Interventions to Improve Peer Interaction in Inclusive Preschool Classrooms: A Systematic Scoping Review

Conceptual Framework

We drew on sociocultural theory (Vygotsky, 1978) to frame the role of social interactions and active supports within children’s zone of proximal development (ZPD) and to characterize the potential mechanisms facilitating or impeding peer interactions that advance social inclusion for CWD. In addition, we utilized Brown and colleagues (2001) peer interaction intervention framework to analyze the types of peer interaction interventions applied. These framings informed the research questions, codebook development, and final analysis. Having a clear theoretical framing supports the coherence and transparency of systematic literature reviews (Cumming et al., 2023).

Vygotsky stated that children can exceed their current capacities and acquire new skills and perceptions through collective activities and interactions scaffolded by adults or “more capable” peers within their (ever-shifting) ZPD (Vygotsky, 1978). The purpose and process of such scaffolding, from a sociocultural perspective, is to facilitate children’s entrance, or inclusion, into their sociocultural context. That is, learning is an enculturation process by which adults and peers model and teach socially meaningful skills using society’s valued signs and tools (e.g., language and behavior; Vygotsky, 1978). Thus, how social interactions are supported (or not) can be viewed as one way that children learn what (and who) is valued in their society. Leonardo and Manning (2017) pointed out that applying a sociocultural lens helps explicate who is considered capable in society and what makes one capable. In the present literature review, applying a sociocultural lens helped us frame the importance of peer interactions and closely examine the social messages imbued in the ways peer interactions are conceptualized and supported (i.e., who and what are valued).

We used Brown and colleagues’ (2001) framework to categorize the intensity and type of peer interaction interventions in inclusive preschool classrooms (see Supplementary Materials—Figure 1). They assign various peer interaction interventions into four categories based on their purposes and level of intrusiveness (Tsao & Sung, 2014). Classroom-wide interventions are the least intrusive intervention type and include the use of developmentally appropriate practices to support social interactions, environmental arrangements (e.g., the availability of some types of social toys such as checkers), and affective interventions to improve the positive attitudes and perceptions of TDC toward CWD (e.g., using children’s books to talk about disability). Naturalistic interventions, the next most intrusive, include incidental teaching of social behaviors and friendship activities (Tsao & Sung, 2014). Incidental teaching is an individualized and contextualized intervention that uses naturally occurring opportunities during the routine of daily activities, such as outdoor play, snack time, or transitions, to prompt behavioral changes, such as appropriate social responses between interaction partners (Alai-Rosales et al., 2017). Friendship activities are teacher-planned activities (e.g., songs or games) intended to teach or embed prosocial behaviors, such as making friendly statements and sharing (Brown et al., 2001).

The next level of intervention in the hierarchy is social integration interventions. In social integration activities, interventionists can plan, arrange, and introduce the activity; assign roles; and monitor children’s interaction. However, they do not participate directly and gradually remove themselves from the activity. Finally, the most intense intervention is explicit social skills interventions. These interventions are adult-led and may involve children engaging in activities outside of typical classroom routines and activities or engaging in said activities in very different ways from what they naturally do. One common type of explicit social skills intervention is PMI, where a child without a disability is taught a particular behavioral chain to interact with a child with a disability in a very specific way (Milam et al., 2021). Brown and colleagues (2001) emphasized the need for less intrusive interventions (i.e., those located at the lower levels of the hierarchy) to reduce the research-to-practice gap. They also acknowledged that different types of interventions can be used together.

Article Search and Screening

Key terms that guided the article search and screening process are defined as follows. Peer interaction is the process of peers acting and influencing one another (e.g., initiation and responses, communicative turns, and play behaviors) through verbal or nonverbal means (Wu et al., 2021). “Inclusive preschool classroom” refers to classrooms where children aged 3 to 5 years with and without disabilities attend together. This definition was not attached to a particular teacher-child ratio or the ratio of children with and without disabilities. Peer interaction intervention means “a program, product, practice, or policy” aimed at improving peer interaction (What Works Clearinghouse, 2002).

In the original search, the first author did the electronic database search to identify articles from Education Research Complete, ERIC, and APA PsycInfo databases using the terms: (preschool* OR child* OR early childhood) AND (disab* OR special need* OR typically develop*) AND (peer interact* OR social interact* OR friend*) AND (interven* OR strateg* OR development). The first author also restricted the search to peer-reviewed articles written in English and published in academic journals between January 1, 2013, and February 1, 2023. To be included, articles had to meet the following criteria: (a) an intervention was conducted to increase peer interaction between children with and without disabilities, (b) the intervention occurred in inclusive preschool settings, (c) the interaction of children with and without disabilities between the ages of 3 to 5 was the primary dependent variable of the intervention, and (d) the study used experimental or quasi-experimental group design or single-case design. See the four-phase flow diagram for the article identification, screening, eligibility, and inclusion processes in Figure 1, adapted from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow (Moher et al., 2009).

OPEN IN VIEWER

Figure 1.

Systematic Review Flow Diagram.

The unit of analysis for screening was the published manuscript (i.e., article), meaning that the entire article had to meet the inclusion criteria, rather than individual research designs or participants within the study. The first author identified a total of 3,516 studies across all three databases and excluded 3,421 of these articles because they did not include an intervention focusing on increasing or improving peer interactions according to their titles and abstracts. After eliminating duplicates (n = 22), 73 studies remained. The first author then reviewed the full text of those studies and excluded 59 articles. Thirty of these fifty-nine articles were excluded due to their participants’ features. For example, the first author excluded articles that included children older than 5 years old (n = 22; e.g., Watkins et al., 2019), younger than 3 years old (n = 3; e.g., Martinez Cueto et al., 2022), and children considered to be “at-risk” but without disabilities (n = 5; e.g., Lemmon & Green, 2015). Also, the first author excluded 29 studies due to the interventions’ features (n = 29), such as interventions not being conducted in inclusive preschool settings (e.g., conducted in special education classrooms; n = 14; Bauminger-Zviely et al., 2020). This process resulted in the inclusion of 14 articles. Then, the first author conducted ancestral searches and a forward search of these articles, which resulted in and included two additional studies (Dueñas, D’Agostino, & Plavnick, 2021; Dueñas, Plavnick, & Goldstein, 2021).

The first author then replicated the original electronic database search using the same search terms and databases, covering publications from February 1, 2023, to June 1, 2025, to include the most current literature. As in the initial review, the unit of analysis for this screening was the published manuscript (i.e., article). This updated search yielded a total of 905 studies. After title and abstract screening, the first author excluded 872 articles for not including an intervention specifically targeting peer interactions. Following the removal of duplicates (n = 14), 19 articles remained for full-text review. Eight of these 19 articles were excluded due to their participants’ features, including children older than 5 years old (n = 7; e.g., Fedewa, Watkins, Carden, & Grbac, 2024) and children who do not have a diagnosed disability (n = 1, Foster et al., 2024). Eleven studies were excluded because the intervention did not focus specifically on peer interaction (n = 4; e.g., researchers intervened on CWD’s involvement in play activities and learning; Lundqvist, 2024). Studies were also excluded if they did not use an experimental, quasi-experimental, or single-case design (n = 4; e.g., Fedewa, Watkins, Barber, & Baggett, 2024). Finally, studies were excluded that were not conducted in inclusive preschool settings (n = 3; e.g., Edgar et al., 2024). Ultimately, no additional studies were identified for inclusion in the current review. The first author also reviewed the articles included in previous systematic reviews (Garrote et al., 2017; Huber & Carter, 2016; Therrien et al., 2016; Zagona & Mastergeorge, 2018), and after screening those articles against the inclusion criteria, she found that none qualified for inclusion in the current review.

Coding

We used a deductive coding process (Fereday & Muir-Cochrane, 2006) guided by a predefined codebook developed for this review. Coding consisted of extracting descriptive excerpts from each article, based on our research questions and purpose (Cumming et al., 2023). Although several manuscripts contained multiple single-case designs, we treated each study as the unit of analysis rather than each design. This decision aligns with the descriptive purpose of this review and with CEC (2014) guidance on assessing the quality of single-case designs. Importantly, for all studies that included more than one design, the designs used the same intervention category, type, and dependent variables; multiple designs served primarily to replicate effects rather than to test substantively different interventions. Therefore, treating the study as the analytic unit prevented overweighting studies with more replications without altering the interpretation of the review’s key findings. We used Microsoft Excel to code.

The articles were coded according to (a) descriptive information about the articles (e.g., title and publication date), (b) study method(s), and (c) participant characteristics for TDC and CWD separately (i.e., number, age, disability type, gender, race, home language, whether their home language is the same as the intervention language, and children’s inclusion criteria [requirements for participation] and developmental characteristics, as described in the articles [if the general inclusion criteria were reported for all participating children, it was listed; however, if there were no general inclusion criteria, the patterns among children’s development characteristics were summarized]; RQ1). In addition, the articles were coded for (d) intervention category according to Brown et al. (2001), which were classroom-wide, naturalistic, social integration, and explicit social skill interventions (RQ2), (e) intervention type and content (i.e., the name of the intervention as stated in the article and the content of the intervention for TDC and CWD; RQ2), (f) intervention context (i.e., the type of activity or routine in which the intervention was implemented; RQ2), (g) intervention agents (i.e., the individuals who implemented the intervention with the target child(ren) or trained the child(ren) to carry out the intervention steps; RQ2), and (h) dependent variables measured in the studies (i.e., participants about whom data was collected and the skills for which data was collected; RQ3). See the codebook in the Supplementary Materials. Although only single-case design studies were included in the final sample, the coding system was designed to accommodate single-case and group design studies.

Study quality was coded based on the Council for Exceptional Children’s (CEC; 2014) quality indicators for single-case design (SCD) research, as these standards are specifically designed to evaluate intervention studies targeting child outcomes in special education. The CEC standards include eight domains (i.e., context and setting, participants, intervention agent, description of practice, implementation fidelity, internal validity, outcome measures, and data analysis) encompassing 22 indicators specific to SCD studies. Each indicator was coded as “met” or “not met.” For each study, the number of indicators met was divided by the total number of applicable indicators (22) and multiplied by 100 to calculate the percentage of indicators met. The average percentage of indicators met was 89.57%, with a range from 72.72% (Hughett et al., 2013) to 100% (Driscoll & Carter, 2018; Dueñas, D’Agostino, & Plavnick, 2021). However, a full rigor analysis was beyond the scope of this review, given its descriptive purpose. This review also did not include a secondary visual or quantitative analysis of outcomes. Rather than independently evaluating the effectiveness of interventions, we coded the types of dependent variables assessed and the participant groups (CWD, TDC, or both) for whom data were collected.

Interrater Agreement and Consensus Procedures

The first author coded all included articles. After the first author coded the articles, she selected the six articles (37.5%) that were the most difficult to code due to factors such as ambiguous language, complex findings, or challenges in applying the existing coding framework, and the second author also coded those articles. The first two authors discussed discrepancies to come to a consensus on the codes and code definitions. For reliability purposes, the third author was then assigned one randomly selected article to complete the training process. She independently coded this article using the codebook. After achieving 100% agreement with the first author on this training article, the third author independently coded five additional randomly selected articles (31.25% of all included studies). To calculate interrater agreement (IRR), agreements were defined as both coders identifying or describing the same information for each code. The IRR was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplying by 100. The IRR scores were computed both by article and by coding variable. The IRR by article averaged 95%, ranging from 91.30% to 100%. The IRR by coding variable also averaged 95%, with individual code-level agreement ranging from 60% (e.g., dependent variables for CWD) to 100%. Due to the small number of double-coded articles (n = 5), IRR percentages for individual codes should be interpreted with caution, as single disagreements can substantially affect the percentage. The overall IRR across all articles and codes was 95%.

Finally, following Shamseer and colleagues (2015), the first and third authors discussed all remaining discrepancies and reached a consensus on the final codes. In combination with documenting initial IRR based on independent coding, reaching consensus through discussion supports the trustworthiness and validity of the findings (Corr et al., 2020). For all the IRR and consensus procedures, the first and third authors met three times for 1 hour each meeting.

Final Analysis

Following Rodgers et al. (2022), we used a descriptive analysis approach for final analyses to identify patterns in the coded excerpts, enabling us to answer our research questions and provide a holistic picture of the studies’ methods and intervention practices. Data (i.e., coded excerpts) were organized separately for CWD and TDC to identify patterns in participant characteristics (RQ1), intervention content and context (RQ2), and dependent variables (RQ3). We then looked at patterns within and across these groups of participants.

Participant Characteristics

See Supplementary Materials—Table 1 for a summary of participant characteristics.

Intervention Category, Type, and Content

We first present intervention categories based on Brown and colleagues’ (2001) framework, as previously described: classroom-wide, naturalistic, social integration, and explicit social skill interventions. Then, we report intervention type and content for each group of children (TDC and CWD), including intervention agent and context. Table 1 presents the results.

OPEN IN VIEWER

Table 1.

Intervention Characteristics.

Study	Intervention category	Intervention type and content for each group of children	Intervention agent	Intervention context
Driscoll & Carter (2018)	Classroom-wide intervention	EA: TDC and CWD: Manipulating the variety of available activities (8 vs. 4 activity conditions)	Researchers	Play activities
Lee et al. (2022)	Naturalistic intervention	Cooperative physical activity = TDC and CWD: The physical activity, including prompting, peer partnering, and an interdependent group contingency (stickers + praise)	Graduate students + teachers	Physical activity
Lee & Lee (2015)	Classroom-wide Intervention + Explicit Social Skills Intervention	EA (modified snack routine and good friend board) + PMI = TDC: social initiation training and correspondence training + Prompted to initiate to CWD when necessary, CWD: Promoted to respond to a peer initiation when necessary	Researchers (trained teachers and TDC) → Teachers (trained TDC and applied EA) → TDC (used initiation skills, appropriate gestures and language, prompted CWD, and directed CWD’s attention)	Daily snack routine
Coogle et al. (2019)	Explicit social skills intervention	PMI = TDC: Prompted to use choice-making toward CWD, CWD: Prompted to respond to the peer when necessary	Undergraduate research assistant	Choice time
Oh-Young et al. (2018)	Explicit social skills intervention	PMI + Video Modeling + Constant time delay and least-to-most prompting = TDC: Verbal social interaction task analysis training + Acting in peer video models + Learning to deliver prompting and praise to CWD, CWD: Watching peer-video model + Receiving peer or teacher-delivered prompts	Researchers (trained teachers and TDC) → Teachers (trained TDC and applied intervention strategies) → TDC (delivered prompting and social interaction strategies to CWD in PN; acted in and modeled target behaviors in PVM videos)	Center time
Katz & Girolametto (2013)	Explicit social skills intervention	PMI = TDC and CWD: Attending social skills training, including reading a book and using communication boards + attending facilitated play sessions	Researchers (trained teachers and both CWD and TDC) → Teachers (trained both CWD and TDC and supervised the play sessions)	Play activities
Hughett et al. (2013)	Explicit social skills intervention	PMI = TDC and CWD: Buddy skills package with SPT training, including visual cards and adult prompts	Paraeducator	Play activities
Thiemann-Bourque et al. (2017)	Explicit social skills intervention	PMI = TDC: SPT training, including using an SGD + Prompted, when necessary, CWD: Prompted, when necessary to use SGD	Researchers (trained TDC and implemented intervention) → TDC (delivered SPT strategies and SGD use to CWD)	Center time
Dueñas et al. (2019)	Explicit social skills intervention	PMI = TDC: Play partner training + joint video modeling, CWD: Joint video modeling with peer	Researcher (trained TDC, delivered video modeling, prompted attention to video) → TDC (delivered modeled interaction strategies to CWD during play sessions after training)	Play activities
Osborne et al. (2019)	Explicit social skills intervention	PMI = TDC and CWD: SPT training + Group contingencies + Recorded reminders	Graduate Students	Play activities
Severini et al. (2019)	Explicit social skills intervention	PMI = TDC and CWD: SPT training with modeled AAC use + token reinforcement	Researchers + Graduate Students	Play activities
Dueñas, Plavnick, & Goldstein. (2021)	Explicit social skills intervention	PMI = TDC: Initial peer training + Self-management with video self-review, CWD: Video modeling + Prompting when necessary	Researchers (Implemented all intervention strategies, including TDC training) → TDC (delivered trained interaction strategies)	Play activities
Dueñas, D’Agostino, & Plavnick. (2021)	Explicit social skills intervention	PMI + Video Modeling = TDC: Training to make a bid to play + Video modeling + Prompting to make a bid to play, CWD: Prompting to accept a bid to play	Researchers (trained the classroom staff, behavior technicians, and TDC) → Classroom staff (presented video models to TDC + Prompting) →TDC (applied PMI strategies to invite CWD to play) + Behavior technicians (Prompted to CWD)	Play activities
Therrien & Light (2016)	Explicit social skills intervention	iPad with an AAC app with visual scene displays and a dyadic turn-taking training = TDC and CWD: Turn-taking training by using an iPad	Researchers	In a small room—not a specific activity type
Stanton-Chapman & Brown (2015)	Explicit social skills intervention	Social-communication intervention: TDC and CWD: Advanced play organizer training, including vocabulary instruction, role assignment, storybook-based social skills modeling (adjacency, initiation, responding), and guided practice	Researchers	Play activities
Mancil et al. (2016)	Explicit social skills intervention	Mand training using two communication devices: CWD: Mand training using Dynavox and iPod Touch as AAC devices	Researchers (Trained children and teachers) → Teachers (trained children)	Play activities

Note. TDC = typically developing children; CWD = children with disabilities; PMI = peer-mediated intervention; SPT = Stay, Play, Talk intervention; EA = environmental arrangements; SGD = speech-generating device; AAC devices = augmentative and alternative communication devices; PN = Peer Networking; PVM = Peer Video Modeling.

Intervention Outcome

Table 2 in the Supplementary Materials provides an overview of dependent variables. Of the 16 included studies, 10 reported data for both TDC and CWD (e.g., Coogle et al., 2019), 5 reported data only for CWD (e.g., Driscoll & Carter, 2018), and one reported data only for TDC (Mancil et al., 2016).

Among the 10 studies that reported data for both TDC and CWD, three of them measured the same dependent variables for both TDC and CWD, such as reciprocal social interactions (Lee & Lee, 2015) or turns in extended interactions (Katz & Girolametto, 2013). In the other seven studies, researchers used different measures to assess each group’s participation in peer interactions. For TDC, dependent variables commonly captured their use of the practices they were taught to prompt interaction with CWD, similar to fidelity checks. For example, researchers assessed how many times TDC provided choices to CWD (Coogle et al., 2019), the percentage of stay, play, and talk behaviors (Severini et al., 2019), and the number of invitations extended to CWD (Dueñas, D’Agostino, & Plavnick, 2021). Dependent variables reported for CWD tended to focus on their responsiveness to peer prompts and participation in reciprocal interaction, such as weighted communication, gestures, and vocalizations (Coogle et al., 2019), acceptance of play invitations (Dueñas, D’Agostino, & Plavnick, 2021), and the duration and frequency of cooperative play (Hughett et al., 2013).

Among the five studies that reported data regarding dependent variables exclusively for CWD, researchers assessed their use of specific skills. Two studies assessed the frequency and/or percentage of verbal and nonverbal social interactions (e.g., Lee et al., 2022), while another measured both scripted and unscripted verbalizations alongside play actions (Dueñas et al., 2019). Three studies examined various aspects of play behaviors, including parallel play (Stanton-Chapman & Brown, 2015) and social-cognitive play (Driscoll & Carter, 2018). In sum, dependent variables for CWD primarily focused on language and play skills as indicators of their peer interactions.

Representation of Child Participants

The majority of CWD in the reviewed studies had an ASD disability label. This finding aligns with other literature reviews on peer interaction interventions that included older children (e.g., Garrote et al., 2017). Although children with ASD may experience challenges with social skills and social isolation, children with other disability labels, such as intellectual disabilities, physical disabilities, visual impairments, and hearing impairments, may also face exclusion or limited peer interactions (e.g., Chen et al., 2019). Therefore, future research on peer interaction interventions should be intentional about including children with other disability labels, abilities, and support needs.

Most studies did not provide information about children’s race or language. For the studies that did provide information about children’s race, most children were White. This is consistent with other peer interaction intervention reviews on PMI (Huber & Carter, 2016). In addition, the few studies that reported children’s home language generally reported it to be English. Considering the increasing number of bilingual learners (National Center for Education Statistics, 2022) and the possible effect of cultural and linguistic differences on peer interaction (e.g., Lalvani & Bacon, 2019), researchers should more comprehensively report children’s demographics, including race and language. In addition, since no study reported the intervention language, the match between children’s home language and intervention language was not analyzed. While we can assume interventions were provided in English, this reporting gap limits our understanding of how language may shape children’s experiences and outcomes in peer interaction interventions. More diverse linguistic representation and reporting both home and intervention language can help the field understand for whom peer interaction interventions are effective and the adjustments that may need to be made to serve a diverse group of children.

Prevalence of PMI

Our review found that explicit social skills interventions, particularly PMI, were the most common type of peer interaction intervention. PMI is considered an evidence-based practice to support peer interactions (Brock & Huber, 2017). Because PMI has most widely been studied with elementary and secondary students, researchers have called for more research on its use with preschool-age children (Zagona & Mastergeorge, 2018), and this review describes the ways PMI has been implemented with preschool children in the studies we identified.

It is important to note that PMI in the studies included in the present review were conducted to increase a narrow set of dependent variables, particularly the use of certain types of peer prompts for TDC and language and play skill development for CWD. While such skill development can support peer interactions, as prior research has pointed out, the skills of CWD are not the only factor that impacts peer interactions (e.g., Diamond & Tu, 2009; Hong et al., 2020). Therefore, to more robustly support peer interactions, researchers should consider and assess what PMI may need to be supplemented with to support outcomes that influence peer interactions beyond those identified in the current review. They may particularly take into account the understanding and attitudes of TDC toward CWD (Diamond et al., 2008; Hong et al., 2013; Yu et al., 2015) and broader environmental supports, including classroom climate and routine, and teacher facilitation strategies (Acar et al., 2017; Hong et al., 2020).

Intrusiveness of Interventions

Most of the included studies were explicit social skills interventions that required direct teaching of social skills and behaviors, the most intensive type of peer interaction intervention. Although they might be effective at improving the measured dependent variables, because these interventions involve directly teaching social behavior to TDC or CWD (Milam et al., 2021), they can be demanding and time-intensive and interrupt typical classroom activities and routines (Brown et al., 2008). The PMI, specifically, requires considerable adult prompting and intensive specialized training for interventionists (Tsao & Sung, 2014). Perhaps because of this, the studies in the present review relied heavily on researchers to administer interventions.

Based on the studies included in this review, it is difficult to determine whether less-intensive or classroom-wide supports were already in place before individualized interventions (i.e., explicit social skills interventions) were introduced, as this information was not reported. Future research should therefore clearly describe the practices occurring prior to intervention implementation, including during baseline, to help readers understand whether individualized interventions were built upon existing classroom-wide supports. Also, while we recognize the importance of being responsive to children’s individual support needs and implementing explicit social skills interventions when necessary, more research is needed about less-intensive interventions (i.e., naturalistic and classroom-wide interventions). In the current review, few studies examined naturalistic or class-wide interventions to support peer interaction. Although such approaches are often preferred by teachers (Hugh et al., 2022) and have been shown to be effective for child outcomes when implemented by teachers in inclusive classrooms (Hemmeter et al., 2021), only a small subset of the studies in this review evaluated their impact.

The Role of Teachers, TDC, and CWD in Peer Interaction Interventions

Teachers are key facilitators of peer interactions in inclusive classrooms (e.g., Acar et al., 2017). However, few studies have involved teachers implementing peer interaction interventions or providing such support. This is consistent with findings from previous literature reviews on peer interaction interventions (e.g., Therrien et al., 2016). With little information about how teachers and other classroom staff can be supported to promote peer interactions, more research is needed that focuses on improving teachers’ awareness, the effective implementation of peer interaction interventions, and their training experiences. Previous literature suggests that classroom staff can be trained to implement interventions effectively. For instance, Camargo and colleagues (2016) compared the effect sizes of studies in which researchers implemented behaviorally based social interaction interventions versus those in which teachers implemented the intervention. The authors found no differences between the two groups. Future research should incorporate more naturalistic intervention agents, particularly teachers and paraprofessionals, and investigate how to best support teachers’ skills in this area.

Meanwhile, some studies in this literature review, especially studies of PMI, positioned TDC as intervention agents expected to deliver intervention strategies to their peers with disabilities. This set-up can inadvertently perpetuate a kind of “helper-recipient of the help” relationship that can reify inequality in children’s friendships (Therrien et al., 2016). This may be particularly troubling given the previously described deficit perceptions TDC may hold toward their peers with disabilities (e.g., Yu et al., 2015). Scholars have long called for researchers and practitioners to attend to TDC’s understanding of disability, equity, and fairness to further improve peer interactions and friendships between children with and without disabilities (e.g., Diamond & Tu, 2009; Diamond & Hong, 2010). Yet, in the current literature set, such dispositions and skills were not addressed in the intervention procedures or dependent variables measured. Little, if any, attention was paid to the perspectives of the TDC toward disability or their peers with disabilities. Moreover, TDC were often taught separately from CWD and taught to interact with them in largely scripted ways to solicit specific behaviors from the CWD. Future research could expand the dependent variables supported and measured in peer interaction interventions to include perceptions and additional skills for TDC.

The differential positioning of children with and without disabilities was also evident in the participant descriptions of the groups included in the interventions. The CWD were largely included and described based on perceived deficits in peer interaction, social skills, language, and communication. We did not identify any studies that included CWD based on their interests or strengths in peer interactions, though researchers in six studies expected CWD to show certain skills related to attention, motor movements, and language to participate in the study. In contrast, TDC were generally included in the study if they showed social, language, or interaction strengths, and, with the exception of one study, they were described based on their strengths and interests in interacting with their peers. Thus, TDC were fairly explicitly positioned as the “more capable peer” in terms of Vygotsky’s ZPD, and while their interests regarding peer interactions were routinely referred to, those of CWD were not. While it can be assumed that CWD were selected for these studies because they had few peer interactions and/or support needs in related skills, that does not preclude the development and assessment of interventions grounded in their existing abilities and interests related to peer interactions. Indeed, these findings point to the need for more research that bases peer interaction intervention on the strengths, social interaction preferences, and agency of CWD; this can support more strengths-based and inclusive practice. To that end, our findings raise an important question that researchers should ask as they consider the next steps for this literature base: Is the purpose of peer interaction interventions only to improve skills for CWD or to create an inclusive classroom that promotes the full membership of all children? If we shift focus to the latter, we must reconsider the ways CWD are understood, described, and supported in both research and practice.

Finally, the perceived role of CWD in peer interactions is further evident in the fact that dependent variables were more often measured for CWD than for TDC. In almost all reviewed studies, the skills or behaviors of CWD were assessed, whereas five studies did not collect data on TDC. It is not inherently problematic for the outcomes of a population to be measured that did not directly receive intervention (e.g., Hemmeter et al., 2021). However, it is important to recognize that CWD were always expected to show changes in behavior and skills, while TDC were not consistently assessed to ensure they displayed the skills taught to support peer interactions. Thus, CWD were more often held responsible for the status of their peer interactions and whether peer interactions increased in frequency, length, or certain features.

A sociocultural analysis of these findings calls us to question who is being valued as capable in these scenarios (primarily TDC) based on whose cultural tools are being employed and upheld (nondisabled ways of being and communicating). Accordingly, to further support peer interactions, researchers should consider what explicit skill instruction could be supplemented with to build on the skills and desires of CWD, promote equal-status relationships (Therrien & Light, 2016), and break the pattern of social exclusion that research continues to document for CWD (e.g., Chen et al., 2020). In combination with other literature that has revealed the dispositions and skills TDC need to support peer interactions with CWD (e.g., theory of mind, understanding of disability and equity; Diamond & Hong, 2010; Lalvani & Bacon, 2019), the current analysis indicates the need for researchers to (re)consider (a) who is in need of skill development to support peer interactions; (b) what skills need to be supported (beyond the social and communication skills of CWD, and even beyond the communication initiation skills of TDC); and (c) what outcomes are targeted and measured for interventions intended to support peer interactions. By expanding how peer interaction interventions are conceptualized and implemented, researchers can support all children’s abilities while ensuring more equitable interactions between children with and without disabilities.