Braveman et al., 2011; Bronfenbrenner, 1977, 1979; Elder, 1998), and thus influence special education eligibility.
The structural space does more than encourage more robust statistical models. It also calls attention to the socially constructed elements of disability (Artiles et al., 2016; McDermott & Varenne, 1995) and the extent to which contextual factors and locale shape disability outcomes (e.g., Elder et al., 2019; Fish, 2019a; Gelb & Mizokawa, 1986; Singer et al., 1989), thus reinforcing the idea that disabilities are boundary objects. However, the structural space does not directly explain how these outcomes relate to variations in contexts and, as such, does little to shape the policies and practices related to disability. For that, we must consider the interactional space.
Even with federal and state policy dictates, local actors have considerable leeway concerning how they engage in their professional practices at the local level (Lipsky, 2010), particularly when it comes to special education (Harry & Klingner, 2014; Kramarczuk Voulgarides, 2018; Tefera & Kramarczuk Voulgarides, 2016). Policies and practices are interpreted and enacted by educators. These educators use and interpret school-based policies and practices related to special education classifications such that student ability and disability are culturally constructed within the context of schools (Mehan et al., 1986). When practitioners refer students for special education assessment, they are operating within the constitutive rules (policies and practices) of schools in defining ability and disability based on their general perceptions of a student within their specific school context (Anderson-Levitt, 1984; Mehan, 1980; Mehan et al., 1986). These perceptions of ability and disability are initially codified and enacted in classrooms and eventually inform special education classifications (Harry & Klingner, 2014; Mehan et al., 1986). And educators, in attempting to comply with IDEA (2004), inadvertently reify racialized inequities through acts of regulatory compliance (Kramarczuk Voulgarides, 2018), which serve to define what ability and disability are in the minds of students and teachers (O’Connor, 2020; Rist, 1970). The interactional space is where the policies, practices, and beliefs of race and disability are enacted, but it is also the space where we are most likely to find solutions. A focus on the interactional space provides a more nuanced understanding of disproportionality and the policies, practices, and beliefs from which solutions may arise.
Importantly, from a solutions-based perspective, Bal, Kozleski, et al. (2014) illustrate the importance of working with a wide range of stakeholders (e.g., educators, family members, community members) to address issues of disproportionality, describing the challenging and sometimes contentious push to shift from deficit discourses about students to more expansive discourses that disrupt racial inequities and facilitate systems change. By focusing on individual agency and privileging the voices of those who are thought to be most affected by disproportionality (e.g., Lewis-McCoy, 2016), we can gain an in-depth understanding of how different components of the special education classification process unfold and consider a range of viewpoints that cannot be easily accounted for in more technical and structural studies.
The mechanisms and relationships that connect structural spaces, interactional spaces, and student outcomes are numerous and complex, and warrant further investigation. Special education classifications or disability labels are the “consequence of institutional practice[s]” (Mehan et al., 1986, p. 159) that are shaped by school contexts (Bray & Russell, 2018). Hibel et al. (2010) assert that within the same context, similar students may appear more or less able relative to the perceived ability of their peers; additionally, the researchers opine that the level of district resources may mediate perceptions of ability and disability. The existing literature points to different ways in which contexts around district demographics may inform policies and procedures related to curriculum, instruction, and special education (Anyon, 1980, 1981; Eitle, 2002; Harry & Klingner, 2014; Kramarczuk Voulgarides, 2018; Meier et al., 1989), as well as how concepts of race are understood within school contexts (Lewis, 2003). It is also essential to consider the idea that the interactional space informs and transfers information into the structural space—reinforcing, resisting, and remaking social structures (e.g., Giddens, 1984; Gorski, 2019; Pollock, 2004).
Conclusions and New Directions for Research
In describing the research on disproportionality, we have outlined important ontological and epistemological sources of the disjointedness mentioned above and offer a shared framework for understanding the field, both to serve our colleagues and to demonstrate the benefits of conceptual frameworks in bridging the different approaches. Within the frame of sociodemographic research, the field has made some necessary and important advances concerning our understanding of disproportionality, showing that the likelihood of being classified as disabled varies by context. Simultaneously, the shifting and sometimes divisive landscape of research on disproportionality has laid bare the importance of different scholarly approaches coming together.
What We Have Learned
Developing high-quality evidence requires scholars to attend to and, when possible, bridge epistemological and ontological gaps in their field. As made apparent throughout this chapter and noted in prior research syntheses, scholarship on special education disproportionality straddles multiple disciplines, approaches, and perspectives and, as such, is marked by disagreements (see Cruz & Rodl, 2018; Waitoller et al., 2010). In this regard, research on special education disproportionality is not unique to broader education research, which also crosses disciplines, approaches, and perspectives and is beset by rifts and chasms (Moss et al., 2009; Phillip, 2006). We hope that the lessons from our small corner of the field can aid our collective thinking about high-quality evidence in education research.
The American Educational Research Association (AERA) has created clear guidelines for the interpretation of individual studies in education (e.g., AERA, 2006, 2008, 2009), but concerning interpretation across a holistic field, the standards become less clear. The disjointed nature of the field of study around disproportionality offered us the chance to explore ideas about high-quality research across groups of studies whose approaches to research are often not aligned. What remains is the idea that high-quality research evidence works to reconcile epistemological and ontological divides. Sociodemographic frames and professional practices frames differ in meaningful ways concerning what data are being collected and analyzed (and for what purpose) and, relatedly, how disability is understood in the research (implicitly or explicitly), ultimately shaping our understanding of special education disproportionality in important and sometimes incomplete and misleading ways. However, those differences can be put in conversation with each other by using a shared framework to help clarify and open up new research opportunities.
Developing high-quality evidence requires scholars to use and engage with a shared framework. The idea that high-quality research links to theory is a tenet of scientific research in education, as put forth by the National Research Council (2002b). But high-quality research also brings together the divisions discussed above and holds a space for common points of conversation and agreement. Moss et al. (2009) aptly propose that “[the] quality of research done by a research community is enhanced when that community agrees on how it will measure the features or variables central to the work” (p. 505). We believe that conceptual and theoretical frameworks enhance research by creating a shared space among a community of scholars and a launching point for future inquiry. Ignoring shared frameworks can exacerbate disagreements and misunderstandings and stymie opportunities for cutting-edge research.
We have attempted to show how research on disproportionality benefits from connecting to broader frameworks that purposefully link micro- and macrosocial forces to make sense of disproportionality and seek meaningful comparisons and conversations across disciplines and approaches, and in shared meaning making. By putting sociodemographic and professional practice research within the same framework, we can bridge epistemological and ontological divides and create a space to debate on and challenge the existing knowledge base, guide new research, and test new ideas in ways that advance the field. This is not to say that our framework or any framework is fixed. Frameworks are meant to be tested, bent, broken, and built again. The key is that the scholarly community shares in this process.
New Directions
In what follows, we discuss possible new directions for the field of research on special education disproportionality based on our understanding of the disagreements in the field and our belief that a shared framework born of those disagreements can help us better advance knowledge on how, when, for whom, and why disproportionality occurs.
Engage More With Sociohistorical and Sociocultural Research Frames
We were disappointed to see that sociohistorically framed research has yet to flourish. Sociohistorical studies consider the power relations that juxtapose minoritized groups against a dominant majority (Waitoller et al., 2010), and sociohistorical frames may prove important in reconceptualizing how structural spaces and interactional spaces change over time. As O’Connor (2020) noted, “the past is necessarily present, even if rendered invisible” (p. 475). It is important to recognize that historically created conditions of inequality, such as housing segregation (Rothstein, 2017) and environmental injustices (Akom, 2011), cannot easily be explicitly accounted for in most sociodemographic studies. If the higher likelihood of disability classification for a particular group is decoupled from historical and political contexts, then individual learning capacities are being erroneously linked to racially charged cultural traits, rendering a deficit view of students (see Valencia, 2012). Thus, while sociodemographic studies elucidate the role that context plays in shaping special education outcomes, they are largely unable to take into account the sociohistorical and sociocultural elements that inform conceptions of students, race, and ability, which in turn shape policy and practice (Artiles, 2013).
Future studies should continue to examine the historical connections between race, class, and learning disability classifications (e.g., Eitle, 2002; Ong-Dean, 2006; Wiley et al., 2013; see Sleeter, 2010, for information on historical shifts in the construction of disability). Given the shared methodologies used in contemporary sociohistorical studies and sociodemographic studies, it may prove relatively easy to bridge the gaps between the two frames. Future research can engage in multisite and multilevel ethnographies to document not just constitutive actions but also their connection to historical and political contexts (O’Connor, 2020). This may include discussions about intersectional oppressions and historical considerations of the intersections between race, disability, and disproportionality (see Annamma et al., 2013; Annamma et al., 2018; Artiles, 2013). It can also include comparative studies that look at the same phenomena across multiple contexts (e.g., Kiru & Cooc, 2017; Kramarczuk Voulgarides, 2018). As scholars come to better understand the role that context plays in special education classifications, there may be a melding of sociodemographic and professional practice studies with sociohistorical studies and a concomitant broadening of conceptual frameworks, variable choices, and research designs to more effectively include the sociocultural and historical conditions of local contexts.
Expand Professional Practice Research
Our framework creates a space for examining the prevalence and mechanisms of bias in the existing literature base. Collectively, professional practice studies provide much-needed insights into the contextualized nature of disproportionality. However, they are limited. Because of the various potential sources of disproportionality within schools, it is difficult to capture the entirety of the special education classification process. Moreover, professional practice research is less equipped to systematically explore the role that context plays in special education classifications in the same way as sociodemographic studies. As a result, research on professional practices may contribute to myopic perspectives of disproportionality, in which we look for localized and technical solutions that either ignore structural racism and inequities or leave unexplored the experiences of minoritized students and their families. However, professional practice–oriented researchers can understand the broader influence of structure that shapes policies, practices, and beliefs. Notably, Cooc’s (2018b) study of teacher perceptions of disability provides an example of what might be a helpful direction forward. Furthermore, Fish (2017) calls for the need to “examine the contexts in which race/ethnicity becomes salient to exceptionality construction” (p. 331). This approach seeks to push the boundaries of agency-oriented research into shared spaces.
Reconceptualize Variables in Sociodemographic Research
The body of research on special education disproportionality is ever expanding. From a sociodemographic perspective, researchers have focused on a range of questions related to disproportionality, describing the extent to which a student’s race (e.g., Sullivan & Artiles, 2011; Zhang et al., 2014), linguistic background (e.g., Estrem & Zhang, 2010; Sullivan, 2011; Umansky et al., 2017), nation of origin (e.g., Cooc, 2019), family immigration history (e.g., Hibel & Jasper, 2012), and socioeconomic background (e.g., Kincaid & Sullivan, 2017) are predictive of being classified as disabled, both in special education generally (e.g., Anderson et al., 2015) and in specific disability categories (e.g., Bal et al., 2019; Dickerson et al., 2017; Mandell et al., 2008; Mandell et al., 2009; Morgan, Farkas, Hillemeier, Li, et al., 2017; Morrier & Hess, 2012; Robinson & Norton, 2019; Travers et al., 2011). There is also an emerging list of studies examining the differential rates in access to early-childhood interventions and supports (e.g., Mann et al., 2007; Morgan et al., 2012; Morrier & Gallagher, 2012).
At the same time, sociodemographically oriented research must contend with the explanatory limits concerning the role that individual actors and the enactment of policies, practices, and beliefs play in the special education classification process. Sociodemographic studies are often limited by the availability and format of data. They tend to rely on large administrative data sets rather than data sets that are purposefully constructed to study disproportionality. As such, researchers in this domain may have reached a limit with respect to their capacity to source new data with unique variables. In this regard, they may need to explicitly collect and analyze data that speak to both the end result—a special education classification—and the special education classification process when attempting to make claims about disproportionality, and reconsider how they conceptualize such variables in their studies.
The use of achievement data is a good place to start. For example, while students’ academic achievement tends to be constant over time (Wu et al., 2014), it is important to remember how both structural and agency-focused studies show that academic performance, socio-emotional well-being, discipline outcomes, and the influence of poverty—individual student characteristics commonly associated with predicting student ability—are malleable and can be influenced by effective school-based programs and other teacher and classroom practices, without the need for special education classification (Bardon et al., 2008; Decker et al., 2007; Lo & Cartledge, 2006; Muschkin et al., 2015; Temple et al., 2010). Moreover, as ecological models show, student ability may similarly be influenced by racial contexts (Benner & Crosnoe, 2011; Benner et al., 2008; Eitle & Eitle, 2004; Ong-Dean, 2006). If researchers view academic achievement as malleable, it may be more difficult to fully argue that studies are comparing students “similar in their clinical need for services” (Morgan, Farkas, Hillemeier, & Maczuga, 2017, p. 307) when making claims about the role of race in special education; two students who are struggling academically may come by their low test scores in different ways, and as such, academic achievement is not a valid predictor of disability. Achievement scores, while commonly used in recent studies of special education disproportionality, would not be used to differentiate between students with disabilities and students who simply struggle academically for other reasons, or what Fish (2019a) describes as “uncategorized, nonmedicalized low performance” (p. 2576). Moreover, by privileging a view of disability as an artifact found within an individual, these studies are unable to explain why and how structural factors operate to produce special education outcomes (Artiles et al., 2010) and how people within their local contexts resist and (re)produce them.
Final Thoughts
While there is potential for researchers to transcend agency–structure dualism and come to a shared definition of disability, as we have noted, many of these restrictions serve a useful purpose. We also acknowledge that it may be difficult to fully bridge the divide, as researchers specialize in specific approaches and disciplines. However, we are hopeful that scholars can effectively stretch their intellectual and analytic capacities, position their work in a manner that captures newer and dualism-collapsing sources of data (or at the very least, interpret their findings relative to this dualism), and thus push the boundaries of agency- and structure-based research. Studies that seek to produce high-quality evidence around disproportionality should, by virtue of their design, be attentive to the whole field. Moreover, we should not be looking to a single set of studies to provide a definitive answer. Instead, we should be working to build more robust and inclusive paradigms to understand the issue. This requires the field to embrace multiple levels and multiple methods of analysis. It may not be possible to attend to both structural and interactional spaces in the same study, but scholars in either space should attempt to situate their studies within this broader organizing framework for understanding disproportionality.
