738761 research-article2017

EDRXXX10.3102/0013189X17738761Educational ResearcherEducational Researcher

Feature Articles

Who Feels Included in School? Examining Feelings of Inclusion Among Students With Disabilities Leanna Stiefel1, Menbere Shiferaw1, Amy Ellen Schwartz2, and Michael Gottfried3

The passage of landmark federal legislation in 1975 guaranteed students with disabilities (SWDs) a free appropriate public education. Over time, reauthorization of this legislation has highlighted the importance of educating SWDs in the general education environment where appropriate, and significant interaction with general education peers who are not receiving special education services (GENs) has followed. The largest U.S. school district—New York City (NYC), the setting of our study—mirrors this national trend toward more inclusive environments. Despite the trend, a critical question remains: Does inclusion in fact feel inclusive? This study provides a district-wide descriptive analysis of feelings of inclusion among SWDs compared to their GEN peers. We rely on detailed, longitudinal administrative and student survey data on approximately 249,000 NYC middle school students attending schools that educate both SWDs and GENs (which we call traditional schools in contrast with schools that educate only SWDs). Our results suggest that while gaps between SWDs overall and GENs are quite small, there is some heterogeneity. Specifically, students with an emotional disturbance (ED) and other health impairments (OH) feel modestly less included with peers but more with adults. There are, however, almost no differences in feelings of inclusion between students assigned exclusive and inclusive services, even for those in the ED and OH groups. In fact, for students with low incidence (LI) disabilities, feelings of inclusion are slightly better when assigned exclusive services.

Keywords: correlational analysis; descriptive analysis; econometric analysis; education policy; learning environments; longitudinal studies; mainstreaming; middle schools; regression analyses; secondary data analysis; special education

T

he passage of landmark federal legislation in 1975 guaranteed students with disabilities (SWDs) a free appropriate public education.1 Over time, reauthorization of this legislation has highlighted the importance of educating SWDs in the general education environment where appropriate, and significant interaction with general education peers who are not receiving special education services (GENs) has followed. For example, by 2014, 62.6% of SWDs nationwide “were educated inside the regular class 80% or more of the day,” an increase of 16.9 percentage points from 1996 (U.S. Department of Education, 2007, 2015). The largest U.S. school district—New York City (NYC), the setting of our study—mirrors the nation, with 60.5% educated inside the regular class in 2014 (NYSED, 2017). Moreover, between 2006 and 2012 in NYC, the percentage of SWDs assigned to self-contained services declined while the percentage assigned to services in the general education environment increased, by 9 percentage points (Stiefel, Shiferaw, Schwartz, & Gottfried, 2017). Educational Researcher, Vol. XX No. X, pp. 1­–16 DOI: 10.3102/0013189X17738761 © 2017 AERA. http://edr.aera.net

Although the delivery of SWD education services involving considerable interaction with GENs (labeled inclusive services hereafter) is a national and NYC trend, there is little empirical evidence supporting the efficacy of the practice. There is, however, some evidence that special education services in general can improve achievement for SWDs (e.g., Cohen, 2007; Hanushek, Kain, & Rivkin, 2002). Perhaps most surprising is the dearth of evidence on whether students educated with inclusive services feel included—important itself and for its effect on academic success (Anderman, 2002). We fill this gap by providing a rich descriptive examination of four questions: Research Question 1: How do SWDs and GENs perceive feeling included in schools that educate both SWDs and 1

New York University, New York, NY Syracuse University, Syracuse, NY 3 University of California, Santa Barbara, Santa Barbara, CA 2

MONTH XXXX    1

GENs (hereafter traditional schools in contrast to schools that educate only SWDs)?2 Research Question 2: Do SWDs’ perceptions differ across major disability classifications? Research Question 3: Are there differences between assigned services that are inclusive rather than exclusive? Research Question 4: Since low-performing students may perceive the school environment differently, to what extent are results driven by differences in students’ academic achievement? Answering these helps policymakers and practitioners understand patterns of responses to the inclusive environment and sheds light on whether steps need to be taken to make inclusion in fact feel more inclusive. We are in a unique position to examine these questions for several reasons. First, NYC public schools have large populations of both SWDs and GENs, increasing the ability to detect nuanced relationships, obscured in smaller data sets. Second, NYC’s heterogeneous students and neighborhoods make it a useful setting for policy and practice that applies to many different kinds of districts, although its exceptional size and complex political environment might also mean that findings are likely more narrowly relevant only for the largest U.S. districts. Third, our data contain rich detail on disability and assigned special education services that allow us to parse more finely the single aggregate category “SWD.” Finally, we have student responses to survey questions about the school environment for both SWDs and GENs. This is particularly important because we can determine how students with different disabilities and assigned different services feel about inclusion in their schools both overall and in comparison to GEN peers.3 We use these data to estimate descriptive regression models that examine SWDs’ feelings of inclusion compared to GENs. Background Inclusion of SWDs in the general education environment has a clear appeal. While much of the literature on this topic was published over 15 years ago, when opportunities for using causal methods were more limited, the correlational findings were generally positive.4 For example, inclusion can offer SWDs greater access to the general education curriculum, which has been linked to increased academic success (Baker, Wang, & Walberg, 1995). Existing evidence supports the notion that inclusion could also boost social development as SWDs may feel a better sense of school belonging and acceptance rather than feeling labeled and isolated, which often results from little (or no) interaction with GENs (Ballard, Corman, Gottlieb, & Kaufman, 1977; Fitch, 1999, 2003; Schnorr, 1990). Moreover, inclusion may result in some instruction in classrooms that offer two teachers, potentially benefitting both SWDs and GENs (Schwab & Gelfman, 2005). That said, many education stakeholders question whether inclusion adequately addresses the needs of SWDs (Moon, Todd, Morton, & Ivey, 2012; Rule, Stefanich, Haselhuhn, & Peiffer, 2009; Supalo, Mallouk, Rankel, Armorosi, & Graybill, 2008). For example, general education teachers are often underprepared to 2    EDUCATIONAL RESEARCHER

educate SWDs (Kauffman & Hallahan, 1993; Moon et al., 2012), there may be a lack of necessary individualization within larger classes (Sandler, 1999), or schools may be unequipped to address SWD needs adequately (Supalo et al., 2008). Additionally, because students rely on classmates to make comparisons and develop selfconfidence (Huguet et al., 2009), SWDs might have lower selfesteem if they compare themselves to higher performing or more socially adjusted GENs (Freeman & Alkin, 2000). There also may be problematic effects for GENs. For instance, some SWDs may be disruptive at school (Fletcher, 2010), slowing the pace of instruction as teachers attend to management (Gottfried, 2014; Lazear, 2001). Finally, increasing diversity might lead to hostility (Banks & Banks, 1995). Despite these concerns, quantitative research on the effects of inclusion is scarce, focuses on academic outcomes, and reports mixed results. For SWDs, some studies find positive effects of inclusion on academic outcomes (Jepma, 2003; Markussen, 2004; Myklebust, 2007; Peetsma, Vergeer, Roeleveld, & Karsten, 2001; Rogers & Thiery, 2003; Rojewski, Lee, & Gregg, 2015; Waldron & McLeskey, 1998), while a few show negative or null findings (Cole, Waldron, & Majd, 2004). For GENs, the literature is also mixed. Fletcher (2009, 2010) finds that having classmates with emotional and behavioral disabilities lowers math test scores. Friesen, Hickey, and Krauth (2010) and Keslair, Maurin, and McNally (2012) find null peer effects of SWDs. Gottfried and Harven (2015) and Gottfried, Egalite, and Kirksey (2016) find that GENs perform worse with SWD classmates, although this effect is moderated by the percentage of female classmates. Hanushek et al. (2002), however, report a higher proportion of SWDs in a grade raises GEN math achievement. Much less is known about how inclusion influences SWDs’ socioemotional well-being, in particular perceptions of inclusion. For GENs, feeling included in school predicts multiple indicators of academic success. For instance, researchers have found that feeling included predicts academic self-efficacy (Uwah, McMahon, & Furlow, 2008), achievement (Anderman, 2002), school completion (Finn, 1989), expectancy of success (Goodenow & Grady, 1993), and transition into postsecondary education (Pittman & Richmond, 2007). But little is known about these patterns for SWDs in general education settings. We believe this oversight is problematic because SWDs are less likely to be socially integrated with peers, have poorer relationships with teachers, and are more likely to be socially isolated than GENs (Biordi & Nicholson, 2012; Murray & Greenberg, 2001; Pearl et al., 1998; Turner & Noh, 1988; Van Gundy & Schieman, 2001). These issues of feeling included become especially poignant in the context of specific assigned services and disabilities. On the one hand, as SWDs receive more inclusive services, SWDs and GENs spend more time together and may feel greater similarity (Freeman & Alkin, 2000) and exhibit improved interpersonal skills and acceptance of individual differences (Gottfried, 2014; Williams & Downing, 1998). On the other hand, inclusion might inhibit feeling included. Pulling SWDs out of general education classrooms to receive services may increase feelings of dissimilarity, decrease participation in curricular activities with GENs, and diminish feelings of acceptance from both groups (Gottlieb, 1981; Schnorr, 1990). Moreover, it is likely that the type of disability plays a role. SWDs with emotional disabilities

involving aggression or disruptiveness may feel more rejected by their peers (Coie, Dodge, & Kupersmidt, 1990; Siperstein & Leffert, 1997), which might be exacerbated with increased time spent with GENs. In this case, feelings of inclusion may decline (Freeman & Alkin, 2000; Kozleski & Jackson, 1993). Other SWDs may experience more (or less) positive peer attitudes (Laws & Kelly, 2005; Nowicki, 2006), low self-esteem, or teasing, for example. Hence, the relationship is complex, requiring detailed attention to service assignment and disability. Issues pertaining to feeling included are especially critical for middle school students—the focus of our study. Students first feel self-consciousness and make social comparisons in middle school (Nicholls, 1990) and feel a greater need for positive and supportive peer and nonparental adult relationships than earlier (Anderman, 2003; Blum, McNeely, & Rinehart, 2002). In middle school, students become more aware of and vulnerable to school social climate (Anderman, 2003; Wang & Holcombe, 2010) and may begin to feel either more included in or alienated from school—feelings that continue into high school (Rumberger, 1995). Young adolescents’ sense of belonging and inclusion has significant ramifications for academic and psychological development (Anderman & Freeman, 2004; Resnick et al., 1997). Therefore, understanding how inclusion contributes to these feelings is important, particularly during a significant transitional period like middle school. It is even more poignant for SWDs facing the aforementioned difficulties in establishing and maintaining social relationships. In summary, although SWDs are increasingly included with GENs in school, almost nothing is known about whether inclusion is linked to feeling included. Because (a) inclusive practices are increasingly prevalent in our nation’s schools and (b) feeling included is important for short- and long-term social and educational success, it behooves researchers to determine to what extent students do in fact feel included. Understanding the processes that facilitate feeling included in school might be effective in increasing educational success for all. To address these issues, we provide rich descriptive evidence from multiple regression models that associate SWD status with measures of feeling included. We probe the associations more deeply by distinguishing SWDs by disability classification and comparing feelings for students receiving exclusive compared to inclusive services.

professionally evaluated, assigned a disability classification (among the 13 federal disability classifications), and where appropriate, an Individualized Education Program (IEP) is developed. For our study, SWDs include all students with an IEP; GENs are all others. We group the 13 disability classifications into the five largest categories in our sample: learning disability (LD, 60.5% of middle school SWDs in our 2012 sample), speech impairment (SI, 26.3%), other health impairment (OH, 7.1%), emotional disturbance (ED, 2.6%), and low incidence (LI, 3.5%), which includes deafness, deaf-blindness, visual impairment, hearing impairment, intellectual disability, multiple handicap, autism, traumatic brain injury, and orthopedic impairment. Shogren, Kennedy, Dowsett, and Little (2014) grouped high-incidence disabilities (learning, emotional, speech, or language impairments and other health impairments) separately from low-incidence groups. Their taxonomy is similar to ours and also demonstrates precedence for collapsing federal categories into analytically manageable groups. Following the information in the IEP, NYCDOE identifies the primary assigned service as the service (of four possible) that is assigned to the SWD for the largest proportion of time, which can be less than 50%. The NYCDOE ranks these from the “least to most restrictive environment” as follows (NYCDOE, n.d.): related services only (RS, 9.1% of SWDs in our 2012 sample), special education teacher support services (SETSS, 40.7%), integrated co-teaching (ICT, 45.0%), and self-contained (SC, 5.3%). See Appendix Table A4 for the percentage of students by disability classification and assigned service for our sample. RS include services such as counseling, physical therapy, and hearing education. Students are assigned RS as their primary service if it is their only service. SETSS take place in general education classrooms and provide pullout or in-class selective services to support SWDs or indirect services to foster collaboration between special education and general education teachers. ICT services take place in classrooms with both general and special education teachers in classes with both GENs and SWDs. SC services are provided to SWDs grouped by needs.6 We create a dummy variable, exclusive, that takes a value of one if a SWD is assigned to SC (the least inclusive service) and zero otherwise (that is, assigned to an inclusive service—RS, SETSS, or ICT).7

Methods

Other student measures. Student sociodemographic characteristics and performance levels include gender, age, race/ethnicity, nativity status, Limited English Proficiency (LEP), home language, free/reduced price lunch eligibility, and proficiency levels on lagged fifth-grade English language arts (ELA) and math examinations. Data include school and grade IDs. In Table 1, we present descriptive statistics for the most recent year of data, 2012, for our analytic sample of students in traditional schools. Recall we do not include students in D75. SWDs make up roughly 11% of the sample and differ from GENs markedly; they are disproportionally male, Hispanic, LEP, native-born, and likely to score below proficiency on math and ELA tests. The majority of SWDs (95%) are assigned inclusive services (RS, ICT, or SETSS). SWDs with each disability are represented in all four assigned

Data and Measures Overall sample. We use administrative and survey data from the New York City Department of Education (NYCDOE). The final sample comprises 248,951 students attending 502 traditional public schools serving Grades 6 through 8 between 2007 and 2012.5 We exclude SWDs attending schools in the special district (D75) serving students with severe disabilities, although results are robust to including them. SWD sample and measures. We identify disability status, classification, and primary assigned service of each student. During the special education assessment process, each student is

MONTH XXXX    3

Table 1 Descriptive Statistics for Sixth- Through EighthGrade Students With Disabilities and General Education Students in NYC Traditional Public Schools, 2012

Female White Black Hispanic Asian and other Home language is English Limited English Proficiency (LEP) Foreign-born Free/reduced lunch Inclusive services Exclusive services Learning disability (LD) Speech impairment (SI) Other health impairment (OH) Emotional disturbance (ED) Low incidence disability (LI) Changed service Changed disability Below proficient in math Below proficient in ELA Average age N of students Percentage of sample

GEN (%)

SWD (%)

53.5 16.2 25.1 38.4 20.3 53.9 8.5 18.3 82.3

38.1 15.8 26.6 49.0 8.6 62.1 15.3 8.9 86.2 94.7 5.3 60.5 26.3 7.1 2.6 3.5 2.2 5.4 43.6 61.3 14.2 8,830 10.5

16.8 25.3 13.9 75,667 89.6

Note. All GEN-SWD differences are statistically significant except for the percentage of students who are White. Proficiency in math and English Language Arts (ELA) are defined by New York State. Below proficient is an indicator for students performing below the standards for their grade. Math and ELA proficiency levels are measured in the fifth grade. NYC = New York City; GEN = general education peers who are not receiving special education services; SWD = students with disabilities.

services, although SETSS and ICT are most common. Relative to other classifications, a larger share of students with emotional disturbances are assigned SC services. Outcomes: Feeling included. In 2006–2007, NYC launched an annual survey of students’ responses to questions concerning their learning environment (Learning Environment Survey; NYCLES). The largest survey of its kind in the United States, it provides unusually rich information on 6th- through 12thgrade students with over 60 questions in four broad reporting categories: academic expectations, communication, engagement, and safety and respect. Response rates increased steadily since first implementation, from roughly 60% in 2007 to 90% in 2012 in our sample. We account for varying response rates across questions and students by weighting all models using the inverse of the predicted probability of response (Seaman & White, 2013; Wooldridge, 2002). Results are robust to excluding 2007, which has relatively low survey response rates. See Appendix Table A5. 4    EDUCATIONAL RESEARCHER

To be sure, “feeling included” is a multidimensional aspect of the learning environment, reflecting engagement, safety and respect, and communication. Five questions that span these three domains of the learning environment cover aspects of the school environment related to feeling welcome (welcome), bullying (bully), harassment (harass), being known (known), and SWD inclusion in school activities (included). As shown in Table 2, the response to each question is either on an agreement scale (strongly agree, agree, disagree, or strongly disagree) or frequency scale (all of the time, most of the time, some of the time, or never). For each, we construct a binary indicator taking a value of one if the student agrees (strongly agree/agree) or the event occurs more frequently (all of the time/most of the time). Other authors also have used binary indicators (Cannon, Jacknowitz, & Painter, 2006) and/or analyzed other ways of categorizing such variables and have found that all the “empirical action” is captured by the binary representation (Gibbons & Silva, 2011).

Empirical Framework To provide descriptive evidence on Research Question 1 (How do SWDs and GENs perceive feeling included in traditional schools?), we first estimate a baseline linear probability model linking feeling included to SWD status: Yigst = β0 + δ SWDigst + θ g + τt + εigst , (1) where igst denotes student i, in grade g, attending school s, in year t. The outcome Y represents a dichotomous response as described previously. θ and t are grade and year effects, respectively, capturing grade-specific idiosyncrasies or common factors in each year such as systemwide accountability policies. In all models, standard errors are clustered at the school by year level. This is consistent with later models that include school by year fixed effects. The reference group is GENs so that δ captures conditional differences in feeling included between SWDs and GENs. Estimation of logit models yields similar results to linear probability models. The next models deepen the analysis by adding a set of student characteristics (X) encompassing gender, age, race/ethnicity, nativity status, LEP, home language, free/reduced price lunch eligibility, and proficiency levels on lagged fifth-grade ELA and math examinations. We substitute school by year fixed effects (πst) for year effects to capture school-specific differences, such as principal leadership, which may change over time: ′ Yigst = β0 + δ SWDigst + X igst β + θ g + πst + εigst . (2)

To answer Research Question 2 (Are SWD reports about feeling included heterogeneous?), we replace SWD with a set of disability indicators, disability, one for each of the five categories described previously: ′ Yigst = β0 + disability ′igst δ + X igst β + θ g + πst + εigst . (3)

Here our coefficients of interest, δ, capture the disparity in feelings of inclusion between SWDs in each disability classification and GENs, accounting for differences in student characteristics and differences across grades and schools by years.

Table 2 Survey Questions That Capture Feelings of Inclusion Questions

Variable Name

1. I feel welcome in my school. 2. Students threaten or bully other students at school. 3. Students in my school are harassed or threatened based on race, color, creed, ethnicity, national origin, citizenship/immigration status, religion, gender, gender identity, gender expression, sexual orientation, or disability. 4. Most of the teachers, counselors, school leaders, and other adults I see at school every day know my name or who I am. 5. Students with disabilities are included in all school activities. (2010–2012 only)

If Student Responds

Welcome = 1 Bully = 1 Harass = 1

Strongly agree/agree All/most of the time All/most of the time

Known = 1

Strongly agree/agree

Included = 1

Strongly agree/agree

Note. Included was only asked in the survey between 2010 and 2012.

Third, to address Research Question 3 (Are there differences in perceptions about feeling included by assigned service?), we estimate a model that includes the interaction between each of the disability indicators and an indicator for being assigned to an exclusive service (disability × EXCL): ′ Yigst = β0 + disabilityigst δ1 + ( disability × EXCL ) ′igst δ2 + ′ X igst β + θ g + πst + εigst .

(4)

As in Equation 3, components of disability equal one for a student with a specific disability and zero otherwise while disability × EXCL equals one if a student with a specific disability is served with an exclusive service and zero otherwise. The δ1 represent the difference in feeling included between SWDs (in inclusive services) and GENs. The δ2 capture the difference between those in exclusive and inclusive services, within disability classifications. Put differently, the δ2 capture the extent to which students with exclusive services report different feelings of inclusion than otherwise similar students receiving inclusive services. Although our sample includes repeated observations for students (either two years or three years), we cannot implement student fixed effects models because too few students change disability classification (4.8%) or service (1.9%). Finally, to address Research Question 4 (To what extent are differences driven by academic performance?), we reestimate Equation 4 including lagged proficiency levels. While proficiency could be related to feelings of inclusion, it may also indicate the need for special education services. Moreover, feelings of inclusion could affect test scores or vice versa. Disentangling these is complicated, and including a control for proficiency in the model is no doubt not fully satisfying. Nonetheless, since recent literature suggests such controls can change the size and signs of important coefficients (Morgan et al., 2015), we report results with lagged, fifth-grade proficiency levels. Results are similar when we include contemporary proficiency levels or test scores. Results

Feeling Included Question 1 asks how SWDs as a whole and GENs perceive feeling included in traditional schools. Results are presented in Figure 1, which shows SWD and GEN responses to each survey question, adjusting for grade and year. All differences between SWDs and GENs are statistically significant. As shown, fairly

high percentages of both SWDs and GENs feel welcome and known, somewhat lower percentages of both feel included, and significant percentages of both report a likelihood of bullying and harassment. Importantly, however, compared to GENs, SWDs are only modestly less likely to report feeling welcome, somewhat more likely to report bullying, and somewhat more likely to report harassment at school. Also, compared to GENs, SWDs are somewhat more likely to report that adults know their name and somewhat more likely to report being included in school activities. All differences are between 1.7 and 6.3 percentage points (ppts). While these differences are small, they may be meaningful for the most vulnerable students who are attempting to navigate their environment as young teens. Subsequent results will also illustrate that averaging across all SWDs masks notable disparities by disability type. To further address Question 1, we next make adjustments to estimate the gap between SWD and GEN feelings of inclusion. From here on, we report only the size of the gap, having shown the absolute magnitudes in Figure 1. We incorporate controls for student characteristics and school by year effects. Figure 2 presents the estimated δ, the SWD-GEN difference. Again, all differences represented by all bars are statistically significant. The grey bars show results from Figure 1 but as gaps, for comparison (Model 1). The black bars show results for Model 2, with student characteristics and school by year effects included. The modest SWD-GEN disparity persists, with the gaps somewhat attenuated. For example, now SWDs are 2.1 ppts more likely than GENs to report bullying and 3.8 ppts more likely to feel they are known. Question 2 asks if SWD perceptions are heterogeneous, differing by major disability. Results are presented in Figure 3, which disaggregates SWDs into five major classifications, still including student characteristics and school by year effects (Model 3). There is heterogeneity across SWDs with different disability classifications. Except for Known, students with EDs and OHs feel least included, with the largest (but still modest) gaps across all classifications of SWDs in Included. For example, the ED-GEN gap is 3.9 ppts in Welcome and a 4.2 ppt gap in Included. Gaps for students with LDs and SIs are a little less consistent than for students with EDs and OHs; while there is a negative gap with GENs for Welcome, Bully, and Harass for included, students with LDs and SIs feel more included than GENs. As for all classifications, students with LDs and SIs feel more known than GENs. Finally, students in the LI MONTH XXXX    5

Figure 1. Do students with disabilities (SWDs) feel differently from students not receiving special education services (GENs)? Baseline average responses Data are student-level, administrative data from the New York City Department of Education. Baseline results are adjusted for year and grade. All SWD-GEN differences are statistically significant at the 1% significance level. All estimates are weighted by the inverse of predicted probability of response.

Figure 2. Are differences driven by student and school by year characteristics? All SWD-GEN differences are statistically significant at the 1% significance level. Standard errors are corrected for school by year clusters. Bars represent estimates from linear probability models that link students’ responses to each question to SWD status, adjusting for just year and grade in the baseline model (grey bar), then adjusting for students’ characteristics and school by year effects (black bar). Student characteristics include gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/reduced lunch eligibility, age, in the sample for two years only, and indicators for students who ever changed disability classification or service setting. All estimates are weighted by the inverse of predicted probability of response. SWD = students with disabilities; GEN = students not receiving special education services. disability group feel more included than other SWDs. The gap with GENs shows more positive or similar feelings in all cases except Harass, which is negative but only significant at the 10% level. Thus, feelings of inclusion vary in both intensity and consistency across SWDs with different disability conditions. Overall, students with EDs and OHs most consistently feel somewhat less included.

The Role of Assigned Service Question 3 asks if there are differences in feelings of inclusion by assigned service. Table 3 presents results; the reference group is

6    EDUCATIONAL RESEARCHER

GENs. All models include student characteristics and grade and school by year fixed effects. The coefficients on EXCL variables, which capture differences between feelings of inclusion among students assigned exclusive compared to inclusive services, are largely insignificant— that is, there are few differences between students in exclusive and inclusive services. Recall that most students in the traditional schools in this study are assigned to inclusive services, so the sample assigned to exclusive services is not large. Only for students in the LI group is there some indication of differences, and these students feel more included (more welcome, less

Figure 3. Heterogeneity in SWD-GEN differences by disability classification Standard errors are corrected for school by year clusters. Bars represent estimates from linear probability models that link students’ responses to each question to a student’s disability classification, adjusting for grade effects, school by year effects, and the following student characteristics: gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/reduced lunch eligibility, age, in the sample for two years only, and indicators for students who ever changed disability classification or service setting. The SWD-GEN disparities are statistically different across the five disability categories at the 1% significance level. All estimates are weighted by the inverse of predicted probability of response. SWD = students with disabilities; GEN = students not receiving special education services. † Not statistically significant from GENs. bullying, and more known) compared to GENs when served with exclusive compared to inclusive services.

The Role of Academic Achievement Question 4 asks to what extent the results are driven by differences in students’ academic achievement. Results are presented in Table 4. For students in the ED, OH, and LI groups, there are essentially no differences when proficiency levels based on test scores are added to the model. For students with LDs and SIs, some coefficients are no longer statistically significant, although all signs are the same, and thus controlling for performance seems to affect them somewhat. In general, from Tables 3 and 4, the most consistent SWD-GEN disparities across all five outcomes arise among students with EDs, OHs, and LIs. These results for students with EDs are consistent with prior research, which has shown that students with EDs are among the most challenging to incorporate into more inclusive settings (MacMillan, Greshman, & Fornes, 1996). Discussion In this study, we examine how feelings of inclusion differ for SWDs and GENs in traditional schools. This is a critical issue to address for several key reasons. First, little research has focused on inclusion for both SWDs and GENs. Our study looks at responses of both groups to questions about feelings of being included. Second, most research has focused on inclusion and

academic outcomes while our study goes beyond testing. Finally, little research examines differences by disability within a single study on inclusion (for an exception, see Gottfried, 2014); we are the first to examine feeling included by disability and by service, made possible by NYC’s unique, large administrative data set with rich detail on disability and service. We derive several key insights. First, among students attending traditional schools, SWDs as a whole feel only slightly less included with classmates (Welcome, Bully, Harass) compared to GENs but somewhat more included with teachers and in school activities (Known, Included). These may reflect educators’ particular efforts to address the education needs of SWDs and draw them into school activities. Sharing classes with GEN peers, however, might make SWDs feel singled out among classmates. This could occur because SWDs receive different services compared to GENs (e.g., having another teacher in the classroom or receiving pull-out services), or this may arise due to observable differences, such as physical impairments, compared to GEN classmates. Together, our results suggest that SWDs in general feel only modestly less included with their classmates but are somewhat more likely to report favorable feelings with regard to teachers, especially feeling known. Second, our findings regarding feelings about classmates correspond with prior research, which has found that SWDs often have poorer peer relationships compared to GENs (Biordi & Nicholson, 2012; Pearl et al., 1998). Our study demonstrates this in findings related to the modest gaps in Welcome, Bully, and Harass as described previously. Thus, it suggests schools and policymakers would do MONTH XXXX    7

Table 3 Disparities by Disability and Assigned Service

  Inclusive  ED    OH    LD    SI    LI   Exclusive-inclusive disparity   ED × EXCL     OH × EXCL     LD × EXCL     SI × EXCL     LI × EXCL   School by year fixed effects Constant   N students N schools N observations

(1)

(2)

(3)

(4)

(5)

Welcome

Bully

Harass

Known

Included (2010–2012 Only)

–0.054*** (0.011) –0.029*** (0.006) –0.013*** (0.002) –0.007** (0.003) –0.001 (0.007)

0.059*** (0.013) 0.048*** (0.007) 0.015*** (0.003) 0.024*** (0.004) 0.020* (0.011)

0.060*** (0.013) 0.016** (0.007) 0.008*** (0.002) 0.012*** (0.004) –0.004 (0.010)

0.084*** (0.010) 0.079*** (0.005) 0.032*** (0.002) 0.029*** (0.003) 0.094*** (0.008)

–0.046** (0.018) –0.021** (0.010) 0.021*** (0.004) –0.001 (0.005) 0.045*** (0.016)

0.026 (0.030) –0.017 (0.030) –0.026 (0.016) –0.017 (0.020) 0.072*** (0.025) Y 0.933*** (0.012) 248,951 502 610,273

–0.025 (0.040) –0.010 (0.048) 0.013 (0.019) 0.040 (0.027) –0.098* (0.052) Y 0.400*** (0.015) 248,951 502 610,273

–0.006 (0.034) –0.017 (0.037) –0.006 (0.017) 0.057** (0.025) –0.050 (0.041) Y 0.192*** (0.014) 248,951 502 610,273

0.022 (0.028) 0.004 (0.025) 0.007 (0.016) 0.005 (0.022) 0.032 (0.023) Y 0.611*** (0.013) 248,951 502 610,273

0.041 (0.056) 0.003 (0.058) 0.002 (0.030) 0.051 (0.037) 0.231*** (0.066) Y 0.359*** (0.023) 143,405 471 324,914

Note. Robust standard errors clustered by school by year in parentheses. All models include grade and school by year effects and the following student characteristics: gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/reduced lunch eligibility, age, in the sample for two years only, and indicators for students who ever switched disability classification or service setting. Coefficients on the inclusive indicators represent the SWD-GEN disparity for SWDs assigned to inclusive services (RS, ICT, and SETSS). Coefficients on the interaction terms represent the exclusive-inclusive disparity. The SWD-GEN disparity for SWDs assigned to exclusive services (SC) can be obtained by summing the coefficients on inclusive and the interaction. Inclusive is set to zero for all GENs. All SWD-GEN disparities by disability classification are statistically different from one another, for SWDs receiving inclusive services, and likewise for SWDs receiving exclusive services across all outcomes (F tests). All models are weighted by the inverse of predicted probability of response. Included, in Column 5, was only asked in 2010–2012. SWD = students with disabilities; GEN = students not receiving special education services; LD = learning disability; SI = speech impairment; OH = other health impairment; ED = emotional disturbance; LI = low incidence; EXCL = exclusive service; RS = related services only; ICT = integrated co-teaching; SETSS = special education teacher support services; SC = self-contained. *p < .10. **p < .05. ***p < .01.

well to understand how inclusive activities that bring students with individual differences together might at the same time differentially affect feeling included at school versus feeling included with peers. The distinction of what it means to feel included (and with whom) becomes important. Since disparities are small, attention to these differences could result in rapid improvements. The third finding provides insights into mechanisms that might drive our results. The analysis by disability group suggests that some of the strongest and most consistent relationships arise for students with an ED. These students have some of the greatest needs when included with GENs (MacMillan et al., 1996). While the results here do not directly support or dispute that claim, we find that 8    EDUCATIONAL RESEARCHER

students with EDs feel among the least included of all SWDs and suggests a previously undocumented source of risk for this group. Most notably, except for students in the OH category, students with EDs compose the only group that is somewhat less likely to report being included in school activities when in traditional schools. Thus, in considering the previously established challenges with regards to including students with EDs (e.g., disruptive behaviors, academic disengagement), additional challenges might arise in connection with feeling included.8 Given this newly surfaced source of risk faced by students with EDs, we call attention to support systems that might be best suited to improve feelings of inclusion specific to this group. For instance, Gottfried and Harven (2015)

Table 4 Does Academic Performance Matter?

  Inclusive  ED    OH    LD    SI    LI   Exclusive-inclusive disparity   ED × EXCL     OH × EXCL     LD × EXCL     SI × EXCL     LI × EXCL   School by year fixed effects Constant   N students N schools N observations

(1)

(2)

(3)

(4)

(5)

Welcome

Bully

Harass

Known

Included (2010–2012 Only)

–0.052*** (0.011) –0.027*** (0.006) –0.009*** (0.002) –0.004 (0.003) 0.000 (0.007)

0.058*** (0.013) 0.045*** (0.007) 0.011*** (0.003) 0.021*** (0.004) 0.019* (0.011)

0.059*** (0.013) 0.014** (0.007) 0.003 (0.003) 0.008** (0.004) –0.006 (0.010)

0.085*** (0.010) 0.078*** (0.005) 0.030*** (0.002) 0.027*** (0.003) 0.093*** (0.008)

–0.048*** (0.018) –0.023** (0.010) 0.018*** (0.004) –0.004 (0.005) 0.044*** (0.016)

0.027 (0.030) –0.016 (0.030) –0.027* (0.016) –0.017 (0.020) 0.069*** (0.025) Y 0.924*** (0.012) 248,951 502 610,273

–0.024 (0.040) –0.008 (0.048) 0.016 (0.019) 0.042 (0.027) –0.087* (0.052) Y 0.410*** (0.015) 248,951 502 610,273

–0.006 (0.034) –0.016 (0.036) –0.003 (0.017) 0.058** (0.025) –0.042 (0.040) Y 0.202*** (0.014) 248,951 502 610,273

0.023 (0.028) 0.008 (0.025) 0.011 (0.016) 0.008 (0.022) 0.045* (0.023) Y 0.614*** (0.013) 248,951 502 610,273

0.041 (0.056) 0.002 (0.058) 0.002 (0.030) 0.050 (0.037) 0.231*** (0.066) Y 0.364*** (0.023) 143,405 471 324,914

Note. Robust standard errors clustered by school by year in parentheses. All models include grade and school by year effects, an indicator for students who scored below proficient on math and English language arts exams in the fifth grade, and the following student characteristics: gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/reduced lunch eligibility, age, in the sample for two years only, and indicators for students who ever switched disability classification or service setting. Coefficients on the inclusive indicators represent the SWD-GEN disparity for SWDs assigned to inclusive services (RS, ICT, and SETSS). Coefficients on the interaction terms represent the exclusive-inclusive disparity. The SWD-GEN disparity for SWDs assigned to exclusive services (SC) can be obtained by summing the coefficients on inclusive and the interaction. Inclusive is set to zero for all GENs. Inclusive is set to zero for all GENs. All SWD-GEN disparities by disability classification are statistically different from one another, for SWDs receiving inclusive services, and likewise for SWDs receiving exclusive services across all outcomes (F tests). All models are weighted by the inverse of predicted probability of response. Included, in Column 5, was only asked in 2010–2012. SWD = students with disabilities; GEN = students not receiving special education services; LD = learning disability; SI = speech impairment; OH = other health impairment; ED = emotional disturbance; LI = low incidence; EXCL = exclusive service; RS = related services only; ICT = integrated co-teaching; SETSS = special education teacher support services; SC = self-contained. *p < .10. **p < .05. ***p < .01.

found that peer composition was critical for classrooms with EDs when it came to improving testing outcomes, and perhaps consideration of classroom assignment is a mechanism to boost feelings of inclusion for this especially vulnerable group of students. Additional supports might include expanding the role of school counselors, shown to improve student behaviors (Carrell & Hoekstra, 2014; Reback, 2010) and could foster the development of feeling included as well. These are just several of numerous ways for schools to improve outcomes for students with ED vis-à-vis feeling included. Fourth, within traditional schools, we do not find differences in feelings of inclusion between students assigned exclusive and inclusive services. This is important for at least two reasons. First, it

indicates that traditional schools need to pay attention to feelings of SWDs across all services and not simply students receiving the most or least inclusive services. Second, with the exception of students with LIs, it indicates that differences in exclusive and inclusive services may not affect feelings of inclusion. These latter findings, however, need to be tempered by the descriptive nature of the study and the small proportion of students assigned exclusive services. Our design does not yield causal predictions of what would happen if significantly more students were assigned inclusive (or exclusive) services, and importantly, we cannot say what outcomes would result if more students were educated in traditional schools instead of exclusive schools. MONTH XXXX    9

Finally, while the small negative disparities between SWDs and GENs uncovered in this study present some challenges for traditional schools, the size of the gaps are not so large as to discourage productive action.9 Limitations There are several limitations that should spark further research in this area. First, the survey used for this study does not distinguish feelings of inclusion in specific academic subjects. Future research might consider a follow-up survey in a single school site (or perhaps focus groups using qualitative inquiry) to test whether feeling included differs across academic subjects. Second, while we believe that our findings have implications for the nation, our study relied on data from one district and focused on middle school, so additional work exploring generalizability to other contexts is warranted. Although NYC is the largest district in the United States, findings in other districts could differ, so it is important to replicate this work in other districts. Additionally, while we focused on middle school, a critical time in young adolescence for the development of feelings of inclusion, future research might examine other critical periods in the education pipeline, such as the extent to which children with disabilities feel included when transitioning into kindergarten or how high school students feel about inclusion at the point in high school when dropout becomes a risk. Finally, this study does not address the academic performance of SWDs in traditional schools, a complex topic that we hope these analyses will inform in the future. Conclusion There are several concluding implications of our study. To begin, it demonstrates the value of exploring administrative data in conjunction with survey data. Prior research in special education has often been limited to using administrative data (on students and classrooms but lacking detail on nonacademic facets) or national data sets (rich in contextual data but lacking a census of students). Our findings extend the examination in both dimensions—we describe nonacademic outcomes for traditional schools in an entire district, including both SWDs and GENs. Our findings also suggest that perceptions of the school environment have the potential to be responsive to policy changes. For example, this study suggests that teachers and programs might help SWDs feel somewhat more included than do peers. Understanding these mechanisms better will be important moving forward as special education policy continues to implement reforms. Finally, much of prior research has treated SWDs as a single aggregate category. There is richness in the diversity of this group of students, however, and ensuring that we address the needs of students within this population becomes obscured when considering the group in the aggregate. Furthermore, we examine these outcomes not only by disability grouping but also assigned service. This refined level of detail provides insights to schools and policymakers about the imperative to address the needs of SWDs no matter what their service. Understanding patterns at this level will enable multiple layers 10    EDUCATIONAL RESEARCHER

of stakeholders to implement better practices and develop supportive and diverse learning environments. In summary, middle school is both a critical time period for education attainment as well as an important life course stage for identity formation and social development (Williams & Downing, 1998). Our focus on disparities in feelings of inclusion can support education practitioners in their efforts to identify areas that require attention to address the needs of all youth during this important life point. Notes

We thank Clive Belfield, Bryant Hopkins, Agustina Laurito, anonymous referees, participants at the 2015 Association for Public Policy and Management and the 2016 Association for Education Finance and Policy conferences, and participants at the NYU-IESP summer seminar series for helpful comments. Generous support from the Spencer Foundation is gratefully acknowledged. All work is that of the authors alone.   1From 1975 to 1990, this legislation was titled the Education for All Handicapped Children Act (EHA). In 1990, the U.S. Congress changed the title to the Individuals with Disabilities Education Act (IDEA). The legislation specifies that students with disabilities be educated with other children in the regular education environment where appropriate. In 1997, this provision was referred to as the least restrictive environment (LRE) requirement. While the trends we report could be consistent with LRE, LRE requires an individual assessment of appropriate education, and national trends may or may not meet that requirement. Thus, we do not use the term LRE here but simply report the large increases in students “educated inside the regular class 80% or more of the day.”   2We do not include schools serving students with severe disabilities (e.g., severe autism, severe emotional disturbance, intellectual disabilities.) These schools are labeled District 75 (D75) in New York City, and they educate exclusively students with disabilities (SWDs). Results for SWDs do not change if students in D75 schools are included, but we cannot make within-school comparisons with students who do not receive special education services (GENs) as we do for most analyses in this paper. Statistics for all sixth, seventh, and eighth graders, including students in D75, are provided in Appendix Table A1.   3Recall that students in this study are all educated in traditional (not D75) schools and most are assigned to inclusive services, so there are not large samples assigned to exclusive services. Those assigned exclusive services differ in observable characteristics from those assigned inclusive services, and these characteristics are controlled for in our models.   4These findings may even have provided some of the motivation for policymakers to increase inclusionary practices.   5This study focuses on sixth-, seventh-, and eighth-grade students in public schools serving both SWDs and GENs and excluding alternative, charter, and D75 schools. From this sample, we make four exclusions. First, we remove SWDs without a disability classification or receiving services from a non–New York City Department of Education (NYCDOE) facility. Second, we exclude five schools that are extreme outliers in response rates to the survey described in the text. In all, 2,473 students are excluded for these two restrictions, and results are robust to these exclusions. Third, we restrict the sample to students who answer four of the five survey questions in each year in order to work with a consistent sample, excluding an additional 236,375 students. The fifth question was asked only beginning in 2010 and thus has a lower sample size and is not subject to this last exclusion. Finally, we restrict the sample to students who are present at least two of three consecutive grades

to avoid students who might rotate in and out or are present in only one grade, excluding 221,005 students. Appendix Table A2 shows results for Table 3 (discussed later) from adding back the 221,005 students from the final exclusion, who were in the sample fewer than two years. Table A3 shows results when students from both the third and final restriction are added back. Exclusions are not mutually exclusive, and students answer different questions; thus, the number of observations and students differ by column and do not add to the sum of each exclusion. Results remain robust, although a few more coefficients are statistically significant, as expected with a larger sample size.   6In all, 4.8% of students in our sample change disability classification, and 1.9% change services. In our models, we include indicators equal to one if a student has ever changed classification (service) and zero otherwise to account for these changes.   7Without a large number of students assigned to exclusive services, we may not be able to detect statistically significant small differences for such students.   8It is possible that students with an emotional disturbance (EDs) are disproportionately disciplined, which might contribute to their lower feelings of inclusion. We are unable to control for disciplinary incidents, but we note that such a correlation could help explain the larger responses of students with EDs. 9 We thank a reviewer for another way to summarize our results: “In sum, . . . overall, SWDs feel almost as welcome as GENs, with those feeling a bit less welcome being those with disabilities associated with behavior problems. Further, those receiving exclusive services do not feel less welcome.” References

Anderman, E. M. (2002). School effects on psychological outcomes during adolescence. Journal of Educational Psychology, 94(4), 795–809. Anderman, L. H. (2003). Academic and social perceptions as predictors of change in middle school students’ sense of school belonging. The Journal of Experimental Education, 72(1), 5–22. Anderman, L. H., & Freeman, T. M. (2004). Students’ sense of belonging in school. In M. Machr & P. Pintrich (Eds.). Advances in motivation and achievement: Motivating students, improving schools: The legacy of Carol Midgley (pp. 27–63). Greenwich, CT: JAI Press. Baker, E. T., Wang, M. C., & Walberg, H. J. (1995). The effects of inclusion on learning. Educational Leadership, 52(4), 33–35. Ballard, M., Corman, L., Gottlieb, J., & Kaufman, M. J. (1977). Improving the social status of mainstreamed retarded children. Journal of Educational Psychology, 69(5), 605–611. Banks, J. A., & Banks, C. A. M. (1995). Handbook of research on multicultural education. New York, NY: Macmillan. Biordi, D. L., & Nicholson, N. R. (2012). Social isolation. In I. M. Lubkin & P. D. Larsen (Eds.), Chronic illness (pp. 97–115). Burlington, MA: Jones and Bartlett Learning. Blum, R. W., McNeely, C. A., & Rinehart, P. M. (2002). Improving the odds: The untapped power of schools to improve the health of teens. Minneapolis, MN: University of Minnesota, Center for Adolescent Health and Development. Cannon, J., Jacknowitz, A., & Painter, G. (2006). If full better than half? Examining the longitudinal effects of full-day kindergarten attendance. Journal of Policy Analysis and Management, 25(2), 299–321. Carrell, S. E., & Hoekstra, M. (2014). Are school counselors an effective education input? Economics Letters, 125(1), 66–69. Cohen, J. L. (2007). Causes and consequences of special education placement: Evidence from Chicago public schools. Washington, DC: Brookings Institution.

Coie, J. D., Dodge, L. A., & Kupersmidt, J. B. (1990). Peer group behavior and social status. In S. R. Asher & J. D. Coie (Eds.), Peer rejection in childhood: Cambridge studies in social and emotional development (pp. 17–59). New York, NY: Cambridge University Press. Cole, C. M., Waldron, N., & Majd, M. (2004). Academic progress of students across inclusive and traditional settings. Mental Retardation, 42(2), 136–144. Finn, J. D. (1989).Withdrawing from school. Review of Educational Research, 59(2), 117–142. Fitch, E. F. (1999). Special education and social censure: A case study of an inclusion program (Unpublished doctoral dissertation). Miami University, Oxford, OH. Fitch, F. (2003). Inclusion, exclusion, and ideology: Special education students’ changing sense of self. The Urban Review, 35(3), 233–252. Fletcher, J. M. (2009). The effects of inclusion on classmates of students with special needs: The case of serious emotional problems. Education Finance and Policy, 4(3), 278–299. Fletcher, J. M. (2010). Spillover effects of inclusion of classmates with emotional problems on test scores in early elementary school. Journal of Policy Analysis and Management, 29(1), 69–83. doi:10.1002/pam.20479 Freeman, S. F. N., & Alkin, M. C. (2000). Academic and social attainments of children with mental retardation in general education and special education settings. Remedial and Special Education, 21, 3–18. Friesen, J., Hickey, R., & Krauth, B. (2010). Disabled peers and academic achievement. Education Finance and Policy, 5(3), 317–348. Gibbons, S., & Silva, O. (2011). School quality, child wellbeing and parents’ satisfaction. Economics of Education Review, 30(2), 312–331. Goodenow, C., & Grady, K. E. (1993). The relationship of school belonging and friends’ values to academic motivation among urban adolescent students. The Journal of Experimental Education, 62(1), 60–71. Gottfried, M. A. (2014). Classmates with disabilities and students’ noncognitive outcomes. Educational Evaluation and Policy Analysis, 36(1), 20–43. Gottfried, M. A., Egalite, A., & Kirksey, J. J. (2016). Does the presence of a classmate with emotional/behavioral disabilities link to other students’ absences in kindergarten? Early Childhood Research Quarterly, 36, 506–520. Gottfried, M. A., & Harven, A. (2015). The effect of having classmates with emotional and behavioral disorders and the protective nature of peer gender. The Journal of Educational Research, 108(1), 45–61. Gottlieb, J. (1981). Mainstreaming: Fulfilling the promise? American Journal of Mental Deficiency, 86, 115–126. Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (2002). Inferring program effects for special populations: Does special education raise achievement for students with disabilities? Review of Economics and Statistics, 84(4), 584–599. Huguet, P., Dumas, F., Marsh, H., Seaton, M., Wheeler, L., Suls, J., . . . Nezlek, J. (2009). Clarifying the role of social comparison in the big-fish-little-pong effect (BFLPE): An integrative study. Journal of Personality and Social Psychology: Interpersonal Relationships and Group Processes, 97, 156–170. Jepma, I. J. (2003). The school career of pupils at risk in primary education. Amsterdam: Thela Thesis. Kauffman, J. M., & Hallahan, D. P. (1993). Towards a comprehensive delivery system for special education. In J. I. Goodlad & T. C. Lovitt (Eds.), Integrating general and special education (pp. 73–102). New York, NY: Merrill. Keslair, F., Maurin, E., & McNally, S. (2012). Every child matters? An evaluation of “Special educational needs” programmes in England. Economics of Education Review, 31(6), 932–948. MONTH XXXX    11

Kozleski, E. B., & Jackson, L. (1993). Taylor’s story: Full inclusion in her neighborhood elementary school. Exceptionality, 4, 153–175. Laws, G., & Kelly, E. (2005). The attitudes and friendship intentions of children in United Kingdom mainstream schools towards peers with physical or intellectual disabilities. International Journal of Disability, Development and Education, 52(2), 79–99. Lazear, E. P. (2001). Educational production. The Quarterly Journal of Economics, 116(3), 777–803. MacMillan, D. L., Gresham, F. M., & Fornes, S. R. (1996). Full inclusion: An empirical perspective. Behavioral Disorders, 21(2), 145–159. Markussen, E. (2004), Special education: Does it help? A study of special education in Norwegian upper secondary schools. European Journal of Special Needs Education, 19(1), 33–48. Moon, N. W., Todd, R. L., Morton, D. L., & Ivey, E. (2012). Accommodating students with disabilities in science, technology, engineering, and mathematics (STEM): Findings from research and practice for middle grades through university education. Atlanta, GA: Center for Assistive Technology and Environmental Access. Morgan, P. L., Farkas, G., Hillemeier, M. M., Mattison, R., Maczuga, S., Li, H., & Cook, M. (2015). Minorities are disproportionately underrepresented in special education: Longitudinal evidence across five disability conditions. Educational Researcher, 44(5), 278–292. Murray, C., & Greenberg, M. T. (2001). Relationships with teachers and bonds with school: Social emotional adjustment correlates for children with and without disabilities. Psychology in the Schools, 38(1), 25–41. Myklebust, J. O. (2007). Diverging paths in upper secondary education: Competence attainment among students with special educational needs. International Journal of Inclusive Education, 11(2), 215–31. New York City Department of Education, Office of Special Education Initiatives. (n.d.). Special education services as part of a unified service delivery system. Retrieved from http://www.uft.org/files/ attachments/doe-continuum-of-services.pdf Nicholls, J. (1990). What is ability and why are we mindful of it? A developmental perspective. In R. Sternberg & J. Kolligian (Eds.), Competence considered (pp. 11–40). New Haven, CT: Yale University Press. Nowicki, E. A. (2006). A cross-sectional multivariate analysis of children’s attitudes towards disabilities. Journal of Intellectual Disability Research, 50(5), 335–348. NYSED. (2017). Special education school district data profile 2015–16. Retrieved from https://data.nysed.gov/specialed/?year=2016&insti d=800000048663#Indicator5 Pearl, R., Farmer, T. W., Van Acker, R., Rodkin, P. C., Bost, K. K., Coe, M., & Henley, W. (1998). The social integration of students with mild disabilities in general education classrooms: Peer group membership and peer-assessed social behavior. The Elementary School Journal, 99(2), 167–185. Peetsma, T., Vergeer, M., Roeleveld, J., & Karsten, S. (2001). Inclusion in education: Comparing pupils’ development in special and regular education. Educational Review, 53(2), 125–135. Pittman, L. D., & Richmond, A. (2007). Academic and psychological functioning in late adolescence: The importance of school belonging. The Journal of Experimental Education, 75(4), 270–290. Reback, R. (2010). Schools’ mental health services and young children’s emotions. Journal of Policy Analysis and Management, 29(4), 698–725. Resnick, M., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., & Jones, J. (1997). Protecting adolescents from harm. Journal of the American Medical Association, 278(10), 823–832. Rogers, D. P., & Thiery, I. M. (2003). Does an inclusive setting affect reading comprehension in students with learning disabilties? Paper presented at Annual Meeting of the Mid-South Educational Research Association, Biloxi, MS. 12    EDUCATIONAL RESEARCHER

Rojewski, J. W., Lee, I. H., & Gregg, N. (2015). Causal effects of inclusion on postsecondary education outcomes of individuals with high-incidence disabilities. Journal of Disability Policy Studies, 25(4), 210–219. Rule, A. C., Stefanich, G. P., Haselhuhn, C. W., & Peiffer, B. (2009). A working conference on students with disabilities in STEM coursework and careers. Retrieved from https://eric.ed.gov/?id=ED505568 Rumberger, R. W. (1995). Dropping out of middle school: A multilevel analysis of students and schools. American Educational Research Journal, 32(3), 583–625. Sandler, A. G. (1999). Short-changed in the name of socialization? Acquisition of functional skills by students with severe disabilities. Mental Retardation, 37(2), 148–150. Schnorr, R. F. (1990). “Peter? He comes and goes . . . ”: First graders’ perspectives on a part-time mainstream student. Journal of the Association for Persons With Severe Handicaps, 15(4), 231–240. Schwab, N., & Gelfman, M. H. (2005). Legal issues in school health services: A resource for school administrators, school attorneys, school nurses. Bloomington, IN: iUniverse. Seaman, S. R., & White, I. R. (2013). Review of inverse probability weighting for dealing with missing data. Statistical Methods in Medical Research, 22(3), 278–295. Shogren, K. A., Kennedy, W., Dowsett, C., & Little, T. D. (2014). Autonomy, psychological empowerment, and self-realization: Exploring data on self-determination from NLTS2. Exceptional Children, 80(2), 221–235. Siperstein, G. N., & Leffert, J. S. (1997). Comparison of socially accepted and rejected children with mental retardation. American Journal of Mental Retardation, 101, 339–351. Stiefel, L., Shiferaw, M., Schwartz, A. E., & Gottfried, M. (2017). Is special education improving? Evidence on segregation, outcomes, and spending from New York City. Manuscript submitted for publication. Supalo, C. A., Mallouk, T. E., Rankel, L., Amorosi, C., & Graybill, C. M. (2008). Low-cost laboratory adaptations for precollege students who are blind or visually impaired. Journal of Chemical Education, 85, 243–247. Turner, R. J., & Noh, S. (1988). Physical disability and depression: A longitudinal analysis. Journal of Health and Social Behavior, 29, 23–37. U.S. Department of Education, Office of Special Education and Rehabilitative Services. (2007). 29th annual report to Congress on the implementation of the Individuals with Disabilities Education Act. Washington, DC: Author. U.S. Department of Education, Office of Special Education and Rehabilitative Services. (2015). 37th annual report to Congress on the implementation of the Individuals with Disabilities Education Act. Washington, DC: Author. Uwah, C., McMahon, H., & Furlow, C. (2008). School belonging, educational aspirations, and academic self-efficacy among African American male high school students: Implications for school counselors. Professional School Counseling, 11(5), 296–305. Van Gundy, K., & Schieman, S. (2001). Looking inward: Introspectiveness, physical disability, and depression across the life course. International Journal of Ageing and Human Development, 53(4), 293–310. Waldron, N., & McLeskey, J. (1998). The impact of a full-time Inclusive School Program (ISP) on the academic achievement of students with mild and severe learning disabilities. Exceptional Children, 64(3), 395–405. Wang, M. T., & Holcombe, R. (2010). Adolescents’ perceptions of school environment, engagement, and academic achievement in middle school. American Educational Research Journal, 47(3), 633–662. Williams, L. J., & Downing, J. (1998). Membership and belonging in inclusive classrooms: What do middle school students have to say?

Journal of the Association for Persons with Severe Handicaps, 23(2), 98–110. Wooldridge, J. M. (2002). Inverse probability weighted M-estimators for sample selection, attrition, and stratification. Portuguese Economic Journal, 1(2), 117–139. Authors

LEANNA STIEFEL, PhD, is professor of economics and education policy at New York University’s Wagner and Steinhardt Schools, 295 Lafayette Street, #3098, New York, NY 10012; [email protected]. Her research focuses on urban education policy and finance, currently special education policy, small school reforms, and student school and residential mobility. MENBERE SHIFERAW, MA, is in the last year of her doctoral program in public policy at The Wagner Graduate School of Public Service, New York University, 295 Lafayette Street, New York, NY 10012; [email protected]. Her research focuses on understanding and promoting the educational and labor market opportunities of special populations, such as English language learners, students with disabilities, immigrant students, and STEM teachers. AMY ELLEN SCHWARTZ, PhD, is the Daniel Patrick Moynihan Chair in Public Affairs at the Maxwell School of Citizenship and Public Affairs at Syracuse University, 426 Eggers Hall, Syracuse, NY 13244; [email protected]. Her research interests span a broad range of issues in education policy, urban economics, and public finance. MICHAEL A. GOTTFRIED, PhD, is an associate professor in the Gevirtz Graduate School of Education at the University of California Santa Barbara, Santa Barbara, CA 93106; [email protected]. edu. His research focuses on the economics of education and education policy, with a focus on children with disabilities. Manuscript received October 21, 2016 Revisions received April 18, 2017, July 25, 2017, and September 29, 2017 Accepted October 2, 2017

Table A1 Descriptive Statistics for New York City Public School Students in Grades 6 Through 8 and Ungraded Students, All Schools, 2012

Female White Black Hispanic Asian Home language is English Limited English Proficiency (LEP) Foreign-born Free/reduced lunch District 75 Inclusive services Exclusive services (SC) Learning disability (LD) Speech impairment (SI) Other health impairment (OH) Emotional disturbance (ED) Low incidence disability (LI) Average age N students Percentage of sample

GEN (%)

SWD (%)

51.8 14.7 28.2 38.6 18.5 55.9 11.0 19.3 81.8

36.4 13.2 31.8 47.6 7.4 66.1 15.6 8.3 87.5 6.0 82.5 17.5 56.1 24.1 7.1 5.8 6.9 13.6 26,871 13.4

13.4 173,262 86.6

Note. Sample includes students in New York City public schools in Grades 6 through 8 and ungraded students assigned Grade Levels 6 through 8 for New York State assessment purposes.

More Detail on Students in Sample by Assigned Services and Disability See Table A4. Steps to Address Survey Nonresponse

Appendix This appendix provides additional detail on the analytical sample used in the paper, how we address survey nonresponse, and robustness checks.

Descriptive Statistics Including District 75 Students Table A1 shows descriptive statistics for all students for comparison to Table 1 in the paper, which includes students in traditional schools only (our sample). Note that the percentage of students assigned exclusive services (17.5%) is higher than in our sample, and recall that District 75 (D75) educates students with disabilities (SWDs) only, and thus these students are not used in the paper’s analyses of students in inclusive schools. Details on Exclusions to Get to Final Sample See Note 9 for details that describe samples in Tables A2 and A3.

In the New York City (NYC) Learning Environment Survey, response rates to the student survey during the years of this study increased steadily from roughly 60% in 2007 to 90% in 2012. For robustness, we estimate all models in the paper excluding the year with the lowest response rate (2007), and our conclusions remain the same. Results for our preferred specifications with this exclusion, which corresponds to Table 3 in the paper, are presented in Appendix Table A5. To adjust for varying response rates across students, we weight all models by the inverse of the predicted probability of response. This is a common statistical procedure recommended by statisticians and commonly used by applied social scientists working with survey data (Seaman & White, 2013; Wooldridge, 2002). We first estimate a logistic regression predicting the probability of response to each survey question, adjusting for students’ gender, race/ethnicity, nativity, home language, English proficiency, free/reduced lunch eligibility, test scores, disability (for SWDs), assigned services (for SWDs), grade, and year. We then use the inverse of predicted probabilities as weights in our analyses. Results are robust to using a probit specification instead.

MONTH XXXX    13

Table A2 Disparities by Disability and Assigned Service

  Inclusive  ED    OH    LD    SI    LI   Exclusive-inclusive disparity   ED × EXCL     OH × EXCL     LD × EXCL     SI × EXCL     LI × EXCL   School by year fixed effects Constant   N students N schools N observations

(1)

(2)

(3)

(4)

(5)

Welcome

Bully

Harass

Known

Included (2010–2012 Only)

–0.058*** (0.008) –0.027*** (0.005) –0.012*** (0.002) –0.006** (0.003) –0.001 (0.007)

0.068*** (0.010) 0.048*** (0.006) 0.015*** (0.002) 0.023*** (0.003) 0.015* (0.009)

0.050*** (0.010) 0.017*** (0.006) 0.009*** (0.002) 0.009*** (0.003) –0.008 (0.008)

0.082*** (0.008) 0.074*** (0.004) 0.033*** (0.002) 0.029*** (0.003) 0.089*** (0.007)

–0.037*** (0.014) –0.015* (0.008) 0.020*** (0.003) 0.005 (0.005) 0.036*** (0.013)

0.036** (0.018) 0.003 (0.020) 0.002 (0.008) –0.002 (0.011) 0.047*** (0.016) Y 0.931*** (0.011) 470,006 507 855,287

–0.038 (0.025) 0.011 (0.031) 0.023** (0.011) 0.026 (0.016) –0.098*** (0.026) Y 0.409*** (0.013) 470,006 507 855,287

–0.019 (0.022) –0.013 (0.026) –0.007 (0.010) 0.008 (0.015) –0.071*** (0.022) Y 0.198*** (0.012) 470,006 507 855,287

0.005 (0.017) 0.013 (0.017) 0.002 (0.009) 0.026** (0.012) 0.004 (0.019) Y 0.593*** (0.011) 470,006 507 855,287

0.059* (0.033) 0.069* (0.037) 0.044** (0.018) 0.058** (0.024) 0.213*** (0.042) Y 0.402*** (0.019) 305,367 477 490,919

Note. Robust standard errors clustered by school by year in parentheses. Table 3 in paper estimated including students not in the sample for 2 or 3 consecutive years. All models include grade and school-by-year effects and the following student characteristics: gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/reduced lunch eligibility, age, indicator for number of years in the sample, and indicators for students who ever switched disability classification or service setting. Coefficients on the inclusive indicators represent the SWD-GEN disparity for SWDs assigned to inclusive services (RS, ICT, and SETSS). Coefficients on the interaction terms represent the exclusive-inclusive disparity. The SWD-GEN disparity for SWDs assigned to exclusive services (SC) can be obtained by summing the coefficients on inclusive and the interaction. Inclusive is set to zero for all GENs. SWD = students with disabilities; GEN = students not receiving special education services; LD = learning disability; SI = speech impairment; OH = other health impairment; ED = emotional disturbance; LI = low incidence; EXCL = exclusive service; RS = related services only; ICT = integrated co-teaching; SETSS = special education teacher support services; SC = self-contained. *p < .10. **p < .05. ***p < .01.

14    EDUCATIONAL RESEARCHER

Table A3 Disparities by Disability and Assigned Service

  Inclusive  ED    OH    LD    SI    LI   Exclusive-inclusive disparity   ED × EXCL     OH × EXCL     LD × EXCL     SI × EXCL     LI × EXCL   School by year fixed effects Constant   N students N schools N observations

(1)

(2)

(3)

(4)

(5)

Welcome

Bully

Harass

Known

Included (2010–2012 Only)

–0.066*** (0.008) –0.029*** (0.004) –0.012*** (0.002) –0.006** (0.002) –0.004 (0.006)

0.066*** (0.010) 0.049*** (0.006) 0.016*** (0.002) 0.023*** (0.003) 0.015* (0.009)

0.051*** (0.010) 0.019*** (0.006) 0.009*** (0.002) 0.008*** (0.003) –0.006 (0.008)

0.082*** (0.008) 0.072*** (0.004) 0.033*** (0.002) 0.030*** (0.003) 0.087*** (0.007)

–0.037*** (0.014) –0.015* (0.008) 0.020*** (0.003) 0.005 (0.005) 0.036*** (0.013)

0.027 (0.017) 0.008 (0.019) –0.003 (0.008) –0.004 (0.011) 0.049*** (0.014) Y 0.922*** (0.010) 496,995 509 939,449

–0.041* (0.023) 0.003 (0.031) 0.028*** (0.010) 0.028* (0.016) –0.080*** (0.025) Y 0.416*** (0.013) 489,198 508 912,336

–0.009 (0.023) –0.011 (0.026) –0.003 (0.010) 0.017 (0.015) –0.065*** (0.023) Y 0.204*** (0.013) 483,428 509 896,708

–0.004 (0.017) 0.036** (0.016) 0.004 (0.009) 0.029** (0.012) 0.003 (0.018) Y 0.581*** (0.012) 491,744 509 926,927

0.059* (0.033) 0.069* (0.037) 0.044** (0.018) 0.058** (0.024) 0.213*** (0.042) Y 0.402*** (0.019) 305,367 477 490,919

Note. Robust standard errors clustered by school by year in parentheses. Table 3 in paper estimated including students not in the sample for 2 or 3 consecutive years and students who answer any survey question each year. All models include grade and school by year effects and the following student characteristics: gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/reduced lunch eligibility, age, indicator for number of years in the sample, and indicators for students who ever switched disability classification or service setting. Coefficients on the inclusive indicators represent the SWD-GEN disparity for SWDs assigned to inclusive services (RS, ICT, and SETSS). Coefficients on the interaction terms represent the exclusive-inclusive disparity. The SWD-GEN disparity for SWDs assigned to exclusive services (SC) can be obtained by summing the coefficients on inclusive and the interaction. Inclusive is set to zero for all GENs. Note that estimates in Column 5 are identical to those in Table A1 because there are no survey exclusions made for the outcome included. SWD = students with disabilities; GEN = students not receiving special education services; LD = learning disability; SI = speech impairment; OH = other health impairment; ED = emotional disturbance; LI = low incidence; EXCL = exclusive service; RS = related services only; ICT = integrated co-teaching; SETSS = special education teacher support services; SC = self-contained. *p < .10. **p < .05. ***p < .01.

MONTH XXXX    15

Table A4 Percentage Distribution of Sixth- Through Eight-Grade Students With Disabilities Across Assigned Services by Disability, Analytic Sample, 2007–2012 Inclusive Services

Exclusive Services

Disabilities

RS

SETSS

ICT

SC

Total %

N

LD SI OH ED LI  

 1.0 17.9 28.6 24.3 39.3

57.9 37.8 32.9 21.4 23.2

37.5 39.8 35.1 37.9 26.5

3.6 4.5 3.4 16.3 11.0

100 100 100 100 100

37,263 15,759  4,398  1,661  2,085 N = 61,166

Note. LD = learning disability; SI = speech impairment; OH = other health impairment; ED = emotional disturbance; LI = low incidence; RS = related services only; ICT = integrated co-teaching; SETSS = special education teacher support services; SC = self-contained.

Table A5 Disparities by Disability and Assigned Service

  Inclusive  ED    OH    LD    SI    LI   Exclusive-inclusive disparity   ED × EXCL     OH × EXCL     LD × EXCL     SI × EXCL     LI × EXCL   School by year fixed effects Constant   N students N schools N observations

(1)

(2)

(3)

(4)

Welcome

Bully

Harass

Known

–0.058*** (0.011) –0.028*** (0.006) –0.012*** (0.002) –0.006** (0.003) –0.002 (0.008)

0.070*** (0.013) 0.049*** (0.007) 0.016*** (0.003) 0.024*** (0.004) 0.016 (0.011)

0.071*** (0.013) 0.015** (0.007) 0.008*** (0.003) 0.011*** (0.004) –0.014 (0.010)

0.079*** (0.009) 0.078*** (0.005) 0.030*** (0.002) 0.028*** (0.003) 0.094*** (0.008)

0.034 (0.029) –0.029 (0.030) –0.017 (0.016) –0.016 (0.021) 0.068** (0.027) Y 0.942*** (0.012) 248,951 493 567,577

–0.032 (0.041) –0.009 (0.049) 0.002 (0.018) 0.046 (0.028) –0.104* (0.054) Y 0.400*** (0.015) 248,951 493 567,577

–0.003 (0.035) –0.009 (0.037) –0.019 (0.017) 0.033 (0.023) –0.044 (0.047) Y 0.192*** (0.014) 248,951 493 567,577

–0.017 (0.027) –0.004 (0.026) 0.014 (0.015) 0.019 (0.017) 0.025 (0.024) Y 0.636*** (0.013) 248,951 493 567,577

Note. Robust standard errors clustered by school by year in parentheses. Table 3 in paper estimated without students in 2007, year with low survey response. All models include grade and school by year effects and the following student characteristics: gender, race/ethnicity, nativity status, home language, Limited English Proficient, free/ reduced lunch eligibility, age, in the sample for two years only, and indicators for students who ever switched disability classification or service setting. Inclusive is set to zero for all GENs. All models are weighted by the inverse of predicted probability of response. Survey question included was not asked in 2007. GEN = students not receiving special education services; LD = learning disability; SI = speech impairment; OH = other health impairment; ED = emotional disturbance; LI = low incidence; EXCL = exclusive service. *p < .10. **p < .05. ***p < .01. 16    EDUCATIONAL RESEARCHER

inclusion study.pdf

Page 1 of 3. Educational Researcher, Vol. XX No. X, pp. 1–16. DOI: 10.3102/0013189X17738761. © 2017 AERA. http://edr.aera.net. MONTH XXXX 1. The passage of landmark federal legislation in 1975 guaran- teed students with disabilities (SWDs) a free appropriate. public education.1. Over time, reauthorization of this ...

405KB Sizes 1 Downloads 283 Views

Recommend Documents

inclusion study.pdf
education services that allow us to parse more finely the single .... ours and also demonstrates precedence for collapsing federal cat- egories into analytically ...

Innovative Inclusion - GSMA
network operator (MNO) Telesom launched. Telesom ZAAD .... FREE-OF-CHARGE SERvICE; FOCUS ON INDIRECT REvENUES. TABLE 1. A comparison of ...

Inclusion of Teachers - Orders issued.
All District EducationalOfficers and Assistant Educational Officers. e6e Executive ... Name of the Teacher. Name of ... Lousha Cyriac, HSA Social Science. SMI{S ...

Brewing Social Inclusion
that allows anyone with an email address to connect to a participating network (Appendix B). Admittedly .... from: http://www.saraharnquist.com/chroniccare.html.

Choosing Inclusion, Peace and Justice.pdf
Choosing Inclusion, Peace and Justice.pdf. Choosing Inclusion, Peace and Justice.pdf. Open. Extract. Open with. Sign In. Main menu.

Inclusion of married mentally disabled.PDF
There was a problem previewing this document. Retrying... Download. Connect more apps... Try one of the apps below to open or edit this item. Inclusion of ...

NATIONAL IDENTITY & INCLUSION PROJECT ...
form of national security demands and enhanced focus on “civic integration” ..... to credit, accommodation and living conditions (2002, 49). The Traveller's ...

Barclay Primary - Inclusion Policy.pdf
... Profile scores and School Pupil Tracker, reading ages and whole‐. school pupil progress data. classroom‐based assessment and monitoring arrangements ...

VACANCY ANNOUNCEMENT Position : Inclusion ... -
Organization and development of tools and processes to facilitate result based management. - Knowledge Management experience: Experience on supporting documenting, identification of innovation, research and capitalization work. Skills and Knowledge:

Inclusion of Victim Advocates.pdf
Page 1 of 1. Page 1 of 1. Inclusion of Victim Advocates.pdf. Inclusion of Victim Advocates.pdf. Open. Extract. Open with. Sign In. Main menu. Displaying Inclusion of Victim Advocates.pdf. Page 1 of 1.

Inclusion of Aadhaar number.PDF
Download. Connect more apps... Try one of the apps below to open or edit this item. Inclusion of Aadhaar number.PDF. Inclusion of Aadhaar number.PDF. Open.

Pharmacological promotion of inclusion formation - Semantic Scholar
Mar 14, 2006 - mg/ml BSA). The luciferase was denatured at 40°C for 15 min, followed by incubation for 10 min on ice and 5-fold dilution into Tris buffer (TB; ...

Synthesis, conformational studies and inclusion ...
spectra were recorded on a Voyager-DETM STR Bio-. spectrometry_Workstation made by Applied Biosys- tems Inc. Azacalixarenes 1 [7d, l] and 5 [7b] and o,o'- bis(hydroxymethyl) p-substituted phenol dimers 9 [7d, l] and 10 [7b] were prepared according to

Hierarchical Inclusion Think ALoud.PDF
There was a problem previewing this document. Retrying... Download. Connect more apps... Try one of the apps below to open or edit this item. Hierarchical Inclusion Think ALoud.PDF. Hierarchical Inclusion Think ALoud.PDF. Open. Extract. Open with. Si

NATIONAL IDENTITY & INCLUSION PROJECT ...
Aug 20, 2008 - works emphasizes the role of cultural interpretation and the ..... the potential that this mis-match can translate into forms of social tension, which ...

NATIONAL IDENTITY & INCLUSION PROJECT ...
formation and identity”; and that “cultures are not internally homogenous” (1995 .... weight multiculturalists suppose, Brian Barry talks about culture in terms of people's ... seen in this more “anthropological” light, we can deal with cul

Libraries, indigenous peoples, identity and inclusion - Acta Académica
Aug 14, 2007 - http://herkules.oulu.fi/isbn9514268539/html/x2692.html [Accessed ... endangered languages; the promotion of literacy and basic bilingual education; warranty of ... Aboriginal projects like Aymara Uta (“Aymara house”, website ...

The Equity and Inclusion Enforcement Act - House Committee on ...
are re-segregating at an alarming rate, Ranking Members Scott (Education and ... programs, including the dissemination of information, technical assistance, ...