You are on page 1of 26

598780

research-article2015
ECXXXX10.1177/0014402915598780Exceptional ChildrenCarter et al.

Article
Exceptional Children
2016, Vol. 82(2) 209­–233
Randomized Evaluation of Peer © 2015 The Author(s)
DOI: 10.1177/0014402915598780
Support Arrangements to Support ec.sagepub.com

the Inclusion of High School


Students With Severe Disabilities

Erik W. Carter1, Jennifer Asmus2, Colleen K. Moss2,


Elizabeth E. Biggs1, Dan M. Bolt2, Tiffany L. Born2,
Matthew E. Brock1, Gillian N. Cattey1, Rui Chen1,
Molly Cooney2, Ethan Fesperman1, Julia M. Hochman1,
Heartley B. Huber1, Jenna L. Lequia2, Gregory Lyons2,
Kerrie A. Moyseenko2, Lindsay M. Riesch2, Rebecca A. Shalev2,
Lori B. Vincent2, and Katie Weir2

Abstract
Enhancing the social and learning experiences of students with severe disabilities in inclusive
classrooms has been a long-standing focus of research, legislative, and advocacy efforts. The
authors used a randomized controlled experimental design to examine the efficacy of peer support
arrangements to improve academic and social outcomes for 51 students with severe disabilities
in high school general education classrooms. Paraprofessionals or special educators recruited,
trained, and supported 106 peers to provide individualized academic and social assistance to
students with severe disabilities throughout one semester. Compared to students exclusively
receiving adult-delivered support (n = 48), students participating in peer support arrangements
experienced increased interactions with peers, increased academic engagement, more progress on
individualized social goals, increased social participation, and a greater number of new friendships.
Moreover, an appreciable proportion of relationships lasted one and two semesters later after the
intervention had concluded. These findings challenge prevailing practices for supporting inclusive
education and establish the efficacy and social validity of peer support arrangements as a promising
alternative to individually assigned paraprofessional support.

Over the past decade, growing attention has cen- of students with multiple disabilities now spend
tered on identifying how best to support students at least 40% of their day in general education
with severe disabilities to access rigorous, rele-
vant learning opportunities within the general 1
Vanderbilt University
education classroom (e.g., McLeskey, Waldron, 2
University of Wisconsin–Madison
Spooner, & Algozzine, 2014; Ryndak, Jackson,
& White, 2013). Myriad legislative, policy, and Corresponding Authors:
Erik W. Carter, Department of Special Education,
research developments have changed expecta-
Vanderbilt University, PMB 228, Peabody College,
tions not only for what students with severe dis- Nashville, TN 37072, USA.
abilities can and should learn, but also where E-mail: erik.carter@vanderbilt.edu
they should receive this instruction. Although
Jennifer M. Asmus, Department of Educational
not disaggregated by disability severity, national Psychology, University of Wisconsin-Madison, 1025 W.
data indicate 43% of students with intellectual Johnson Street, Madison, WI 53706, USA.
disability, 57% of students with autism, and 28% E-mail: asmus@wisc.edu
210 Exceptional Children 82(2)

classes in secondary schools (U.S. Department support arrangements involve equipping one
of Education, 2012). An underlying presump- or more peers to provide ongoing academic
tion is that enrollment in general education and/or social support to their classmate with a
classrooms expands the social opportunities of severe disability throughout the semester as
students, creates shared learning experiences they work together on activities designed for
alongside peers, and increases access to educa- all students by the classroom teacher (Carter,
tors with content expertise. Moreover, the Cushing, & Kennedy, 2009). Peers are
opportunities peers have to learn alongside and recruited from within the same classroom,
develop relationships with their classmates with participate in an initial orientation, provide
severe disabilities are presumed to shape their individualized supports outlined in a written
attitudes toward and expectations of people with plan, and receive ongoing guidance and feed-
disabilities (Carter, Hughes, Copeland, & Breen, back from paraprofessionals and educators as
2001; Scior, 2011). they assist their classmate. Socially, peers
At the same time, considerable concern has model age-appropriate social and communi-
been raised regarding the prevailing approaches cation skills, increase opportunities to con-
used to support students with severe disabilities tribute to task- and social-related discussions
in general education classrooms. The U.S. in the classroom, and diminish proximity and
Department of Education (2012) has reported attitudinal barriers within the classroom (Ban-
that more than 411,000 full-time equivalent dura, 1977; Carter, Sisco, & Chung, 2012).
special education paraprofessionals work with Academically, working alongside peers may
students ages 6 to 21. Regional studies suggest increase the amount of individualized sup-
most of these staff work one-to-one with stu- ports, corrective feedback, and response
dents at least weekly and almost half provide opportunities students with severe disabilities
some support in general education classrooms receive as they learn course content, as well as
(Carter, O’Rourke, Sisco, & Pelsue, 2009; the extent to which the materials and instruc-
Fisher & Pleasants, 2012). When assigned as tion they receive aligns with what general
the primary or exclusive source of assistance educators provide to all students in the class
for students with severe disabilities in general (Carter & Kennedy, 2006). Conceptually,
education classrooms, paraprofessionals may paraprofessionals and peers work in tandem
inadvertently hinder the very social and aca- to enable the focus student to learn relevant
demic gains they are present to promote content from the classroom teacher. In other
(Giangreco, 2010). This concern may be espe- words, support from these individuals should
cially heightened at the high school level as supplement, rather than supplant, instruction
curricular content becomes more challenging provided by a highly qualified professional
and the close proximity of adults often sup- (Giangreco, Carter, Doyle, & Suter, 2010).
presses adolescent interactions. Descriptive A small collection of single-case studies
data from both observational studies and the has documented the promise of these inter-
baseline phases of intervention studies often ventions for supporting the general education
paint a portrait of students with disabilities who participation of adolescents with severe dis-
are present in general education classrooms abilities (Carter, Cushing, Clark, & Kennedy,
without having a presence and enrolled without 2005; Carter, Moss, Hoffman, Chung, &
being meaningfully engaged (e.g., Carter, Sisco, 2011; Carter, Sisco, Melekoglu, &
Hughes, Guth, & Copeland, 2005; Carter, Kurkowski, 2007; Shukla, Kennedy, & Cush-
Sisco, Brown, Brickham, & Al-Khabbaz, 2008; ing, 1998, 1999). Students with disabilities in
Feldman, Carter, Asmus, & Brock, 2015; these studies interacted substantially more
Hughes et al., 2013). often with their classmates when working
The present study explores the efficacy and with peers relative to when supported entirely
acceptability of peer support arrangements as by paraprofessionals. Moreover, they accessed
an alternative to an exclusive reliance on a wide range of social, learning, and behav-
assistance from special education staff in high ioral supports from peers. Although documen-
school general education classrooms. Peer tation of learning outcomes is limited, these
Carter et al. 211

interventions promoted academic engagement Method


at least as well as, if not better than, individu-
ally assigned paraprofessional supports. Participants
Additional research is needed to provide This multiyear, randomized controlled study
deeper insights into the implementation and involved 99 high school students with severe
impact of these interventions in high school disabilities, 106 peer partners, and 51 school
settings. First, the current literature is com- staff.
posed entirely of within-participant evalua-
tions involving a total of just 15 adolescents
with severe disabilities. Randomized con- Students with severe disabilities. To be included,
trolled studies involving larger samples of stu- students must have (a) received special educa-
dents served across states could enable tion services under the categories of intellectual
stronger claims about the efficacy and general- disability and autism spectrum disorder (ASD),
izability of this intervention approach. Second, and/or qualified for the state’s alternate assess-
an expanded range of outcome measures is ment; (b) been enrolled in at least one general
needed to elucidate the diverse ways in which education class (other than physical education);
students are impacted by these arrangements. and (c) received individually assigned parapro-
Although much attention has focused on broad fessional or special educator support in that
social interaction behaviors, little is known class. After receiving university and district
about how these interventions affect the goal research approvals, we worked with school staff
attainment, skill acquisition, and friendship for- to identify all students meeting these criteria,
mation of students with disabilities. Moreover, randomly assigned them to one of three study
scant attention has been given to capturing how conditions (i.e., peer support arrangements,
students, peers, and staff view these interven- paraprofessional/adult support only, peer net-
tions. Third, most evaluations of peer-mediated works), and obtained relevant consent/assent.
interventions have been limited to a single The present article focuses on the first two con-
semester. Follow-up data are needed to deter- ditions to examine the effect and acceptability of
mine whether any new peer relationships endure an intervention focused on classroom participa-
across multiple semesters. Fourth, prior studies tion (i.e., peer networks take place outside of
have involved researchers either in directly instructional time).
implementing key aspects of the intervention or From the 206 students originally invited to
providing intensive support to school staff. participate in the peer support and comparison
Ensuring practitioners can deliver these inter- conditions, we obtained a final sample of 51
ventions with adequate fidelity and limited and 48 students, respectively. The total num-
external support is essential to establishing sup- ber of students enrolled in the study each
port models as tractable in typical schools. semester was 15 in fall 2011, 24 in spring
The purpose of this study was to evaluate 2012, 25 in fall 2012, 26 in spring 2013, and 9
the efficacy and social validity of peer support in fall 2013. Table 1 displays participant char-
interventions for high school students with acteristics by group. No differences were
severe disabilities. We sought to answer the fol- found between groups on the variables of gen-
lowing research questions: How do peer sup- der, χ2(1, n = 99) = 1.053, p = .305; race/eth-
port arrangements affect the academic and nicity (i.e., White vs. non-White), χ2(1, n =
social participation, social skills, goal attain- 99) = .043, p = .835; eligibility for free or
ment, and peer connections of high school stu- reduced-price meals, χ2(1, n = 97) = 1.16, p =
dents with severe disabilities? What do the peer .281; age, t(97) = 0.880, p = .381; or adaptive
interactions and supports exchanged look like behavior composite scores, t(97) = 0.72, p =
within these interventions? How do participat- .473. However, the Childhood Autism Rating
ing students, peers, paraprofessionals, and edu- Scale–2 (Schopler, Van Bourgondien, Wellman,
cators view the acceptability and feasibility of & Love, 2010) scores of students with ASD in
these interventions? the peer support condition indicated more
212 Exceptional Children 82(2)

Table 1.  Focus Student and Peer Partner Demographics.

Students with severe disabilities 

Comparison Peer support Peer partners


  (n = 48) (n = 51) (n = 106)
Gendera
 Female 15 (31.2%) 21 (41.2%) 33 (12.3%)
 Male 33 (68.8%) 30 (58.8%) 73 (68.9%)
Grade levela
 9th 9 (18.8%) 9 (17.6%) 30 (28.3%)
 10th 19 (39.6%) 19 (37.3%) 23 (21.7%)
 11th 15 (31.3%) 11 (21.6%) 23 (21.7%)
 12th 5 (10.4%) 12 (23.5%) 30 (28.3%)
Race/ethnicitya
  European American 32 (66.7%) 34 (66.7%) 81 (76.4%)
  African American 8 (16.7%) 7 (13.7 %) 13 (12.3%)
  Asian American 3 (6.3%) 4 (7.8%) 4 (3.8%)
  Native or Alaskan American 1 (2.1%) 2 (3.9%) 1 (0.9%)
  Hispanic or Latino/a 0 (0.0%) 0 (0.0%) 4 (3.8%)
  Other or multiple 3 (6.3%) 3 (5.9%) 3 (2.8%)
Eligible for free/reduced-price mealsa
 Yes 18 (37.5%) 14 (27.5%) —
 No 29 (60.4%) 36 (70.6%) —
  Missing information 1 (2.1%) 1 (2.0%) —
Primary or secondary disability categorya
 ASD 17 (35.4%) 15 (29.4%) —
  Intellectual disability 22 (47.9%) 28 (54.9%) —
  ASD and intellectual disability 3 (6.3%) 7 (13.7%) —
  Multiple disabilities 1 (2.1%) 1 (2.0%)  
  Other developmental disabilitiesb 4 (8.3%) 1 (2%)  
Participation in state assessmentsa
  Alternate assessment 39 (81.3%) 47 (92.2%) —
  Assessment with accommodations 6 (12.5%) 3 (5.9%) —
  Missing information 3 (6.3%) 2 (2.0%) —
CARS-2c
  t-score 43.1 (7.8) 48.5 (9.2) —
  Percentile ranking 27.5 (19.8) 47.1 (24.2) —
SSISc
  Social skills standard scores 80.8 (16.9) 83.3 (14.7) —
  Problem behaviors standard scores 118.5 (13.8) 118.7 (13.5) —
  Academic competence standard scores 73.6 (15.9) 76.1 (17.5) —
Vineland-IIc
  Communication standard scores 64.5 (11.8) 62.8 (10.2) —
  Daily living standard scores 67.3 (12.0) 64.8 (11.8) —
  Socialization standard scores 70.6 (12.4) 69.1 (12.4) —
  Adaptive behavior composite scores 65.2 (11.0) 63.3 (10.8) —

Note. ASD = autism spectrum disorder; CARS-2 = Childhood Autism Rating Scale, 2nd ed. (Schopler, Van
Bourgondien, Wellman, & Love, 2010); SSIS = Social Skills Intervention System (Gresham & Elliot, 2008); Vineland-II =
Vineland Adaptive Behavior Scales-Second Edition (Sparrow, Cicchetti, & Balla, 2005).
a
Frequency (percentage). bStudent either had only a primary category other than autism spectrum disorder/intellectual
disability (ASD/ID), or student had both a primary and secondary category other than ASD/ID. cMean (standard
deviation).
Carter et al. 213

severe symptoms than students in the com- across schools ranged from 1.3% to 54.9%
parison condition, t(41) = –2.06, p = .046. African American (M = 13.2%), 0% to 0.6%
American Indian (M = 0.2%), 0% to 10.5%
Peer partners.  For each student assigned to the Asian American (M = 3.9%), 36.5% to 94.4%
peer support condition, one or more peers European American (M = 72.7%), 1.5% to
(M = 2.1, range = 1–4) from the same class- 18.9% Hispanic/Latino/a (M = 7.4%), and 0%
room were invited to serve as peer partners. We to 9.8% multiple races (M = 3.6%). The per-
asked teachers and paraprofessionals to solicit centage of students eligible for free or
input from the focus student and to consider reduced-price meals ranged from 5.6% to
peers who they anticipated would benefit from 63.1% (M = 28.7%).
such involvement, were interested in getting to Our observations and interventions focused
know the student, would serve as a positive on a single general education classroom for
role model, and who had good attendance. Of each focus student. According to teachers, stu-
the 152 peers initially invited by staff, 46 opted dents were enrolled in an average of 3.3 (SD =
not to participate. Once agreeing to participate, 1.7; range = 1–7) general education classes
all but one of the peer partners remained in this during the semester they entered the study (M =
role throughout the semester. In all, 16 (15.1%) 3.1 for peer support condition and M = 3.5 for
peers reported having had prior experience as a the comparison condition). When students
peer partner (see Table 1 for demographics). were enrolled in more than one general educa-
Average percentage of days peers were in tion course, we selected one classroom based
attendance during the semester of their involve- on scheduling considerations (e.g., avoiding
ment was 95.2% (SD = 5.4) and average GPA classes during the same period of the day to
was 3.2 (SD = 0.8). End-of-semester grades in accommodate observations) and input from
the classes in which they served as peer part- school staff (e.g., classes in which adult sup-
ners were As (66.0%), Bs (20.8%), Cs (8.5%), port was available). We observed 40 students
Ds (1.9%), and Fs (2.8%). (40.4%) in core academic classes (i.e., math,
science, language arts, social studies) and 59
Adult facilitators. Peer support arrangements students (59.5%) in elective and related arts
were facilitated by 42 paraprofessionals and classes (e.g., music, chorus, band, art, theater,
nine special educators assigned to support the dance, digital arts, computer technology, well-
focus student. Most paraprofessionals were ness, culinary arts). No differences were found
female (90.5%) and White (94.1%); they aver- in type of classroom (i.e., core academic versus
aged 10.3 years (SD = 6.7) of total educational elective/related arts) by group, χ2(1, n = 99) =
experience and 6.8 years (SD = 5.6) at their cur- 0.326, p = .568. The average length of these
rent school. All of the paraprofessionals worked classes was 56.1 min (SD = 18.7), and the aver-
under the direct supervision of a special educa- age enrollment was 24.2 students (SD = 7.4). In
tor at their school. Most participating special all, 15 schools used 50-min class periods and 6
educators were female (77.8%) and all were schools used an alternating day block schedule
White (100%); they averaged 9.2 years (SD = of 100-min class periods.
5.8) of total educational experience and 5.0
years (SD = 5.0) at their current school. These
particular special educators did not supervise the Study Conditions
participating paraprofessionals. We used a randomized controlled trial with
pre, post, and follow-up measures to evaluate
Schools and Classrooms the efficacy of peer support interventions on
the social and academic outcomes of students.
Students attended 21 high schools within 12
districts across two states. Using demograph- Peer support arrangements. This intervention
ics from the final year each high school par- condition involved school staff establishing
ticipated, student enrollment averaged 1357 peer support arrangements (Carter, Cushing,
(SD = 449). Average race/ethnicity of students & Kennedy, 2009) for focus students.
214 Exceptional Children 82(2)

Facilitator training. All facilitators partici- increasing involvement in classroom activities,


pated in an initial training lasting an average and decreasing reliance on adult support; (d)
of 2.5 hr. This didactic training was presented the importance of confidentiality and respect-
by members of the research team and included ful language; (e) expectations specific to the
oral instruction, structured presentation slides, classroom and support strategies from the peer
handouts, and guided discussion. The content support plan; and (f) guidance about when to
addressed the goals of the intervention, strate- seek assistance (Carter, Cushing, & Kennedy,
gies for recruiting peers, creating peer support 2009). Project coaches were present.
plans, orienting and supporting peers, fading
direct support, and the role of the intervention Peer support arrangements. Following the
coach. We covered 100% of topics across all initial orientation meeting, students changed
trainings. seats to be in physical proximity. As the stu-
dents worked together daily throughout the
Peer support planning. Prior to schedul- remainder of the semester (M = 8.4 weeks,
ing a planning meeting for each focus stu- SD = 2.4), peer partners supported students
dent, an assigned coach (i.e., a member of the academically (e.g., encouraging contributions
research team) observed the focus student in to class and group discussion, sharing materi-
his or her respective general education class- als, collaborating on class assignments) and
room. Subsequently, facilitators and coaches socially (e.g., conversing about school and
met to develop an individualized peer sup- other activities, modeling appropriate social
port plan, which included support strategies skills, making introductions to other class-
(e.g., modeling ways to interact, highlighting mates) to participate in ordinary class activi-
similarities, teaching interaction skills) that ties, as outlined on a written support plan and/
could be used throughout the class period or prompted by facilitators. In other words,
(i.e., beginning, middle, end) and in a variety the facilitators supported peer partners and
of instructional formats (e.g., whole group, focus students in working together on activi-
lecture, small group, labs) to help students ties organized by the classroom teacher for all
access learning and other expectations within students.
the classroom. Each written plan listed strat- Facilitators modeled support strategies and
egies appropriate for the focus student, peer provided feedback as students initially began
partners, and facilitator (Carter et al., 2011) working together. As students gained experi-
and was shared with the general and special ence with one another, the facilitator was
education teacher (if not already part of the encouraged to fade back direct support as
planning). Planning meetings ranged from 25 appropriate. Coaches were present in the
to 90 min (M = 44 min) and a support plan classroom at least twice (range = 2–5) to mon-
was developed for 100% of students. itor implementation and provide suggestions,
answer questions, encourage use of facilita-
Peer partner training. The initial training tive strategies, and problem-solve any con-
for peer partners was led by the facilitators cerns. We provided facilitators an intervention
and averaged 34 min (range = 15–60 min) in checklist containing specific strategies for the
length. These trainings were delivered out- focus student, peer partners, and facilitators.
side of the classroom in small groups to peer The self-monitoring checklist was filled out
partners (11.2% of peer partners completed weekly and returned to the coach.
the orientation individually); focus students
were not present. Facilitators followed a Treatment fidelity. A summary of each
written outline to ensure they addressed (a) aspect of the peer support intervention—as
the rationale for peer support strategies; (b) reported by the intervention coach or the
background about the focus student; (c) gen- facilitator and averaged across students—is
eral goals of increasing the number of stu- displayed in Table 2. Core intervention com-
dents with whom the focus student interacts, ponents are bolded. Data on four of these
Carter et al. 215

Table 2.  Fidelity Items for Peer Support Intervention.

Data source

Intervention coach Peer support facilitator

Fidelity indicators (abbreviated) M (%) SD (%) M (%) SD (%)


Researcher-implemented items
  Intervention facilitator completed the 100.0 0.0 — —
  full traininga
 Intervention facilitator received a 100.0 0.0 — —
  minimum of 2 coaching sessionsa
Facilitator-implemented items
 The facilitator recruited at least one 100.0 0.0 — —
  peer for the interventiona
 The facilitator addressed all topics at 100.0 0.0 — —
  the initial peer orientation meetinga
 The facilitator supported peer partners 89.6 24.7 97.5 8.6
  and the studenta
   The facilitator provided reminder/feedback 66.6 32.3 95.3 12.9
  to peer partners before, during, or after classb
   The facilitator facilitated interactions during 64.0 33.8 91.7 20.4
  class when appropriateb
  The facilitator monitored students during classb 87.2 26.1 — —
   The facilitator provided praise and feedback 62.4 33.4 94.1 10.6
  to students during or outside of classb
Peer-implemented items
 Peer partners were in close proximity 90.0 20.5 96.7 10.5
  to the focus student during classa
  The students sat next to each otherb 86.4 25.1 89.3 20.5
   The students remained in close proximity 65.5 39.4 83.7 24.7
  during out-of-seat class activitiesb
   During group activities, the students joined 63.8 43.1 83.9 27.4
  the same groupb
 Peer partners interacted with the 90.8 20.3 95.8 16.1
  focus student in classa
  Peer partners greeted the focus studentb 80.4 26.4 93.0 18.8
  The students engaged in conversationb 72.5 34.3 85.0 26.9
   Peer partners included the student 21.0 29.6 60.8 33.6
  in interactions with other peersb
 Peer partners assisted the focus 82.5 24.9 91.6 20.9
  student academicallya
  Peer partners helped the student participate 63.6 35.0 84.3 25.9
  in class activitiesb
  Peer partners repeated or rephrased 44.4 33.5 69.1 37.6
  instructions for the studentb
   Peer partners appropriately prompted 62.6 35.8 77.5 28.9
  the studentb
   Peer partners provided appropriate feedback 50.6 35.0 80.8 28.4
  to the studentb
  The students worked together on 42.0 35.7 76.9 33.2
  classroom activitiesb
  The students shared work materialsb 34.7 31.2 65.0 39.3

Note. Percentage standard deviations are calculated from the means of fidelity measures collected for each participant.
a
Core intervention components identified for peer support interventions. bDenotes a subcomponent of the core
component under which it is listed.
216 Exceptional Children 82(2)

main components were collected as students We converted frequency counts to rates (per
worked together within peer support arrange- hour) and durational measures to percent-
ments: (a) the facilitator supported peer part- ages of total class time.
ners and the student, (b) peer partners were
in close proximity to the focus student dur- Social interactions and peers contacted. We
ing class, (c) peer partners interacted with the coded all social exchanges between the focus
focus student in class, and (d) peer partners student and other classmates without dis-
assisted the focus student academically. We abilities as social interactions. Interactive
considered the intervention to be sufficiently behaviors included any verbal or nonverbal
implemented during a given class when the behavior with clear communicative intent pro-
first two components and either of the last two duced by the focus student to another student
components were observed. Using this defi- without severe disabilities (or vice versa). We
nition, fidelity averaged 96.2% (SD = 11.4%) coded each interactive behavior according to
and 87.6% (SD = 16.5%) across participants its function (i.e., initiation, response), topic
based on facilitator and coach reports, respec- (i.e., task- or social-related), as well as source
tively. and recipient (i.e., focus student, peer sup-
port, classmates without disabilities). Interac-
Individually assigned adult support.  This “busi- tive behaviors were task-related when they
ness as usual” comparison condition involved addressed activities, materials, or expectations
students receiving individually assigned sup- related to the current class; all other interac-
port from paraprofessionals or special educa- tive behaviors were considered social-related.
tors during ongoing class activities planned We coded behaviors as initiations if they were
by the general educator. This reflected the pre- preceded by at least 5 s without an interaction,
vailing support for students in these class- they reflected a change from task- to social-
rooms as indicated in their individualized related topic, or they occurred with a new
education programs. Although students may peer; all other interactions were responses.
have naturally received occasional help from We used the total interactions for the sum of
classmates, no new peer-mediated interven- all initiations and responses between the focus
tions were introduced to these students and we student and peers, whereas focus student con-
observed no formal pairings of students tributions refers to only those initiations and
throughout the study. Data describing this responses made by the focus student. To cap-
condition are displayed in Tables 3 and 4. ture peers contacted, we tallied the number
of different peers interacting with the student
each class period.
Measures and Procedures
Classroom observations. For each focus stu- Academic engagement. We coded students
dent, we observed three full-length class with disabilities as engaged in consistent
periods toward the beginning of the semester activities when they were actively involved in
(pre) and three full-length class periods or attending to instruction or classroom activi-
toward the end of the semester (post; i.e., 590 ties that were aligned with those provided by
total observations; four students missed one the general education teacher to the major-
post-observation). Each set of three observa- ity of the class. Students with disabilities
tions was collected over a 2- to 3-week were engaged in inconsistent activities when
period and lasted an average of 56.8 min of actively involved in or attending to instruction
classroom time per observation. We used or activities that were not aligned with those
tablet computers equipped with MOOSES provided to the majority of the class, but were
(Version 4) to collect durational (i.e., aca- assigned by a paraprofessional or teacher. Stu-
demic engagement, instructional format, dents were not engaged when not attending to
proximity) and frequency (i.e., social inter- any activities or materials related to instruc-
actions) data second-by-second in real time. tion or when there was no instruction.
Carter et al. 217

Table 3.  Descriptive Statistics of Classroom Observational Data by Condition.

Comparison Peer support


group group

Peer
Observational measures Pre Post Pre Post comparison
Total interactionsa 17.2 (43.1) 30.6 (64.2) 12.1 (30.9) 50.5 (58.2) 97.7 (100.3)
  Social-related interactions only 7.9 (24.5) 17.6 (50.0) 4.7 (15.7) 20.5 (32.3) 61.1 (82.4)
  Task-related interactions only 9.3 (24.1) 13.0 (33.5) 7.3 (19.8) 29.9 (43.6) 36.6 (47.4)
Focus student contributions 8.4 (20.5) 15.5 (31.8) 5.2 (14.8) 22.3 (28.7) 49.8 (60.0)
onlya
  Total initiations 3.1 (7.5) 5.6 (11.3) 2.0 (5.8) 6.3 (8.9) 14.8 (14.2)
  Total responses 5.3 (15.1) 10.0 (23.0) 3.2 (9.3) 16.1 (21.6) 35.0 (38.3)
Peer contributions onlya 8.8 (23.0) 15.1 (32.9) 6.9 (18.2) 28.1 (30.4) 47.9 (49.6)
  Total initiations 3.3 (7.6) 4.8 (11.6) 3.3 (12.4) 11.5 (11.8) 11.4 (11.5)
  Total responses 5.6 (15.0) 10.3 (22.6) 3.5 (10.7) 16.6 (22.8) 36.5 (40.2)
Observer ratings
 Reciprocityb 2.3 (0.7) 2.3 (0.7) 2.3 (0.7) 2.4 (0.7) 2.9 (0.3)
 Appropriatenessb 2.8 (0.9) 2.9 (0.4) 2.9 (0.4) 2.8 (0.5) 3.0 (0.0)
 Affectb 2.6 (0.6) 2.5 (0.5) 2.5 (0.6) 2.7 (0.5) 2.9 (0.3)
  Response relevanceb 2.9 (0.2) 2.9 (0.2) 3.0 (0.2) 2.9 (0.3) 3.0 (0.2)
  Overall qualityc 3.6 (1.1) 3.7 (1.1) 3.6 (0.9) 3.8 (0.9) 4.8 (0.5)
Peers contactedd 1.18 (1.63) 1.61 (2.09) 1.16 (1.58) 2.51 (1.93) 4.5 (3.2)
Academic engagemente 62.2 (31.3) 58.3 (34.6) 64.2 (30.3) 71.1 (26.8) 82.9 (21.7)
Proximity to direct supporte 48.7 (34.7) 42.3 (35.8) 50.6 (35.2) 31.1 (32.6) —
Proximity to peer partnerse — — — 62.9 (29.4) —
Gonee 14.8 (20.4) 18.2 (23.9) 15.9 (18.9) 11.2 (16.3) 0.7 (3.2)
a
Mean per hour (SD). bMean ratings on 3-point Likert-type scale (SD). cMean ratings on 5-point Likert-type scale (SD).
d
Mean (SD). ePercentage of class (SD).

Proximity to others. We coded whether high), the appropriateness of content (i.e.,


the focus student was in proximity to peer inappropriate, neutral, appropriate), affect of
partner(s), other classmate(s), and the adult the focus student (i.e., positive, neutral, nega-
assigned to provide direct support. We defined tive), and response relevance (i.e., not related,
proximity as being oriented in a position and somewhat related, mostly related). We also
distance that allowed interactions (about 3 used a 5-point Likert-type scale to rate the
feet). overall quality of the interaction between the
peers and focus student (i.e., low, medium–
Gone. When the focus student was not low, medium, medium–high, high). Ratings
present in the classroom at any point between were not made if no interactions occurred dur-
the start and end bells, we coded the student as ing an observation.
gone. This information was collected as one
of seven mutually exclusive and exhaustive Support behavior checklist. We used an
instructional formats (see Table 5). observational checklist (see Table 4) to docu-
ment the occurrence of support behaviors
Observer ratings.  At the end of each obser- provided by any peers and paraprofessionals
vation, we rated different dimensions of the toward the focus student at any time during
focus student’s interactions with peers: reci- each observed class period. We categorized
procity of interactions (i.e., low, medium, these as academic-related supports, social-
218 Exceptional Children 82(2)

Table 4.  Percentage of Class Periods in Which Academic, Social, and/or Other Support Behaviors
Were Observed by Source and Group.

Comparison group Peer support group

Adult- Peer- Adult- Peer-


  delivered delivered delivered delivered

Pre Post Pre Post Pre Post Pre Post


Abbreviated support behaviors (%) (%) (%) (%) (%) (%) (%) (%)
Academic-related supports
  Assist with completing class assignments 54.2 47.9 10.4 12.0 68.6 43.7 13.1 43.0
  Motivate/encourage the student 27.1 26.8 6.9 12.7 47.7 35.8 3.3 33.8
  Redirect when the student is off task 36.8 27.5 4.9 4.9 41.2 39.7 2.6 25.8
 Help the student participate in a group 13.9 12.7 6.3 7.7 11.8 7.3 3.3 13.9
 activity
  Help student keep organized 21.5 16.2 1.4 2.8 20.3 25.8 0.7 11.3
  Sharing class materials other than notes 0.7 0.7 6.3 6.3 1.3 0.7 2.6 9.9
  Assist in taking notes or share notes 4.9 2.8 2.1 1.4 2.6 0.7 0.7 7.3
  Prompt the student to answer a question 0.7 3.5 1.4 2.1 3.3 4.0 0.0 7.3
  Paraphrase lectures or discussions 4.2 4.2 0.0 0.7 6.5 0.7 0.7 2.6
  Modify/adapt assignments during class 22.2 16.9 1.4 0.0 21.6 18.5 0.0 4.0
  Read aloud a book section or assignment 2.1 3.5 0.7 0.0 3.3 4.0 0.7 2.0
 Write down answers given orally/with 2.1 0.0 0.0 0.0 5.2 4.0 0.0 2.0
  a device
  Other academic-related supports 4.9 3.5 2.1 1.4 7.8 0.7 0.7 4.6
 None of these supports observed 26.4 32.4 81.9 76.8 19.0 33.8 82.4 34.4
  during a period
Social-related supports
 Prompt the student to interact 5.6 2.8 1.4 2.8 9.2 28.5 0.0 4.0
  with classmates
 Encourage classmates to interact 2.8 2.1 0.7 1.4 7.2 22.5 0.0 4.0
  with the student
 Praise social/communication 0.0 1.4 0.0 2.8 2.0 6.6 0.0 4.0
 attempts/behaviors
  Provide emotional support or give advice 0.0 1.4 0.0 0.0 0.7 2.6 0.7 4.0
  Prompt the student to use AAC device 0.7 0.0 0.0 0.0 0.0 2.6 0.0 2.0
 Explicitly teach the student specific 0.0 0.7 0.7 1.4 2.0 3.3 0.0 0.7
  social skills
  Other social-related supports 2.1 2.1 0.0 0.7 2.6 0.7 0.0 1.3
 None of these supports observed 93.8 93.7 97.9 96.5 86.9 63.6 99.3 88.1
  during a period
Other supports
 Help the student self-manage 10.4 9.9 4.9 2.8 20.9 10.6 0.7 7.9
  own behaviors
  Explain class rules 0.0 0.7 0.7 0.0 1.3 0.0 0.0 0.0
  Explain class schedule 2.8 2.8 0.7 0.0 2.6 2.0 0.7 0.0
  Other 0.7 4.2 0.0 0.0 3.3 2.0 0.0 1.3
 None of these supports observed 88.2 88.0 94.4 97.2 77.8 88.1 99.3 92.1
  during a period
Total number of any type of support 2.1 1.8 0.5 0.6 2.8 2.6 0.3 1.9
per class M (SD) (1.8) (1.8) (1.3) (1.4) (1.9) (2.4) (0.7) (1.9)

Note. Percentages sometimes exceed 100% because more than one support could be delivered in a class period.
AAC = augmentative and alternative communication.
Carter et al. 219

related supports, or other supports (Carter appropriate interactions and the skills needed
et al., 2011). to manage social situations. It comprises seven
subdomains (i.e., communication, cooperation,
Peer comparisons. Using the same obser- assertion, responsibility, empathy, engagement,
vational measures, we collected normative self-control). Teachers rated the frequency with
peer comparison data at the beginning of the which the focus student exhibited each of the
semester in 53 of the classrooms and at the 46 social skill items using a 4-point Likert-type
end of the semester in the remainder. Observ- scale (i.e., never, seldom, often, and almost
ers selected peers who (a) did not have severe always). With our own sample, Cronbach’s
disabilities and (b) were in close enough prox- alphas for the seven subdomain raw scores
imity to allow for accurate observation. One were .92 and .89 for pre- and postintervention
peer was chosen for the first half of the class scores, respectively. We report standard scores.
and a different peer was observed for the sec-
ond half; observations were combined to form Social and academic goal attainment. We
a class-length observation. used goal attainment scaling (Schlosser, 2004)
to measure progress on students’ individual-
Interobserver agreement.  A total of 21 grad- ized goals within the general education class-
uate students, staff, and faculty served as class- room as a way to standardize scoring across
room observers. Observers discussed coding content areas and students. Toward the begin-
definitions and scored 100% on quizzes prior ning of the semester, special educators crafted
to observing. We required observers to meet one academic and one social-related goal, out-
(a) 90% agreement on practice videos and lining five measurable outcome levels for each
(b) 80% agreement with a trained observer in goal. The outcome expected for a student by
classrooms. For more than one third (33.7%) the end of the semester was assigned a score
of all observations—balanced across students of 0. An outcome slightly worse than expected
and time points—a second observer indepen- was assigned a –1 and an outcome much worse
dently collected data. We calculated point-by- than expected was assigned a –2. Conversely,
point agreement for event-based codes using an outcome slightly better than expected was
a 5-s window around the primary observer’s assigned a +1 and an outcome much better than
data files. We determined total percentage of expected was assigned a +2. At the end of the
agreement by dividing the number of agree- semester, special educators received the origi-
ments by the number of agreements plus dis- nal goals and outcomes and reported the level
agreements, and multiplying by 100%. We attained by the student in each area.
compared duration-based codes on a second-
by-second basis and calculated agreement Social participation. We collected two mea-
using the same formula (see Table 5). When sures to assess the social participation of stu-
both observers agreed on the nonoccurrence dents within general education classrooms
of interactions, we recorded agreement on fre- and the development of new friendships.
quency codes for that class period.
Classroom participation ratings. General edu-
cators completed ratings of each focus student’s
Student skills.  We asked the special education
social participation toward the beginning and
case managers for focus students to complete
end of the semester. Using a complete list of all
a packet of assessments at the beginning and
students, teachers rated the extent to which each
end of the semester.
student in the classroom (a) talked with other
peers in this class, (b) actively participated in that
Social skill ratings.  The Social Skills class, (c) worked with peers in that class, and (d)
Improvement System (SSIS; Gresham & had friends in that class. Response options for
Elliot, 2008) is a widely used, psychometrically the first three items were never, rarely, some-
strong assessment of behaviors that promote times, or frequently; options for the last item
220 Exceptional Children 82(2)

Table 5.  Interobserver Agreement for Direct Observation Measures.

Observations of all Observations Observations of


students and peer of focus students peer comparisons
Measure comparisons only only
Social interactions
  Focus students
  Task-related initiation 88.1 (0–100) 87.1 (0–100) 92.5 (0–100)
  Task-related response 84.0 (0–100) 82.3 (0–100) 89.2 (50–100)
  Social-related initiation 90.2 (0–100) 86.4 (0–100) 93.3 (67–100)
  Social-related response 82.9 (0–100) 82.4 (0–100) 81.0 (0–100)
  Classmates without disabilities
  Task-related initiation 90.1 (0–100) 90.8 (0–100) 92.2 (50–100)
  Task-related response 84.2 (0–100) 82.2 (0–100) 90.9 (50–100)
  Social-related initiation 87.8 (0–100) 82.9 (0–100) 93.1 (50–100)
  Social-related response 84.2 (0–100) 81.3 (0–100) 85.4 (50–100)
  Peer partners
  Task-related initiation 90.1 (33–100) 90.5 (33–100) n/a
  Task-related response 82.9 (0–100) 83.5 (0–100) n/a
  Social-related initiation 91.0 (0–100) 91.6 (0–100) n/a
  Social-related response 86.1 (50–100) 83.3 (0–100) n/a
  Number of peers contacteda 0.90 (0.02) 0.92 (0.02) 0.81 (0.07)
Academic engagement
 Engaged-consistent 98.4 (15–100) 98.0 (0–100) 99.9 (99–100)
 Engaged-inconsistent 84.1 (0–100) 87.9 (0–100) 99.8 (99–100)
  Not engaged 96.7 (0–100) 96.1 (0–100) 96.8 (63–100)
Proximity to
  Peer partners (post only) 97.4 (46–100) 97.5 (46–100) n/a
  Classmates without disabilities 98.0 (0–100) 97.3 (0–100) 99.8 (98–100)
  Direct support personnel 93.2 (0–100) 93.2 (0–100) 97.2 (97–98)
Instructional format
  Large group 97.4 (0–100) 96.7 (0–100) 99.4 (92–100)
  Small group 94.6 (0–100) 92.8 (0–100) 91.0 (32–100)
  Independent work 97.9 (0–100) 96.2 (0–100) 99.2 (95–100)
  1:1 peer 95.0 (69–100) 91.6 (0–100) n/a
  1:1 adult 96.9 (44–100) 95.5 (0–100) 100
  No instruction 92.9 (0–100) 90.8 (0–100) 98.9 (96–100)
 Gone 98.7 (54–100) 98.7 (54–100) 95.8 (88–100)
Support behaviors
  Adult-delivered supports n/a 95.6 (82–100) n/a
  Peer-delivered supports n/a 94.5 (73–100) n/a
Observer ratingsa
 Reciprocity 0.80 (0.03) 0.77 (0.04) 0.80 (0.03)
 Appropriateness 0.89 (0.03) 0.88 (0.03) 0.89 (0.03)
 Affect 0.82 (0.03) 0.82 (0.04) 0.82 (0.03)
  Response relevance 0.90 (0.03) 0.89 (0.07) 0.90 (0.03)
  Overall quality 0.69 (0.04) 0.64 (0.04) 0.68 (0.04)

Note. Values are mean (range).


a
Overall Cohen’s kappa (standard error) is reported for number of peers contacted as well as observer ratings of
interaction quality.
Carter et al. 221

were none, few, some, or many. To gauge the we completed this measure at the midpoint of
participation of each focus student with respect each of the subsequent two semesters (i.e., one
to the distribution of scores for each classroom, semester and 1 year later).
we standardized each classroom’s distribu-
tion of ratings for each of the four items and a Social validity.  We asked focus students, peers,
composite score using the standard z-score for- facilitators, and general educators to share
mula. For each classroom, we created cutoffs their views on the intervention toward the end
to identify a percentile rank for students with of the semester. Focus students with disabili-
high, above average, average, below average, ties were asked questions about their peer
or low levels of social participation using 0.5 partners, friendships, and enjoyment of
standard deviations. Cronbach’s alphas for the school. Responses were recorded as yes, no,
four items were .83 and .85 at pre- and postint- unsure, and unclear (i.e., the student could not
ervention, respectively. clearly communicate a response); special edu-
cators read questions to students, elaborating
Friendship gains.  We collected data on friend- and recording answers as needed. The facilita-
ships by having special educators complete our tor survey included 20 statements addressing
Social Connections and Relationships Assess- the amount of time and support required to
ment (adapted from Kennedy & Itkonen, implement the intervention, their interest and
1996). We asked them to list every social con- motivation to implement the intervention, and
tact (i.e., an interaction lasting at least 15 min their perceptions of benefits for participating
around a shared activity) the focus student had students, all rated on a 5-point Likert-type
at school during the prior two weeks. For each scale. Peer partners completed a similar sur-
contact, we requested (a) the first name of the vey that also asked if they would recommend
peer, (b) the length of time the student had participation to other students and if the
known the peer (i.e., less than or more than school should have more peer supports in the
1 month), (c) the perceived importance of the future. The general educator survey included
peer to the focus student (i.e., not very, some- questions similar to the facilitator survey.
what, very important), (d) whether or not the
peer also had a severe disability, (e) whether
Data Analysis
or not the peer was considered a friend (i.e.,
someone the focus student considered to be We used descriptive statistics to summarize
socially important and whom they liked), and (a) all observational measures by group and
(f) whether the peer was a peer partner. In a data collection wave, (b) social validity find-
separate section, we asked respondents to list ings, and (c) intervention fidelity measures.
other peers with whom the focus student had To evaluate the efficacy of peer support
not had a social contact in the prior 2 weeks— arrangements relative to the comparison con-
but were considered to be friends—along dition, we ran a series of hierarchical linear
with items (a) through (e) above. We encour- models (HLMs; Raudenbush & Bryk, 2002)
aged respondents to ask the focus student and evaluating the effects of condition for each
other school staff who knew the student well of 11 targeted outcome variables using the
to provide input when completing the measure. software Hierarchical Linear and Nonlinear
From these responses, we determined the total Modeling 7 while accounting for nesting of
number of different peers without disabilities students within schools. The targeted outcome
considered to be friends of the focus student, variables were identified prior to data analysis
new peers identified as friends toward the end and selected based on our interest in under-
of the semester, and peers no longer identified standing the impacts of peer support arrange-
as friends toward the end of the semester. Thus, ments on classroom behavior, skill acquisition,
friendship gains reflect the number of new and connections with peers. All data were
friends at the end of the semester. To gauge the screened prior to analysis for nonnormality
durability of social contacts and relationships, and potential outliers.
222 Exceptional Children 82(2)

Classroom observations. Observational vari- expected change in the comparison condition,


ables were collected for all students three while β11k represents the difference in expected
times prior to the intervention and three times change between the peer support and com-
after the intervention. This data structure per- parison conditions. The terms r0jk and r1jk rep-
mits a three-level analysis, where the first level resent residual student-level variance.
considers repeated measures within student The Level 3 model attends to school effects
(i.e., up to six observations per student) that are and is written as
modeled as a function of a single dummy vari-
able reflecting the postintervention (versus pre- β00k = γ 000 +u00k
intervention) time period. This modeling β01k = γ 010
framework introduces a subject parameter rep-
β10k = γ 100 +u10k
resenting the pre to post change in the observa-
tional variable which can in turn be modeled as β11k = γ 110
a function of the intervention condition to
which the student was assigned. Statistically, where γ000 represents the overall average pre-
the Level 1 model can be written as intervention score in the comparison condi-
tion, γ010 the average difference between peer
OBSERVtjk = π 0 jk + π 1 jk * ( POSTtjk ) + etjk support and comparison students at preinter-
vention, γ100 the average change from pre to
post in the comparison condition, and γ110 the
where OBSERVtjk represents the measure of average difference in change between peer
an observational variable at time t for student support and comparison students. The random
j from school k, POSTtjk is a dummy variable intercept terms, u00k and u10k account for
(0 = preintervention, 1 = postintervention) potential school-level variability in both the
identifying the time period in which the mea- overall preintervention scores and mean
sure was observed, π0jk represents the expected change in the scores from pre to post interven-
value for OBSERV for the student during the tion across both the comparison and peer-sup-
preintervention period, π1jk is the expected port conditions. Our primary hypothesis
change from the pre- to postintervention concerning the effects of the peer support
period, and etjk is an error term. The π0jk and intervention on change from pre to post is
π1jk effects are modeled at Level 2 (student tested through the γ110 parameter, where non-
level) as zero values imply a differential change in out-
come related to peer support. We analyzed six
observational variables with this framework:
π 0 jk = β00k + β01k * ( PSGROUPjk ) + r0jk total interactions, focus student interactions,
π 1 jk = β10k + β11k * ( PSGROUPjk ) + r1jk number of different peers contacted, academic
engagement activities, proximity to adults,
and time gone from class.
where PSGROUPjk is a dummy variable indi-
cating whether student j from school k is in the Student skills and social participation. For the
control (= 0) or peer support (= 1) condition. five teacher-reported measures (i.e., social
As a result, β00k and β01k correspond to the skills, classroom participation, social goal
expected preintervention scores in the com- attainment, academic goal attainment, friend-
parison condition, and the expected difference ship gains), we collected either a single pre-
in preintervention scores between the peer and a single postintervention measurement
support and comparison conditions, respec- (for the first two measures) or only a postint-
tively, for school k. The β10k and β11k coeffi- ervention measurement (the last three mea-
cients correspond to the expected change from sures). We therefore fit two-level models in
preintervention to postintervention within which students were nested within schools
school k. Specifically, β10k represents the and the postintervention measurement is the
Carter et al. 223

outcome variable. For the first two measures models, respectively) is divided by the pooled
only, we introduce the preintervention score within-group standard deviation of the respec-
as a covariate. The Level 1 models can thus be tive dependent variable. Effect size magni-
written as tudes of 0.20, 0.50, and 0.80 are generally
considered small, medium, and large effects,
( a ) POST_Score jk = β0k + β1k * ( PSGROUPjk ) respectively (Cohen, 1988).
Although our primary analyses assume nor-
+ β 2k * ( PRE_Score jk ) + rjk , or
mality, we also carried out robustness checks in
( b ) POST_Score jk = β0k + β1k * (PSGROUPjk ) + rjk cases where normality seemed suspect. Spe-
cifically, for two of the count-based outcome
variables (i.e., total interactions, peers con-
where POST_Scorejk and PRE_Scorejk repre-
tacted) we also conducted Poisson regression
sent the measurements at post and pre for stu-
analyses. We also conducted further explor-
dent j in school k, and PSGROUPjk is again a
atory analyses aimed at testing whether an
dummy variable identifying intervention con-
ASD classification moderated the effects of
dition (0 = comparison; 1 = peer support). The
peer supports on each of the outcome variables.
parameter β0k represents either the expected
In these analyses we added an effect coded
residualized gain score in school k for the
ASD variable (1 = ASD, –1 = not ASD), and a
comparison condition when controlling for
product variable between ASD and the peer
the pre score (model a), or the expected score
support dummy variable, as predictors in the
in school k for the comparison condition
student-level equations of the model. Our pri-
(model b). The parameter β1k indicates the
mary interest was in the significance of the
expected change in either of these scores for
product variable, as such an observation would
students in the peer support condition, and rjk
imply a differential intervention effect for stu-
is a residual term. At Level 2 (the school
dents with versus without ASD.
level), the Level 1 effects are modeled as

β0k = γ 00 +u0k Results


β1k = γ 10
Classroom Observations
( β 2k = γ 20 )
Descriptive statistics for all classroom obser-
where the equation in parentheses is only vation measures are displayed in Table 3,
included for analyses involving a pre score, alongside normative comparison data from
namely the first two outcomes. The parame- peers without disabilities. We focused a priori
ters γ00, γ10, and γ20 reflect the average residu- group comparisons on six of these measures.
alized gain (alternatively, the average Table 6 summarizes estimates of preinterven-
outcome) across schools in the comparison tion fixed effects across the outcome variables
condition, the average difference between as well as effect size estimates based on the
peer support and comparison students on the three-level HLM analysis. The two groups did
residualized gain (alternatively the average not differ significantly on any of these obser-
outcome), and the fixed effect of the preinter- vational measures at preintervention, as evi-
vention score on postintervention score, denced by the lack of significance of the γ010
respectively. Hypotheses regarding interven- coefficients.
tion effects attend to the significance of the γ10 In terms of postintervention effects, the
coefficient. fixed effect γ100 estimates the average change
We examined effect sizes in terms of a of the outcome variable from pre- to postinter-
Cohen’s d coefficient, where the unstandard- vention for the comparison group, whereas
ized coefficient for the peer support group at γ110 estimates the average difference in the
postintervention (the γ110 and γ10 coefficient pre- to postintervention change for the peer
estimates in the three-level and two-level support group relative to the comparison’s
224 Exceptional Children 82(2)

average postintervention change (see Table comparison group, this change was not signifi-
6). Students in the comparison group had sig- cant (γ100 = –0.06, p = .081). For students in the
nificant increases in total interactions across peer support group, proximity decreased by
the semester (γ100 = 13.10, p = .004); however, 19.5% to an average of 31.1% of the class
students in the peer support intervention had a period, but this change was not significantly
significantly greater change in total interac- different from that of the comparison (γ110 =
tions relative to the comparison group (γ110 = –0.13, p = .071; d = –0.38). Differences were
25.05, p < .001; d = 0.42). As shown in Table also noted in amount of time students were
3, a somewhat higher proportion of the inter- gone from the classroom. For the comparison
actions of students in the peer support group group, the amount of time students were gone
focused on task- versus social-related topics. from the classroom significantly increased by
Total interactions took place at slightly more 3.4% to 18.2% of the class period (γ100 = 0.04,
than half the rate of peer comparisons. Gains p = .024); for the peer support group, the per-
were not due solely to the interactive behavior centage of time students were gone decreased
of peers, as the contributions of the focus stu- significantly to 11.2% of the class period (γ110 =
dents to interactions also increased a signifi- –0.09, p < .001; d = –0.45). As noted earlier,
cantly greater amount in the peer support each of these analyses controlled for school
group relative to the comparison group (γ110 = effects related to these outcomes.
10.07, p = .005). Students receiving peer sup-
ports initiated an average of 6.3 times per hour
Student Skills
(versus 11.5 times per hour for peers) and
responded an average of 16.1 times per hour We focused our group comparisons on social
(versus 16.6 times per hour for peers), repre- skills ratings, social goal attainment, and aca-
senting a small to medium sized effect (d = demic goal attainment. Only the SSIS mea-
0.34). Observers rated the interactions as sure had a preintervention score and the two
fairly reciprocal, appropriate, response rele- groups did not differ significantly from one
vant, and of moderately high quality overall. another on this measure, t(97) = –0.76, p =
In terms of the number of different peers with .44. Improvements on social goals were sig-
whom the focus student had social contact nificantly higher for students in the peer sup-
during class periods, there was a significant port group than students in the comparison
gain in the comparison condition (γ100 = 0.39, group (γ10 = 0.80, p < .001; d = 0.79). On a
p = .007), but students in the peer support con- scale ranging from –2 to +2, mean scores for
dition again showed a statistically greater students in the comparison group were 0.25
increase (γ110 = 0.91, p = .001; d = 0.50). At compared to 0.98 for students in the peer sup-
postintervention, students in the comparison port group. Although improvements in aca-
and peer support groups interacted with an demic goals were also slightly higher for
average of 1.6 and 2.5 classmates per period, students in the peer support group versus
respectively. comparison group (0.55 versus 0.17), those
Academic engagement for students in the differences were not significant (γ10 = 0.38,
comparison group decreased by 3.9% at pos- p = .065; d = 0.35). No differences in improve-
tintervention to an average of 58.3% of the ments on the SSIS were found across groups
class period, but this change was not significant (γ10 = 2.33, p = .147; d = 0.17).
(γ100 = –0.04, p = .087). For students receiving
peer support arrangements, academic engage-
Social Participation
ment increased by 6.9% to an average of
71.1% of the class period, a significant Only the classroom social participation mea-
improvement relative to the comparison group sure had a preintervention score, and the
(γ110 = 0.11, p = .012; d = 0.31). Although prox- groups did not differ significantly from one
imity to direct support staff decreased by 6.4% another here, t(95) = –0.60, p = .55. Although
to an average of 42.3% for students in the students in the comparison group improved
Carter et al. 225

Table 6.  Fixed Effects of Peer Support Intervention.

Dependent variable Independent variable Estimate SE t-ratio df p value ESa


Total interactions Intercept, γ000 17.19 5.89 2.92 20 .008  
  PSGROUP, γ010 –5.15 7.11 –0.72 55 .472  
  POST, γ100 13.10 4.08 3.21 20 .004  
  POST*PSGROUP, γ110 25.05 7.12 3.52 55 <.001 0.42
Focus student contributions Intercept, γ000 8.38 2.77 3.02 20 .007  
  PSGROUP, γ010 –3.19 3.44 –0.93 55 .357  
  POST, γ100 6.96 2.12 3.27 20 .004  
  POST*PSGROUP, γ110 10.07 3.43 2.93 55 .005 0.34
Peers contacted Intercept, γ000 1.13 0.22 5.19 20 <.001  
  PSGROUP, γ010 –0.05 0.29 –0.19 55 .849  
  POST, γ100 0.39 0.13 3.01 20 .007  
  POST*PSGROUP, γ110 0.91 0.27 3.41 55 .001 0.50
Academic engagement Intercept, γ000 0.62 0.04 15.50 20 <.001  
  PSGROUP, γ010 0.01 0.04 0.33 55 .740  
  POST, γ100 –0.04 0.02 –1.80 20 .087  
  POST*PSGROUP, γ110 0.11 0.04 2.60 55 .012 0.31
Proximity to direct support Intercept, γ000 0.48 0.04 13.09 20 <.001  
  PSGROUP, γ010 0.03 0.05 0.60 55 .542  
  POST, γ100 –0.06 0.04 –1.84 20 .081  
  POST*PSGROUP, γ110 –0.13 0.07 –1.84 55 .071 –0.38
Gone from classroom Intercept, γ000 0.16 0.04 4.62 20 <.001  
  PSGROUP, γ010 0.00 0.03 0.02 55 .984  
  POST, γ100 0.04 0.02 2.45 20 .024  
  POST*PSGROUP, γ110 –0.09 0.02 –3.63 55 <.001 –0.45
Social skill ratings Intercept, γ00 24.04 3.96 6.07 20 <.001  
  PSGROUP, γ10 2.33 1.59 1.46 76 .147 0.17
  PRE_SSIS, γ20 0.73 0.05 15.74 76 <.001  
Social goal attainment Intercept, γ00 0.23 0.17 1.37 20 .186  
  PSGROUP, γ10 0.80 0.18 4.46 77 <.001 0.79
Academic goal attainment Intercept, γ00 0.17 0.17 1.04 20 .313  
  PSGROUP, γ10 0.38 0.20 1.87 76 .065 0.35
Classroom participation Intercept, γ00 –0.09 0.12 –0.76 20 .450  
  PSGROUP, γ10 0.48 0.17 2.18 74 .006 0.42
  PRE_ZScore, γ20 0.70 0.06 11.04 74 <.001  
Friendship gains Intercept, γ00 0.60 0.19 3.14 20 .005  
  PSGROUP, γ10 1.65 0.38 4.39 77 <.001 1.02

Notes. Final estimation of fixed effects with robust standard errors.


a
Cohen’s d using unstandardized regression coefficients.

slightly in their social participation scores by group gained significantly more friends with-
the end of the semester, their ratings did not out disabilities than students in the compari-
change significantly (γ00 = –0.09; p = .45). In son group (2.25 versus 0.60; γ10 = 1.65; p <
contrast, students in the peer support group .001; d = 1.02).
received significantly greater gains in ratings The extent to which relationships with peer
over the course of the semester relative to the partners maintained beyond the semester of
comparison group (γ10 = 0.48; p = .006; d = the intervention was somewhat high. Among
0.42). Similarly, students in the peer support the 106 peer partners who stayed involved
226 Exceptional Children 82(2)

until the end of the semester, 40% (n = 42) were well within their classroom. However, their
reported to have had an extended social contact readiness to use the strategies in the future
with the student during a 2-week window was more tempered (see Table 8).
toward the middle of the following semester
and 43% (n = 45) were reported as being
Discussion
friends. During the 1-year follow up 16% (n =
17) of these peer partners were reported to Despite decades of debate and discussion,
have had an extended social contact with the critical questions endure about how best to
student during a 2-week window toward the support the meaningful participation of stu-
middle of the semester and 40% (n = 42) were dents with severe disabilities in general edu-
reported to be friends. cation classrooms. Prevailing practice
involves assigning paraprofessionals as a pri-
mary pathway for supporting students to
Exploratory Analyses Related to ASD access instruction from the general education
To explore the extent to which intervention teacher. Yet an absence of empirical backing
impact might be different for students with for this support model—along with a host of
severe disabilities based on whether they had conceptual concerns—has challenged the
ASD. we also ran all HLM models studying field to explore alternative approaches. We
ASD as a moderator variable. Coefficients evaluated the efficacy and social validity of
and results of statistical tests were consistent peer support arrangements as one approach
with results from our primary analyses and for supporting the relationships and learning
failed to suggest many meaningful differences of adolescents with severe disabilities. Across
in the effects of the peer support intervention a large sample of students with and without
for students with ASD. In particular, the inter- disabilities attending a diverse set of class-
action terms related to the ASD and peer sup- rooms and schools, we found that school staff
port variables were consistently nonsignificant, implementation of peer support arrangements
suggesting the intervention had a similar offered distinct advantages over an exclusive
impact for students with and without autism. reliance on assistance from individually
Data are available by request from the authors. assigned special education staff. Findings
from this study extend the literature on sup-
porting general education access in several
Social Validity ways.
As shown in Table 7, the majority of focus First, the social and academic benefits of
students indicated they enjoyed working with peer support arrangements were fairly strik-
their peers, considered the peers their friends, ing. One perennial argument for the inclusion
and wanted to continue working together. of students with severe disabilities has been
Peer partners generally agreed that they felt the new communication and relationship
confident and effective in their roles, found it opportunities general education participation
easy to get their own work done, would rec- is presumed to introduce (e.g., Carter &
ommend this role for other peers, wanted to Brock, 2015; McLeskey et al., 2014). Peer
serve in this role again, and considered the support arrangements were found to substan-
focus student to be a friend. Overall, facilita- tially expand these opportunities by produc-
tors considered the implementation time to be ing significant increases in the interactions
reasonable, they reported understanding the students with disabilities had with peers, the
procedures, they were motivated to use the number of different classmates with whom they
intervention with other students, and they felt conversed, and the extent to which they made
participating students with and without dis- progress on social-related goals. Although prior
abilities benefited. General educators were studies have also documented improvements
also fairly positive, noting the amount of time in social interactions using within-participant
required was reasonable and the strategy fit comparisons (e.g., Shukla et al., 1998, 1999),
Carter et al. 227

Table 7.  Perspectives of Students With Severe Disabilities.

Focus student response

Social validity items Yes (%) No (%) Unsure (%) Unclear (%)
Do you like going to school? 84.0 12.0 0.0 4.0
Do you have friends at school? 90.0 4.0 2.0 4.0
Do you like this class? 86.0 4.0 4.0 6.0
Did you learn new things in this class? 90.0 4.0 4.0 2.0
Did you like working with [names of peer 94.0 0.0 2.0 4.0
supports] in this class?
Did working with [names of peer supports] 86.0 2.0 6.0 4.0
help you learn new things?
Are [names of peer supports] your friends? 94.0 0.0 2.0 4.0
Would you like to keep working with 92.0 0.0 4.0 4.0
[names of peer supports]?

Note. Data are missing for one student (n = 50). Numbers reflect percentage of focus students endorsing each option.
Unsure was recorded if the student communicated he or she was not certain of the answer. Unclear was recorded if
the student could not reliably communicate his or her perspective.

our use of a randomized design enabled us to and without severe disabilities and the empha-
document semester-long gains beyond those sis placed on exchanging academic-related
naturally occurring for students not receiving supports. Indeed, written peer support plans
targeted intervention. In other words, students outlined specific ways peers could encourage
in the comparison group also experienced participation in ongoing class activities. We
more peer interactions as the semester pro- emphasize that peers are not replacements for
gressed, albeit at modest levels (i.e., nearly instruction from classroom teachers, but
doubling their rate of interaction versus more instead support involvement in shared learning
than quadrupling for students receiving the opportunities provided by the classroom
intervention). teacher. Rather than pulling students with dis-
Prior single-case studies have provided a abilities away from instruction in pursuit of
mixed picture of the academic benefits of these social outcomes, working with peers appears to
interventions; some students maintained prior engage students more fully in activities planned
levels of academic engagement, while others by the general educator for all members of the
improved (e.g., Carter et al., 2007; Carter et al., class.
2011). In our study, toward the end of the
semester students involved in peer support Peers are not replacements for
arrangements were engaged in consistent activ- instruction from classroom teachers,
ities (i.e., the same activities provided by the
but instead support involvement in
classroom teacher to all students) significantly
more than students in the comparison group shared learning opportunities provided
(71% vs. 58% of the class period). Moreover, by the classroom teacher.
students spent significantly less time out of the
classroom and were rated by general educators Second, we examined more closely the
as having more active classroom participation. nature of students’ interactions and relation-
Such gains came even as the close proximity ships as part of peer support interventions. For
and direct academic assistance of special edu- example, our normative comparison data of
cation staff diminished. Increased engagement students without severe disabilities indicate
may be explained in part by the prominence of peer interactions—especially those focused
task-related conversations among students with on social-related topics unrelated to class
228 Exceptional Children 82(2)

Table 8.  Perspectives of Intervention Facilitators, General Educators, and Peer Partners.

General Peer
Facilitators educators partners
Social validity items (n = 51) (n = 49) (n = 106)
The student with a disability [my partner] benefitted socially 4.5 (0.7) 4.4 (0.8) 4.0 (0.7)
from having a peer support (e.g., talks more with peers, has
more friends).
The student with a disability [my partner] benefitted 3.7 (0.9) 3.6 (0.9) 3.8 (0.7)
academically from having a peer support (e.g., participates
more in class, learns new skills).
The peers without disabilities [I] benefitted socially from being 4.2 (0.7) 3.9 (0.7) 3.9 (0.8)
a peer support.
The peers without disabilities [I] benefitted academically from 3.4 (0.8) 3.4 (1.0) 3.5 (0.8)
being a peer support.
Overall, I enjoyed participating [being] in this project. 4.6 (0.5) 4.2 (0.7) 4.5 (0.6)
The amount of time required to use this strategy was 4.3 (0.6) 4.4 (0.7) —
reasonable.
I would need ongoing consultation to keep implementing this 2.2 (0.8) 3.0 (1.0) —
strategy.
Implementation of this strategy required considerable support 2.8 (1.1) 2.9 (0.9) —
from other school staff.
I am motivated to continue using this strategy. 4.4 (0.6) 4.0 (0.6) —
I would not be interested in implementing this strategy again. 1.8 (1.0) 2.2 (1.2) —
This strategy was a good way to address the educational 4.2 (0.7) 4.3 (0. 7) —
needs of the student with a disability.
This strategy fits well within this classroom. 4.1 (0.9) 4.3 (0.8) —
I understood the procedures of this strategy. 4.4 (0.6) 3.9 (0.9) —
I would know what to do if I was asked to implement this 4.6 (0.5) 3.7 (0.9) —
strategy again in the future.
The student with a disability has more friends as a result of 4.0 (0.9) 3.8 (0.9) —
this project.
This strategy negatively impacted other students in the class. 1.6 (0.7) 1.7 (0.8) —
I could use the strategies I learned through this project with 4.5 (0.5) 4.0 (0.7) —
other students.
The amount of time for record keeping with this strategy was 4.3 (0.5) — —
reasonable.
I feel I was effective in this role. 4.2 (0.7) — —
I implemented this strategy with a good deal of enthusiasm. 4.3 (0.6) — —
I often use cooperative learning strategies with students in my — 4.1 (0.7) —
classroom.
The peer support strategy would be feasible for me to — 3.3 (1.0) —
implement if additional school staff were not in my
classroom.
This strategy was a good way to address the educational — 3.8 (0.7) —
needs of students without disabilities.
At first, I was excited to become a peer partner. — — 4.2 (0.7)
I felt confident serving in this role — — 4.1 (0.7)
I had enough help from a teacher or teaching assistant (i.e., — — 4.4 (0.7)
paraprofessional) to do this role well.
This was too much work for me. — — 1.6 (0.8)
I feel I was effective in this role. — — 3.8 (0.7)
(continued)
Carter et al. 229

Table 8. (continued)

General Peer
Facilitators educators partners
Social validity items (n = 51) (n = 49) (n = 106)
It was easy to get my own work done while part of this — — 4.2 (0.7)
project.
The initial orientation meeting with a teacher/paraprofessional — — 4.2 (0.7)
was helpful.
Other students in the class should also do this. — — 4.2 (0.7)
I would be a peer support again in the future. — — 4.3 (0.6)
I understand why the teachers thought peer supports would — — 4.2 (0.7)
be helpful for my partner with a disability.
Our school should have more peer supports for students with — — 4.5 (0.7)
disabilities.
I consider my partner with disabilities to be a friend. — — 4.3 (0.6)
I would recommend being a peer support to my other friends. — — 4.3 (0.7)
My views about students with disabilities have changed for the — — 4.4 (0.7)
better.
I also spend time with other students who have similar — — 3.4 (1.0)
disabilities at my school.

Note. Adult wording [peer wording, if different]. Numbers reflect means (standard deviations) on a 5-point scale
ranging from strongly disagree (1) to strongly agree (5). A blank cell indicates the question was not asked of that
stakeholder.

content—may be fairly typical within high 2012). We found that “friendships” best charac-
school classrooms. Although overall rates of terized the relationships of substantial num-
peer interaction for students with severe dis- bers of participating students with and without
abilities receiving the intervention were disabilities. This conclusion was affirmed by
lower than those of their classmates without students with severe disabilities (94% of
disabilities, their rates of task-related inter- whom said their partners were their friends),
actions were comparable. In other words, the reciprocated by peer partners (96% of whom
intervention appears to facilitate content- agreed the focus student was a friend), and tri-
appropriate peer interactions in classrooms angulated in the social network assessments
already replete with potential interaction and end-of-semester questionnaires com-
opportunities. Questions that once were pleted by paraprofessionals, special educa-
directed to a paraprofessional may now be tors, and general educators. Still somewhat
extended to a peer, and feedback offered surprising, however, was the extent to which
only by a paraprofessional may instead be students with disabilities and their peers met
shared by classmates. together socially and were considered friends
Particularly intriguing are data speaking to after the intervention concluded and any
the types of peer relationships emerging from external expectations for connecting with one
these interactions. Questions have been raised another were removed. Amidst changing class
about the extent to which peer-mediated inter- schedules and the crossing of a summer, we
ventions will lead to friendships (Rossetti, 2011) were struck by the percentage of peer partners
and the longer-term impact of different educa- with whom students still had some social con-
tional support models has been largely unex- tact one and two semesters later, as well as the
plored (Ryndak, Alper, Hughes, & McDonnell, percentage of peer partners considered to be
230 Exceptional Children 82(2)

friends over this same time frame. Additional We also found that most students with dis-
research is needed to explore more closely abilities wanted to continue working with their
those factors leading to the formation and peers, most peer partners indicated an interest
maintenance of these relationships over time. in assuming this role again in the future, most
Third, available social validity and fidelity paraprofessionals were motivated to continue
data suggest peer support interventions are using the strategies, and most teachers said the
feasible to implement within secondary school intervention fit well within their classroom.
classrooms and are viewed favorably by Collectively, these stakeholders affirmed the
critical stakeholders. A long-standing concern benefits of these interventions for students with
within the field of special education has been disabilities, peers, and staff alike. When such
the degree to which interventions recom- strong endorsements are considered alongside
mended in the literature are doable and desir- the free cost and limited out-of-class time
able in real schools. We sought to equip school required to plan and implement these interven-
staff to lead in the planning and facilitation of tions, one might expect peer support arrange-
peer support interventions while minimizing ments to be widely adopted in secondary
the involvement of the research team. schools. Yet we see few examples of these
Although we provided the initial training of peer-mediated models implemented outside of
the intervention facilitators, attended the first formal research partnerships. Additional atten-
orientation meeting with peers, and were peri- tion should focus on understanding what spurs
odically present in the classroom, our role was and sustains the implementation of evidence-
to observe and offer coaching to the facilita- based strategies in schools and how best to
tors as they served as the primary imple- ensure effective interventions ultimately pene-
menter. All of the facilitators were successful trate practice.
in recruiting peers, carrying out initial orienta-
tion meetings, and supporting participating Additional attention should focus on
students as they worked together. Amidst the
understanding what spurs and sustains
fairly modest amount of training and coaching
we provided, these facilitators were ultimately the implementation of evidence-based
able to carry out the semester-long interven- strategies in schools and how best to
tions in ways that led to meaningful improve- ensure effective interventions
ments in social and academic outcomes. ultimately penetrate practice.
However, further research focused on the
“scaling up” of these interventions is still Fourth, a close examination of implementa-
sorely needed. Brock and Carter (2015) tion data highlights some of the complexities
showed that special educators could effec- and considerations associated with document-
tively guide paraprofessionals to implement ing intervention fidelity and delineating the
peer support arrangements without extensive critical elements of this approach. Peer sup-
researcher involvement. Yet increasing gen- port arrangements are flexible interventions
eral educators’ engagement with and active that can and should be adapted to meet the
responsibility for all students enrolled in their individualized needs of a student with a dis-
classes is critical to ensuring these interven- ability enrolled in a particular general educa-
tions will be implemented and sustained in tion classroom. Although we delineated the
typical schools. Although all general educa- core elements of this model (i.e., the recruit-
tors in our study were supportive of peer sup- ment, training, and proximity of peers; the
port arrangements, a stronger team-based provision of social and/or academic supports
approach may need to be incorporated at the by those peers; the facilitation of shared work
outset of this intervention to enable more by school staff), the specific supports peers
widespread adoption. and paraprofessionals provide can vary by
Carter et al. 231

student, activity, classroom, and day (see curriculum through this intervention. Fourth,
Tables 3 and 4). Thus, we described rather than we did not explore the effect of these interven-
prescribed these supports in an effort to charac- tions on the social and academic participation
terize more fully the implementation of peer of the peers who provided support. Although
supports under conditions that more closely peers said they personally benefitted from
resemble actual practice. This flexibility— involvement and affirmed the intervention did
reflected in the variability with which particu- not cause them to fall behind, additional stud-
lar intervention components and supports were ies are needed to better capture how the learn-
observed both within and across students— ing and social affiliations of peers are impacted
may enhance the social validity of this support by serving in this role. Finally, we did not collect
model. But it also leaves open the possibility data on the extent to which special educators,
that opportunities to optimize intervention paraprofessionals, and general educators col-
impact are being missed. Striking the best bal- laborated during the planning and implementa-
ance between intervention fidelity and indi- tion of peer-mediated and other classroom
vidualization is an important issue that supports throughout the semester. We share
warrants additional attention. concerns cited elsewhere in the literature
about the relative autonomy many paraprofes-
Limitations and Future sionals are given in relation to the education
of students with severe disabilities. Future
Research studies should explore how strengthening these
Future research is needed to address several collaborations affects implementation fidelity,
limitations of this study. First, friendship data student outcomes, and the long-term sustain-
were reported by adults rather than students. ability of these approaches in typical schools.
Although these findings were corroborated in
social validity ratings, respondents were asked
Conclusion
to consult students, and educators have a
prominent presence in the lives of students Findings from this study—in combination with
with severe disabilities; it is possible adults a growing number of published single-case
under- or overreported the social connections evaluations—challenge the burgeoning reliance
of students. Future studies are needed to on individually assigned adults as the primary
explore alternative approaches for document- avenue for supporting the active participation of
ing these social affiliations. Second, although adolescents with severe disabilities in general
this is the first study to look at the longer-term education classrooms. For secondary schools,
friendship outcomes of peer-mediated inter- this study describes one promising pathway for
ventions, our exploration of the generalization meeting mandates related to supporting access
of intervention outcomes was limited to this to the general education curriculum for students
measure. Future studies should explore how with severe disabilities. At the same time, it
more proximal social and learning outcomes should prompt reflection among school leaders
spill over to any other general education regarding the roles paraprofessionals are
classes in which the focus student is enrolled. asked—or left—to assume in the education of
Third, our ability to capture changes in learn- students with severe disabilities. Like all of their
ing was limited to two measures: observations classmates without disabilities, students with
of academic engagement and ratings of prog- severe disabilities should have well-supported
ress on a single academic-related goal. opportunities to learn from the deep knowledge
Although the present study revealed promis- and expertise of general educators. Paraprofes-
ing gains on both measures and addresses the sionals are not an appropriate substitute for
extent to which students are accessing the access to highly qualified educators—and nei-
general curriculum, future studies are needed ther are peers. Instead, peers and paraprofession-
to better document the extent to which stu- als together can play a valuable supplementary
dents are making progress in the academic role in helping students with severe disabilities
232 Exceptional Children 82(2)

access the rich learning and social opportunities support strategies. Research and Practice for
general education teachers provide in their Persons With Severe Disabilities, 31, 284–292.
classrooms. Carter, E. W., Moss, C. K., Hoffman, A., Chung,
Peer support arrangements are only one com- Y., & Sisco, L. G. (2011). Efficacy and social
validity of peer support arrangements for
ponent of high-quality inclusion for students
adolescents with disabilities. Exceptional
with severe disabilities. Future research is
Children, 78, 107–125. doi:10.1177/00144
needed to explore how to best to improve other 0291107800107
aspects of inclusion, including ongoing collabo- Carter, E. W., O’Rourke, L., Sisco, L. G., & Pelsue,
ration between general and special educators, as D. (2009). Knowledge, responsibilities, and
well as data-based instruction of goals related to training needs of paraprofessionals in elemen-
the general education curriculum and the stu- tary and secondary schools. Remedial and
dent’s individualized education program. Special Education, 30, 344–359. doi:10.1177/
0741932508324399
Carter, E. W., Sisco, L. G., Brown, L., Brickham,
References D., & Al-Khabbaz, Z. A. (2008). Peer inter-
Bandura, A. (1977). Social learning theory. actions and academic engagement of youth
Morristown, NJ: General Learning Press. with developmental disabilities in inclusive
Brock, M. E., & Carter, E. W. (2015). Efficacy of middle and high school classrooms. American
teachers training paraprofessionals to imple- Journal on Mental Retardation, 113, 479–494.
ment peer support arrangements. Exceptional doi:10.1352/2008.113:479-494
Children. Advance online publication. Carter, E. W., Sisco, L. G., & Chung, Y. (2012).
doi: 10.1177/0014402915585564 Peer-mediated support strategies. In P. A.
Carter, E. W., & Brock, M. E. (2015). Promoting Prelock & R. McCauley (Eds.), Treatment of
social competence and peer relationships. In F. autism spectrum disorders: Evidence-based
Brown, J. McDonnell, & M. E. Snell (Eds.), intervention strategies for communication and
Instruction of students with severe disabilities social interactions (pp. 221–254). Baltimore,
(8th ed., pp. 371–403). Upper Saddle River, MD: Brookes.
NJ: Merrill. Carter, E. W., Sisco, L. G., Melekoglu, M., &
Carter, E. W., Cushing, L. S., Clark, N. M., & Kurkowski, C. (2007). Peer supports as an
Kennedy, C. H. (2005). Effects of peer sup- alternative to individually assigned parapro-
port interventions on students’ access to the fessionals in inclusive high school classrooms.
general curriculum and social interactions. Research and Practice for Persons With
Research and Practice for Persons With Severe Disabilities, 32, 213–227. doi:10.2511/
Severe Disabilities, 30, 15–25. rpsd.32.4.213
Carter, E. W., Cushing, L. S., & Kennedy, C. H. Cohen, J. (1988). Statistical power analysis for the
(2009). Peer support strategies: Improving all behavioral sciences (2nd ed.). Mahwah, NJ:
students’ social lives and learning. Baltimore, Lawrence Erlbaum.
MD: Brookes. Feldman, R., Carter, E. W., Asmus, J., & Brock, M.
Carter, E. W., Hughes, C., Copeland, S. R., & E. (2015). Presence, proximity, and peer inter-
Breen, C. (2001). Differences between high actions of adolescents with severe disabilities
school students who do and do not volunteer in general education classrooms. Exceptional
to participate in peer interaction programs. Children. Advance online publication.
Journal of the Association for Persons With doi: 10.1177/0014402915585481
Severe Handicaps, 26, 229–239. Fisher, M., & Pleasants, S. L. (2012). Roles,
Carter, E. W., Hughes, C., Guth, C. B., & responsibilities, and concerns of para-
Copeland, S. R. (2005). Factors influencing educators: Findings from a statewide survey.
social interaction among high school students Remedial and Special Education, 33, 287–
with intellectual disabilities and their general 297. doi:10.1177/0741932510397762
education peers. American Journal on Mental Giangreco, M. F. (2010). One-to-one paraprofes-
Retardation, 110, 366–377. doi:10.1352/0 sionals for students with disabilities in inclusive
895-8017(2005)110[366:FISIAH]2.0.CO;2 classrooms: Is conventional wisdom wrong?
Carter, E. W., & Kennedy, C. H. (2006). Promoting Intellectual & Developmental Disabilities, 48,
access to the general curriculum using peer 1–13. doi:10.1352/1934-9556-48.1.1
Carter et al. 233

Giangreco, M. F., Carter, E. W., Doyle, M. B., & needs: K-12 inclusive-education research and
Suter, J. C. (2010). Supporting students with implications for the future. Inclusion, 1, 28–
disabilities in inclusive classrooms: Personnel 49. doi:10.1352/2326-6988-1.1.028
and peers. In R. Rose (Ed.), Confronting Schlosser, R. W. (2004). Goal attainment scaling as
obstacles to inclusion: International responses a clinical measurement technique in communi-
to developing inclusive schools (pp. 247–263). cation disorders: A critical review. Journal of
London, England: Routledge. Communication Disorders, 37, 217–239.
Gresham, F., & Elliot, S. (2008). Social Skills Schopler, E., Van Bourgondien, M., Wellman, G.,
Improvement System. San Antonio, TX: & Love, S. (2010). Childhood Autism Rating
Pearson. Scale (2nd ed.). San Antonio, TX: Pearson.
Hughes, C., Kaplan, L., Bernstein, R., Boykin, Scior, K. (2011). Public awareness, attitudes and
M., Reilly, C., Brigham, N., . . . Harvey, M. beliefs regarding intellectual disability: A sys-
(2013). Increasing social interaction skills of tematic review. Research in Developmental
secondary school students with autism and/ Disabilities, 32, 2164–2182. doi:10.1016/j.
or intellectual disability: A review of inter- ridd.2011.07.005
ventions. Research and Practice for Persons Shukla, S., Kennedy, C. H., & Cushing, L. S.
With Severe Disabilities, 37, 288–307. (1998). Adult influence on the participation
doi:10.2511/027494813805327214 of peers without disabilities in peer support
Kennedy, C. H., & Itkonen, T. (1996). Social rela- programs. Journal of Behavioral Education, 8,
tionships, influential variables, and change 397–413. doi:10.1023/A:1022801215119
across the lifespan. In L. Koegel, R. I. Koegel, Shukla, S., Kennedy, C. H., & Cushing, L. S.
& G. Dunlap (Eds.), Positive behavioral sup- (1999). Intermediate school students with
port: Including people with difficult behavior severe disabilities: Supporting their social
in the community (pp. 287–304). Baltimore, participation in general education classrooms.
MD: Brookes. Journal of Positive Behavior Interventions, 1,
McLeskey, J., Waldron, N. L., Spooner, F., & 130–140. doi:10.1177/109830079900100301
Algozzine, B. (Eds.). (2014). Handbook of Sparrow, S., Cicchetti, D., & Balla, D. (2005).
effective, inclusive schools: Research and Vineland Adaptive Behavior Scales (2nd ed.).
practice. New York, NY: Routledge. San Antonio, TX: Pearson.
Raudenbush, S. W., & Bryk, A. S. (2002). U.S. Department of Education. (2012). IDEA data.
Hierarchical linear models: Applications and Washington, DC: Author. Retrieved from
data analysis methods (2nd ed.). Thousand http://www.ideadata.org
Oaks, CA: SAGE.
Rossetti, Z. S. (2011). “That’s how we do it”:
Friendship work between high school stu- Authors’ Note
dents with and without autism or develop- Support for this research came from the Institute of
mental disability. Research and Practice for Education Sciences, U.S. Department of Educa-
Persons With Severe Disabilities, 36, 23–33. tion, through Grant R324A100391 to Vanderbilt
doi:10.2511/rpsd.36.1-2.23 University and the University of Wisconsin–Madi-
Ryndak, D. L., Alper, S., Hughes, C., & son. Authors are listed alphabetically from the
McDonnell, J. (2012). Documenting impact of fourth through final author. We appreciate the con-
educational contexts on long-term outcomes tributions of Stacey Blasko, Kristen Bottema-
for students with severe disabilities. Education Beutel, Mary B. Carlson, Lindsay Clark, Jeremy
and Training in Autism and Developmental Dietmeier, Rebecca L. Feldman, Blair Lloyd, Eliz-
Disabilities, 47, 127–138. abeth Schroeder, and Kate Szidon to this project.
Ryndak, D., Jackson, L. B., & White, J. M. (2013).
Involvement and progress in the general cur- Manuscript received December 2014; accepted
riculum for students with extensive support June 2015.
Copyright of Exceptional Children is the property of Sage Publications Inc. and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.

You might also like