Investigating the Impact of Embedded Learning Opportunities on the Engagement of Children With Autism and Intellectual Disability

ABSTRACT

Background

Engagement is essential for fostering learning and development in young children with disabilities, including those with autism spectrum disorder (ASD) and intellectual disability (ID).

Methods

This study used a non‐concurrent multiple‐baseline across participants design to examine the effects of embedded learning opportunities (ELOs) on engagement and learning in four preschool‐aged children diagnosed with ASD and ID.

Results

Results showed a consistent increase in engagement for each child following the introduction of ELOs by teachers, with children generalising this engagement to other settings and sustaining high engagement levels during follow‐up. Additionally, enhanced engagement was linked with improvements in independent performance of target behaviours. Social validity data from teachers further supported the intervention's effectiveness.

Conclusions

These findings highlight ELOs as a promising and individualised approach to enhancing engagement and learning outcomes for preschool children with dual diagnoses of ASD and ID.

Summary.

• Embedded learning opportunities (ELOs) significantly increased engagement and learning in children with dual diagnoses of autism spectrum disorder (ASD) and intellectual disabilities (ID).

• ELOs demonstrated the potential for generalisation, with children applying acquired skills to various classroom activities.

• Both experienced teachers and student teachers strongly endorsed the acceptability and effectiveness of ELOs, highlighting their potential as valuable tools for enhancing engagement and learning in inclusive preschool settings.

• Individuals with dual diagnoses of ASD and ID can benefit significantly from ELOs, which can enhance their engagement and, subsequently, their learning experiences.

1. Introduction

Active engagement plays a pivotal role in fostering skill development and promoting meaningful participation in inclusive educational environments (Meindl et al. 2020; Rakap 2013). For children diagnosed with both autism spectrum disorder (ASD) and intellectual disability (ID), achieving and maintaining appropriate levels of engagement during classroom activities can be a complex endeavour (Bellini et al. 2007; Koegel et al. 2012). This complexity arises from various factors, including communication and social interaction difficulties, heightened sensory sensitivities and individualised preferences that can vary widely among these children (Hodges et al. 2020; Thye et al. 2018). Consequently, children with this dual diagnosis often find it challenging to naturally engage in activities within inclusive preschool settings (Rakap 2013).

Communication and social interaction difficulties represent a major obstacle to active engagement (Koegel et al. 2012). Children with dual diagnoses often face significant challenges in expressing their thoughts, needs and desires effectively (Howlin 2000). Such difficulties can lead to frustration, withdrawal or even disruptive behaviours as a way of compensating for limited expressive abilities (Oliver et al. 2022; Waizbard‐Bartov et al. 2023). For example, a child may struggle to request a toy, seek help from a teacher, or respond to peer interactions, which hinders their ability to participate meaningfully in shared activities. Without effective communication, initiating or sustaining social exchanges, following multi‐step instructions, or navigating social cues becomes a considerable challenge. These communication barriers contribute to a sense of social isolation and impede opportunities for engagement with peers and adults (Ng et al. 2016), making it essential to embed communication supports within classroom routines.

Sensory processing differences, particularly heightened sensory sensitivities, can significantly influence engagement in children with ASD and ID. These children often exhibit atypical responses to sensory input such as noise, lighting, textures or movement (Joosten and Bundy 2010; Mills et al. 2016). For example, a child may become overwhelmed by the sound of a vacuum cleaner or fluorescent lights in the classroom or may avoid activities involving tactile materials like finger paint or sand. Such sensitivities can lead to withdrawal from group activities or behavioural outbursts, further limiting the child's access to learning opportunities (Leekam et al. 2011). To address these challenges, interventions like embedded instruction that consider individual child characteristics such as sensory preferences and sensitivities can be particularly effective. Because embedded instruction is implemented within natural routines and is highly individualised (Brodzeller et al. 2018), it allows educators to adapt the timing, setting and materials of instruction to reduce sensory overload and promote greater engagement in inclusive settings (Meindl et al. 2020; Rakap 2013).

In addition to communication and sensory challenges, children with ASD and ID frequently demonstrate strong individualised preferences and a reliance on routines, which can directly affect their level of engagement (Keen 2009; Meindl et al. 2020; Ruble and Robson 2007). These preferences may manifest as intense interests in specific topics, toys or activities, while disruptions to established routines can lead to distress or disengagement (Leekam et al. 2011). For example, a child might become fixated on lining up cars or require a consistent schedule to transition smoothly between activities. When classroom environments fail to accommodate these preferences and routines, children may resort to repetitive behaviours or exhibit resistance to participation. Without intentional planning and support that align with the child's interests and need for predictability, educators may struggle to maintain engagement. Embedded instruction offers a flexible, individualised approach that aligns with children's preferences and routines to reduce resistance, boost motivation and support sustained engagement in inclusive settings (Rakap 2013).

1.1. Engagement

Engagement has been defined as the amount of time children spend interacting with their social and non‐social environments at different levels of competence and in a developmentally and contextually appropriate manner (McWilliam and Bailey 1992). Researchers have consistently emphasised the importance of engagement, describing it as a necessary condition for promoting development and learning (De Kruif and McWilliam 1999; Pianta et al. 2012; Raspa et al. 2001; Skinner and Pitzer 2012). Behavioural engagement, a particularly observable form of engagement, involves children's active participation and involvement in learning activities. It reflects their attention, effort and persistence, often including positive attitudes toward tasks and the absence of disruptive behaviours (Keen 2009). Behavioural engagement is not only essential for academic performance but also influences children's broader cognitive and emotional experiences. For young children with special needs, engagement is especially critical, as it serves as a foundational skill in early intervention settings (Bevill et al. 2001; Ridley et al. 2000; Strain et al. 1995). Engagement acts as a prerequisite for learning, helping children build essential skills and fostering resilience in classroom activities.

Despite the recognised importance of engagement in early childhood development, much of the existing empirical research has primarily examined how contextual variables such as disability type, service delivery models, environmental arrangements and teacher–child interactions influence engagement (e.g., Coelho et al. 2019; Hojnoski et al. 2020; Losh et al. 2022; Sam et al. 2016; Williford et al. 2013). While these descriptive studies provide valuable insights, fewer studies have explored the effectiveness of specific interventions in actively promoting engagement among young children with disabilities, particularly those with ASD and ID.

Among the interventions that have demonstrated potential, several approaches have shown positive impacts on engagement across different learning contexts. Adult‐mediated strategies, such as joint attention and time delay (Malmskog and McDonnell 1999) and instructional supports like picture cues and reinforcement (Bevill et al. 2001), have been linked to increased engagement during play and free‐choice activities. Similarly, video self‐modelling has been shown to enhance peer‐related social engagement, especially when paired with peer training components (Bellini et al. 2007, 2016). Other studies have emphasised the role of interest‐based strategies, such as embedding children's preferences into learning tasks, which have been effective in promoting sustained attention and proximity to peers (Koegel et al. 2012).

Additionally, interventions that target engagement through structured group play and physical interaction have yielded promising results. For instance, imitation and play expansion strategies have been found to improve both the complexity and duration of engagement during block play (Barton et al. 2018), while sensory and physical activity interventions have supported engagement by reducing distractions and improving focus during teacher‐directed tasks (Ledford et al. 2016; Pokorski et al. 2019). Physical activity as a precursor to learning tasks has also been associated with improved on‐task behaviour and responsiveness (Harbin et al. 2022).

Embedded instruction, in particular, has been highlighted as a promising naturalistic strategy. Studies such as VanDerHeyden et al. (2005) have demonstrated that embedding instructional trials into classroom routines can increase engagement to levels comparable to typically developing peers, especially when engagement is measured through meaningful interactions with peers or materials. More recent work (e.g., Kretzmann et al. 2015; Bateman et al. 2023) continues to highlight the potential of contextually relevant interventions for enhancing engagement in inclusive environments.

Collectively, these studies reflect a growing recognition of engagement as a multidimensional construct that can be actively shaped through diverse intervention strategies. However, the literature also reveals several ongoing gaps. Many interventions have focused either on general developmental delays or solely on autism, without considering the unique challenges faced by children with co‐occurring ASD and ID. Furthermore, relatively few studies have investigated how evidence‐based interventions like embedded instruction can be systematically implemented by preschool teachers within inclusive classroom routines. There remains a pressing need for research that not only addresses the dual challenges of ASD and ID but also examines practical, individualised strategies that are feasible within real‐world preschool settings.

1.2. Embedded Instruction

Embedded instruction is among the approaches that can be used in inclusive preschool settings to support the development and learning of young children with disabilities, including those with ASD and ID. This approach involves integrating learning opportunities into children's daily routines and activities, guided by the fundamental principle that young children with disabilities learn most effectively when engaged in meaningful, functional activities that are part of their daily lives (Horn and Banerjee 2009). Preschool teachers can harness the power of embedded instruction by infusing learning opportunities throughout their ongoing classroom activities. This approach empowers children with disabilities to develop essential knowledge and skills in a meaningful context, promoting engagement and motivation. As learning becomes a natural and relevant part of their daily routines, children are more likely to generalise these newfound skills to other settings, fostering a holistic developmental process (Frazuer‐Cross et al. 2004; Rakap et al. 2024). Furthermore, embedded instruction is inherently individualised, capable of being tailored to meet the unique needs and abilities of each child. It offers comprehensive support across a wide spectrum of skills, encompassing communication, social development, cognitive abilities and physical skills (Gulboy et al. 2023; Rakap and Parlak‐Rakap 2011; Snyder et al. 2015). Additionally, it actively promotes the principles of inclusive education, facilitating the participation of all children, regardless of their abilities, in shared and meaningful learning experiences (Jimenez and Kamei 2015).

Embedding instruction within and across daily activities and routines has long been recommended practice in the field of early childhood special education (Division for Early Childhood 2014). A recent systematic review of the literature underscores its status as an evidence‐based practice (EBP) to support the development and learning of young children with disabilities in inclusive settings (Gulboy et al. 2023). This meta‐analysis concludes that, when implemented by practitioners during ongoing activities in early childhood settings, embedded instruction proves to be an EBP practice for supporting children with disabilities in acquiring new skills across various learning domains, encompassing language, pre‐academic and social skills. Furthermore, the findings from these reviews underscore the substantial benefits of embedded instruction, as it not only facilitates skill acquisition but also promotes the generalisation of these skills across different settings, activities and individuals. Despite these valuable insights, it is noteworthy that none of the studies included in this review specifically examined the impact of embedded instruction on the engagement of young children with disabilities.

1.3. Purpose of the Study and Research Questions

The purpose of this study, therefore, is to investigate the impact of embedded instruction on the engagement levels of preschool‐aged children diagnosed with ASD and ID. Engagement plays a pivotal role in fostering learning and social development in children with these complex needs, and the identification of effective instructional strategies is paramount for enhancing their educational outcomes. This research aims to assess the effectiveness of embedded instruction as a tailored and individualised approach to augment engagement in children with dual diagnoses and to explore its potential for long‐term effects and generalisability across diverse classroom settings. The following questions guided the present study: (1) How does embedded instruction influence the engagement levels of preschool‐aged children with dual diagnoses of ASD and ID? (2) To what extent does the increased engagement observed during embedded instruction sessions generalise to other settings and is maintained over time? (3) To what extent is the increased engagement observed during embedded instruction sessions maintained over time? (4) What is the relationship between increased engagement levels and improvements in independent demonstration of target behaviours in preschool‐aged children with dual diagnoses of ASD and ID during embedded instruction? (5) What are the perceptions and experiences of teachers and caregivers regarding the effectiveness and feasibility of implementing embedded instruction for enhancing engagement in children with dual diagnoses?

2. Methods

2.1. Participants

2.1.1. Children

Four children with dual diagnosis of ASD and ID enrolled in inclusive preschool classrooms located in four different public schools participated in this study. Each child was nominated by their teachers as displaying a lower‐than‐desired level of engagement during classroom activities. Matt, a 61‐month‐old child, presented with mild to moderate delays in the domains of language and communication, cognitive functions and adaptive behaviours. Matt benefited from pull‐out speech and occupational therapy services, with 45 and 60 min of therapy per week, respectively. During play activities, Matt often engaged in solitary or parallel play, struggling to meaningfully participate in small or large group activities. He displayed a tendency to wander around during centre activities without actively engaging with the social or physical aspects of the environment. Before the study, Matt had been in the inclusive preschool classroom for 4 months, consistently attending school for 6–7 h a day, 3–5 days a week. His ABILITIES Index (Simeonsson and Bailey 1991) score was 34, reflecting mild to moderate delays in five of the nine measured areas.

Jason, a 54‐month‐old boy, grappled with mild to moderate delays encompassing cognition, language and communication and adaptive behaviours. He received 1 h of pull‐out speech and occupational therapy services each week. Throughout free play activities, Jason frequently engaged in solitary play and aimlessly wandered around the classroom, with the exception being when he was engrossed in activities involving blocks or computer use. Before the study, Jason had been in the inclusive preschool classroom for 2 months, consistently attending school for 5–7 h a day, 3–5 days a week. His ABILITIES Index score was 36, indicating mild to moderate delays in four of the nine areas assessed.

Charlie was a 56‐month‐old child with mild to moderate delays in communication, adaptive behaviours and cognitive domains. He received 1 h of pull‐out speech and physical therapy services weekly. Charlie encountered difficulties in sustaining his engagement during classroom activities that extended beyond a few minutes. Transitioning from one activity to another proved challenging, often necessitating external support. In the face of these challenges, he frequently exhibited temper tantrums when asked to commence a new activity. Charlie had been a part of the inclusive preschool classroom for 2.5 months prior to the study, consistently attending school for 7 h a day, 4–5 days a week. His ABILITIES Index score of 38 reflected mild to moderate delays in four of the nine areas assessed.

Troy, at the age of 60 months, grappled with mild to moderate developmental delays across language and communication, adaptive behaviours and cognitive domains. He received 1 h of pull‐out speech and occupational therapy services each week. During play activities, Troy frequently engaged in solitary play, either with the same toys or materials, or he would aimlessly wander around the classroom during centre activities, lacking meaningful engagement with materials, peers or adults. Before the study, Troy had been part of the inclusive preschool classroom for 3 months, attending school regularly for 5–7 h a day, 4–5 days a week. His ABILITIES Index score was 43, indicating mild to moderate delays in four of the nine areas assessed.

Before the study commenced, the Gilliam Autism Rating Scale‐3 (GARS‐3; Gilliam 2013) and Stanford‐Binet Intelligence Scale‐5 (SB‐5; Roid 2003) assessments were administered to determine the autism index and intelligence scores of the participating children. The GARS‐3 scores for Matt, Jason, Charlie and Troy were 81, 77, 92 and 98, respectively, indicating substantial support needs. Meanwhile, their scores on the SB‐5 were 67, 65, 62 and 62, respectively, indicating mild ID. Although we did not administer an adaptive behaviour measure specifically for this study, the school staff conducted the Vineland Adaptive Behaviour Scales‐3 (Sparrow et al. 2016) within 45 days of the study's onset as part of their standard assessment procedures. The Adaptive Behaviour Composite score was 54 for Matt, 51 for Jason, 55 for Charlie and 59 for Troy, which are well below the normative mean of 100, indicating significant challenges in adaptive functioning across these children.

2.1.2. Intervention Providers

Four female preschool teachers implemented embedded instruction practices throughout the study. These educators served as the lead teachers in the classrooms of the participating children. All four held bachelor's degrees in early childhood education and fell within the age range of 24–33 years (M = 29.5 years). They had 2–12 years of overall teaching experience, including 2–6 years of specific experience working with children diagnosed with ASD and ID. Moreover, these teachers possessed prior training and experience in implementing embedded instruction practices within ongoing activities of inclusive preschool settings.

2.2. Setting, Materials and Instructional Arrangements

The current study occurred in four inclusive preschool classrooms located in different public preschool programmes. These classrooms accommodated 12–16 preschool children, including 2–3 children with disabilities. Alongside the lead teacher, a student teacher was present in each classroom throughout the school day and across the study. The training sessions for preschool teachers were conducted on a one‐on‐one basis in a dedicated meeting room situated within the preschool programmes where the teachers were employed. During the study, teachers utilised available classroom activities and materials to deliver embedded learning opportunities (ELOs) to child participants based on their individual goals (see Table 1) on the days when children were present. No Supporting Information were introduced into the classrooms by the research team for the use of either teachers or children during the study.

TABLE 1.

Goals by developmental area targeted during embedded instruction.

Child	Cognitive goal	Language/communication goal	Motor/adaptive goal	Social/emotional goal
Matt	Sort objects by shape and colour during playtime to improve sequencing abilities	Improve expressive language by using simple sentences to describe actions and objects during play	Increase fine motor skills through activities like threading beads or using tongs to transfer items	Take turns during small group play activities
Jason	Develop spatial awareness and reasoning by building complex structures with blocks	Articulate needs and choices using words during play activities	Improve hand‐eye coordination through computer games that require mouse or touchscreen manipulation	Initiate requests for pieces and responding to others during block‐building activities
Charlie	Follow simple instructions/directions during circle time to improve attention	Initiate short dialogues with adults and peers, using prompts and visual aids	Throw/catch balls to improve gross motor coordination	Transition between activities or settings by following visual schedules and responding to prompts
Troy	Identify different shapes and colours during sorting activities	Express his wants and needs using words or visuals	Improve fine motor skills by cutting with scissors or manipulating small objects	Maintain a play activity with peers for 3 min

Preschool teachers implemented embedded instruction practices during the morning centre‐time/free‐play activities, which usually lasted 30–45 min. During this child‐directed activity, children could choose the activity centre or area to play and move between them without permission as long as the area they moved in had enough space. Each classroom provided, at minimum, options for pretend play, art, literacy, blocks, sensory and discovery centres that the children could select from. Generalisation sessions were conducted during mealtime routines (snack or lunch) or small group activities, which often lasted 25–30 min across classrooms. During the mealtime routines, the children sat on child‐sized furniture and ate either the food they brought from home or meals provided by the school. The children were encouraged to actively participate in the mealtime routine by setting the table, serving food to each other, pouring juice or milk, spreading jam or butter on bread and taking responsibility for cleaning up once they were finished eating. Small group activities consisted of three to four children engaging in various activities, including board games, counting and sorting exercises, letter‐related tasks, as well as drawing and colouring.

2.3. Response Definitions and Recording Procedures

2.3.1. Child Engagement

The Individual Child Engagement Record‐Revised (ICER‐R; Kishida et al. 2008) was developed to quantify the engagement behaviours of participating children. The ICER‐R includes four codes for engagement type (active engagement, passive engagement, active nonengagement and passive nonengagement). Active engagement refers to a state in which a child is fully involved and participating in a learning or play activity. During active engagement, the child is typically focused, responsive and actively involved in manipulating materials, interacting with peers or adults, or engaging in verbal communication. Examples of active engagement in a preschool setting include a child who is actively manipulating art or craft materials during free play, conversing and interacting with peers during mealtime, or responding to questions posed by the teacher during group activities. Passive engagement is characterised by a child interacting with the learning environment but in a more reserved or less active manner. During passive engagement, the child may observe or take in the environment without engaging in active manipulation or verbalisation. For example, a child may quietly watch the teacher or stimulus materials during group time, observe a peer while they eat during mealtime, or watch other children playing during free play. While they are engaged to some extent, their level of activity or interaction is more subdued compared to active engagement. Active nonengagement occurs when a child is involved in behaviours that do not contribute to the intended learning or play activity. These behaviours may include repetitive actions or aggressive behaviour. For instance, a child may engage in repetitive actions like spinning an object repeatedly, or they may display aggressive behaviour such as hitting or pushing peers. Additionally, active nonengagement can involve behaviours like wandering around the room during structured routines, playing with unrelated objects like shoelaces during story time, or any other actions that do not align with the intended activity. Passive nonengagement is observed when a child's attention is directed away from the intended learning or play activity. This might involve the child looking at something outside of the group context or focusing on a peer rather than engaging in the planned activity. For example, passive nonengagement can occur when a child is looking at a picture on the wall during group time or watching a peer playing with a toy instead of actively participating in the group or play activity. In passive nonengagement, the child may not actively disrupt the activity but is not engaged in a productive way.

2.3.2. Child Responses to ELOs

The Embedded Teaching Coding Form (ETCF; Rakap and Balikci 2017) was utilised to assess children's progress on individualised goals during embedded instruction sessions. This form includes four distinct codes to classify child responses: an unprompted correct response, where the child performs the target behaviour correctly within 3 s of the teacher's antecedent; a prompted correct response, where the child performs the behaviour accurately after a prompt following the antecedent; an incorrect response, where the child exhibits a different behaviour in reaction to the antecedent or prompt; and no response, where the child does not display any behaviour after the antecedent or prompt is given.

2.4. Experimental Design, Dependent and Independent Variables

A non‐concurrent multiple‐baseline across participants design (Watson and Workman 1981) was used to examine the effects of embedded instruction (independent variable) on the engagement behaviours (dependent variable) of children diagnosed with ASD and ID. Consistent with the multiple‐baseline design, the intervention was introduced systematically for one child at a time, with a staggered number of sessions for each participant. The non‐concurrent design offers the necessary flexibility for intricate social contexts while upholding experimental control by demonstrating the presence of stable data across all participants prior to the intervention and establishing functional relationships once the intervention was systematically introduced (Harvey et al. 2004).

2.5. Procedures

2.5.1. Baseline

Prior to the first baseline session, teachers were asked to provide a list of four goals they were currently working on with the target children. During the baseline sessions, teachers were instructed to follow their regular procedures for teaching skills related to the target children's IEP goals. No additional guidance or information on embedded instruction practices was given to the teachers at this stage. A minimum of five baseline sessions were conducted for each child. These baseline sessions were carried out to evaluate the initial engagement levels of the child participants before their teachers began the systematic implementation of embedded instruction practices. Teachers' implementation of ELOs was also examined.

2.5.2. Intervention

Before the first intervention session, teachers individually participated in a 2‐day training that was designed based on the behavioural skills training model, including instruction, modelling, rehearsal and feedback (Reid and Parsons 1995). For the instruction phase, teachers were provided with a comprehensive training manual (Rakap 2017) outlining the principles and components of ELOs 1 week prior to the session and asked to review it. During the 3‐h‐long initial training session, researchers focused on the components of ELOs and how to implement them during the centre‐time/free‐play activities to teach skills associated with target children's individualised education programme goals.

An ELO encompasses an antecedent that initiates the learning opportunity, setting the stage for a child to exhibit the desired behaviour. Following the child's correct execution of the desired behaviour, a consequence occurs immediately, reinforcing the child's performance (Macy and Bricker 2006; Rakap and Balikci 2023). For example, during playtime, the teacher arranges a variety of objects (e.g., blocks, toy animals and cars) in different shapes and colours within the child's play area. As the child begins to interact with these items, the teacher provides an antecedent by saying, “Matt, please find all the red blocks and put them together” This prompts the child to start sorting based on colour. Once the child correctly sorts the red blocks, the teacher provides positive reinforcement by saying, “Great job finding all the red blocks, Matt! Now, can you find the square ones?” This sequence continues, prompting the child to sort items by shape or colour and reinforcing each correct action.

During the modelling phase, researchers demonstrated the practical application of ELOs by showing teachers how to implement them with children in their classrooms. This involved role‐playing scenarios to illustrate the correct introduction, reinforcement and monitoring of ELOs. During the second day of the training, rehearsal and feedback, teachers had the opportunity to practice implementing ELOs within the ongoing activities and routines of their classrooms with another child who did not participate in the study. Researchers observed and provided immediate feedback and guidance. This allowed teachers to refine their skills and address any concerns.

At the end of the training session, the researcher and teachers collaboratively reviewed the IEP goals of the target children in their respective classrooms. They collectively identified four specific skills that could be embedded into the daily classroom activities. A planning matrix was developed to ensure a minimum of 10 ELOs were implemented across the four target skills during the centre‐time/free‐play activities. Following the training, teachers were instructed to use ELOs during centre‐time/free‐play activities at least 3 days a week and employ the teacher's guide throughout the study. They were also asked to submit their weekly planning matrix to the research team on the first day of school each week until each child reached the criterion level of performance. Intervention sessions continued until each child attained the performance criterion, defined as engagement (active + passive) in at least 80% of the observation intervals across three consecutive sessions. Of the three sections where the child was engaged at least 80% of the time, the percentage of intervals in which the child demonstrated active engagement had to be higher than passive engagement.

2.6. Generalisation and Maintenance

Three data collection sessions were conducted to evaluate whether the engagement behaviours of target children were generalised across other settings within the preschool classrooms. While the first session was conducted during the baseline phase, the second session was conducted when a child was engaged at least 80% of the time in a second intervention session. The final generalisation session was conducted on the same day as the 4‐week maintenance session. Across‐setting generalisation was evaluated during mealtimes for Matt and Troy and during small group activities for Charlie and Jason. Maintenance data collection sessions were conducted 2, 4 and 8 weeks after the last intervention session for each child.

2.7. Social Validity

An adapted version of the Intervention Rating Profile (IRP; Martens et al. 1985) was employed to assess teachers' perceptions about the acceptability and effectiveness of embedded instruction practices upon study completion. This form consisted of 25 items, and participants rated them on a 6‐point Likert‐type scale, with response options ranging from 1 (strongly disagree) to 6 (strongly agree).

2.8. Data Collection and Analysis

Data in relation to child engagement and learning were collected by videotaping the first 20 min of the morning centre‐time/free‐play activities once a week, usually on Wednesday or Thursday. The ICER‐R was applied to videotaped data to determine children's engagement levels. The ICER‐R utilises a 15‐s momentary time sampling (MTS) method, which has been demonstrated in research to provide more accurate estimates of engagement compared to other time sampling procedures such as partial‐interval recording or whole‐interval recording (Prykanowski et al. 2018; Wood et al. 2015). With the MTS method, the observer examines the 15th second of each interval and records only the engagement behaviour occurring at that specific moment. Due to the nature of the MTS coding, engagement codes were mutually exclusive. At the end of each session, the percentage of intervals in which each specific engagement behaviour was observed is calculated and reported. The ETCF was used to analyse videotaped data to assess children's learning related to individualised goals listed in Table 1. In each data collection session, the percentage of unprompted correct responses across four target behaviours was calculated based on the total number of embedded instruction trials presented. To measure children's learning, the mean percentage of unprompted correct responses was calculated and reported for each phase of the study.

Visual analyses of graphed data were conducted to assess the impact of embedded instruction on the engagement behaviours of young children with ASD and ID. Visual analysis entails a thorough visual examination of the graphed data, focusing on elements such as level, trend, variability within and across phases, and overlap and immediacy of effects between phases (Barton et al. 2018). To enhance our analysis, we also calculated and reported a nonparametric effect size estimate, Tau‐U, as a supplementary measure (Parker et al. 2011; Rakap 2015).

2.9. Interobserver Agreement (IOA)

Two special education graduate students were trained to code study videos employing the ICER‐R and ETCF by the first author. These students were kept blind to the specific purpose and conditions of the study to ensure unbiased observations. Each student was assigned the role of lead observer for two of the participating children while acting as the secondary observer for the other two children. To establish IOA, secondary observers coded a minimum of 33% of the videos within and across all phases for every participant. For the reliability assessment, videos were randomly chosen, and the primary observers remained unaware of which videos were coded by the secondary observers, eliminating potential bias. To ensure coding accuracy, coders were required to achieve a minimum of 80% agreement with the researcher on the master codes after independently coding three training videos before proceeding to code the study videos. The point‐by‐point method was then employed using the following formula: (total number of agreements/the total number of agreements + disagreements) × 100.

The IOA results for ICER‐R were consistently high across all phases. Specifically, for ICER‐R coding during the baseline phase, the mean IOA score was 94% (range = 91–100) for Matt, 91% (88–100) for Jason, 92% (88–98) for Charlie and 97% (95–100) for Troy. In the intervention phase, the mean IOA score was 93% (88–100) for Matt, 96% (94–100) for Jason, 94% (86–100) for Charlie and 94% (90–100) for Troy. For the generalisation phase, the mean IOA scores were 96% (94–98) for Matt, 97% (91–100) for Jason, 92% (86–100) for Charlie and 95% (90–100) for Troy. In the maintenance phase, the mean IOA scores were 94% (91–97) for Matt, 94% (90–100) for Jason, 94% (90–98) for Charlie and 93% (89–96) for Troy.

For ETCF coding, IOA results were similarly high across all phases. In baseline, the mean IOA score was 92% (range = 89–98) for Matt, 90% (87–98) for Jason, 93% (89–97) for Charlie and 96% (94–99) for Troy. In the intervention, the mean IOA score was 91% (87–99) for Matt, 94% (90–98) for Jason, 92% (88–97) for Charlie and 93% (89–98) for Troy. For generalisation, the mean IOA score was 95% (92–98) for Matt, 96% (90–99) for Jason, 91% (87–96) for Charlie and 94% (91–98) for Troy. In the maintenance, the mean IOA scores were 93% (90–96) for Matt, 92% (89–97) for Jason, 93% (90–95) for Charlie and 91% (88–95) for Troy. These results show a high level of consistency in coding accuracy across observers for all study phases, indicating the reliability of the ICER‐R and ETCF coding.

2.10. Procedural Fidelity

Procedural fidelity data were collected to gauge the degree to which teacher training sessions (n = 4) adhered to the research's predetermined plans and whether teachers used ELOs with precision. A special education graduate student was present at all teacher training sessions, meticulously following a procedural fidelity checklist to validate the precise execution of training activities. Procedural fidelity scores were computed using the subsequent formula: (total number of correctly executed components/the total number of components listed in the checklist) × 100. The results clearly demonstrated that the researcher executed all planned training activities during these sessions, achieving a procedural fidelity score of 100%.

Procedural fidelity in relation to teachers' implementation of ELOs was assessed using the videos collected to evaluate child engagement. Two trained coders reviewed the videotapes and determined whether teachers implemented the component of ELOs accurately. This included delivering the antecedent, providing consequences following the child's correct response or using an error correction procedure when an incorrect or no response was provided. The mean procedural fidelity score was 95%, 94%, 97% and 95%, respectively, for teachers with a range of 90%–100%.

3. Results

3.1. Effect on Child Engagement

The percentage of engagement for each child is shown in Figure 1 and Table 2. Matt's mean percentage of engagement during baseline was 41%, evenly distributed between active (21%) and passive (20%) engagement. As illustrated in Figure 1, his engagement performance remained relatively stable during the baseline (range = 38%–45%). Following the onset of the intervention, there was an immediate increase in his engagement, which continued to exhibit an upward trend throughout the intervention. In the ninth session, he achieved the performance criterion with a mean engagement score of 79%, ranging from 68% to 89% during this phase. During the intervention, his active engagement surpassed passive engagement in eight of the nine sessions. During the maintenance phase, his engagement ranged between 76% and 82%, with a mean of 79%. Matt demonstrated generalisation across settings, as his engagement increased from 35% in the baseline to 68% and 65% during the intervention and maintenance, respectively. Overall, Matt's data showed no overlapping data points across baseline and intervention or maintenance, resulting in a Tau‐U score of 1.0.

FIGURE 1.

Percentage of intervals engaged by children and study phases. Note: Maintenance data were collected 2, 4 and 8 weeks after the intervention was concluded.

TABLE 2.

Mean percentage of engagement by study phases and children.

Participant/engagement	Baseline	Intervention	Maintenance
Matt
Active engagement	21.0 (16–25)	50.2 (37–62)	44.0 (39–49)
Passive engagement	19.6 (16–29)	28.6 (14–40)	34.7 (33–37)
Active non‐engagement	36.4 (23–41)	9.4 (4–15)	9.7 (9–10)
Passive non‐engagement	23.0 (15–37)	11.8 (7–19)	11.7 (9–14)
Total engaged	40.6 (38–45)	78.8 (68–89)	78.7 (76–82)
Total non‐engaged	59.4 (55–62)	21.2 (11–32)	21.3 (18–24)
Jason
Active engagement	19.9 (17–25)	45.7 (28–61)	44.3 (40–49)
Passive engagement	17.9 (14–23)	27.7 (20–34)	28.7 (26–32)
Active non‐engagement	36.1 (26–42)	14.2 (6–26)	12.3 (10–15)
Passive non‐engagement	26.1 (20–37)	12.5 (5–20)	14.7 (13–18)
Total engaged	37.7 (32–44)	73.3 (59–89)	73.0 (72–75)
Total non‐engaged	62.3 (56–68)	26.7 (11–41)	27.0 (25–28)
Charlie
Active engagement	11.9 (7–20)	44.1 (26–65)	38.3 (34–45)
Passive engagement	17.2 (10–30)	29.4 (15–41)	37.3 (34–40)
Active non‐engagement	37.8 (31–51)	12.4 (4–20)	11.0 (8–14)
Passive non‐engagement	33.1 (17–44)	14.2 (4–26)	13.3 (10–18)
Total engaged	29.1 (23–37)	73.4 (55–90)	75.7 (74–79)
Total non‐engaged	70.9 (63–77)	26.6 (10–45)	24.3 (21–26)
Troy
Active engagement	16.5 (10–21)	41.1 (28–53)	48.7 (42–53)
Passive engagement	16.7 (8–23)	29.8 (13–40)	33.3 (26–40)
Active non‐engagement	26.8 (22–32)	14.9 (6–26)	8.7 (6–13)
Passive non‐engagement	40.0 (30–55)	14.3 (5–25)	9.3 (7–11)
Total engaged	33.2 (22–44)	70.8 (53–86)	82.0 (77–87)
Total non‐engaged	66.8 (56–78)	29.2 (14–47)	18.0 (13–23)

Jason exhibited a mean engagement level of 38% during the baseline, with active (20%) and passive (18%) engagement distributed fairly evenly. His engagement remained relatively stable throughout the baseline, with percentages fluctuating within a range of 32%–44%, as depicted in Figure 1. Upon the introduction of the intervention, there was an immediate and consistent increase in Jason's engagement. This upward trend continued, and in the 12th session, he met the performance criterion with a mean engagement score of 73%, which ranged from 59% to 89% throughout the intervention phase. Over the course of the intervention, his active engagement exceeded passive engagement in 10 out of 12 sessions. During the maintenance, Jason's engagement ranged from 72% to 75%, with a mean of 73%. He displayed generalisation across settings, with his engagement rates increasing from 30% during the baseline to 73% and 60% during the intervention and maintenance, respectively. There was no overlap between data points in Jason's baseline and intervention or maintenance, resulting in a Tau‐U score of 1.0.

Charlie's engagement data revealed a mean engagement level of 29% during the baseline, with passive engagement (17%) surpassing active engagement (12%). Throughout the baseline, his engagement exhibited some variability, ranging between 23% and 37%. With the introduction of the intervention, Charlie's engagement immediately increased and maintained an upward trajectory. By the 14th session, he reached the performance criterion with a mean engagement score of 73%, ranging from 55% to 90% during the intervention. Notably, in 11 out of 14 sessions during the intervention, his active engagement surpassed passive engagement. During the maintenance, Charlie's engagement ranged from 74% to 79%, with a mean of 76%. His ability to generalise the acquired skills across settings was evident, with engagement rates increasing from 37% during the baseline to 68% and 64% during the intervention and maintenance, respectively. There was no overlap between data points in Charlie's baseline and intervention or maintenance, resulting in a Tau‐U score of 1.0.

Troy displayed a mean engagement level of 33% during the baseline, with active (17%) and passive (17%) engagement distributed evenly. Throughout the baseline, Troy's engagement exhibited variability, fluctuating between 22% and 44%. Upon the introduction of the intervention, there was an immediate increase in Troy's engagement, which maintained an upward trajectory. By the 16th session, he met the performance criterion with a mean engagement score of 71%, ranging from 53% to 86% during the intervention. Significantly, in 13 out of 16 sessions during the intervention, his active engagement exceeded passive engagement. In the maintenance, Troy's engagement ranged from 77% to 87%, with a mean of 82%. His ability to generalise the acquired skills across settings was evident, with engagement rates increasing from 40% during the baseline to 75% and 71% during the intervention and maintenance, respectively. Importantly, there was no overlap between data points in Troy's baseline and intervention or maintenance, resulting in a Tau‐U score of 1.0.

3.2. Effect on Child Learning

As illustrated in Table 3, during the baseline phase, Matt, Jason, Charlie and Troy had low mean percentages of unprompted correct responses in relation to target behaviours: 11% (range 7–15), 9% (5–20), 8% (4–14) and 8% (0–12), respectively. With the introduction of the intervention, each child showed substantial improvement. Matt's mean rose to 82% (65–100), Jason's to 85% (72–100), Charlie's to 80% (68–100) and Troy's to 86% (71–100). In the maintenance phase, high performance was sustained across all children, with mean percentages of 85% (80–92) for Matt, 81% (75–90) for Jason, 84% (77–89) for Charlie and 91% (85–97) for Troy. These results indicate significant gains from baseline to intervention, with each child maintaining strong performance in the maintenance phase.

TABLE 3.

Mean percentage of children's unprompted correct responses across individualised goals by study phases.

	Baseline	Intervention	Maintenance
Matt	11 (7–15)	82 (65–100)	85 (80–92)
Jason	9 (5–20)	85 (72–100)	81 (75–90)
Charlie	8 (4–14)	80 (68–100)	84 (77–89)
Troy	8 (0–12)	86 (71–100)	91 (85–97)

Matt demonstrated steady progress across all four of his individualised goals. For his cognitive goal, sorting objects by shape and colour, he improved from 10% unprompted correct responses during baseline to 85% by the final intervention session. His language goal, describing actions using simple sentences, increased from 7% at baseline to over 90% during intervention. In the motor/adaptive domain, threading beads or transferring items with tongs rose from 15% to 100%. Finally, in the social–emotional domain, his ability to take turns during group play increased from 12% to 86%. These improvements were maintained during follow‐up, with performance across all four goals ranging from 80% to 92%.

Jason also showed consistent growth across his goals. Initially, he had difficulty engaging in structured block‐building (cognitive goal), with only 5% unprompted responses; by the end of the intervention, his accuracy had reached 89%. His ability to articulate needs during play rose from 9% to 85% unprompted correct responses. For his motor goal, involving touchscreen games requiring hand‐eye coordination, he advanced from 13% to 100%. His social–emotional goal, initiating and responding during peer block play, improved from 8% at baseline to 82% by the final intervention session. These gains were sustained during maintenance, with all skills remaining above 75%.

Charlie began the intervention with relatively low performance across all goals, averaging between 4% and 14% unprompted correct responses. His cognitive goal—following instructions during circle time—rose from 8% to 84%. His communication goal, initiating short dialogues using prompts, increased from 6% to 90%. His ability to catch and throw balls for gross motor coordination rose from 14% to 100%. For his transition goal, using a visual schedule to shift activities, he improved from 4% to 75%. These gains were maintained, with a mean of 84% during follow‐up sessions.

Troy displayed notable improvements across all domains. His sorting goal (cognitive) improved from 0% at baseline to 86% during intervention. His communication goal, expressing needs using words or visuals, grew from 8% to 92%. His fine motor skills, such as using scissors or manipulating small objects, advanced from 12% to 100%. Finally, his ability to sustain peer play for 3 min rose from 10% at baseline to 85% during intervention. Troy's maintenance scores were the highest of all participants, ranging from 85% to 97% across goals.

3.3. Social Validity

The results from the social validity questionnaires demonstrated a substantial consensus among preschool teachers and student teachers regarding the acceptability and effectiveness of ELOs in improving the engagement behaviours of young children with ASD and ID. Both groups perceived embedded instruction as highly beneficial in the context of inclusive preschool classrooms. The average ratings from the four participating teachers, with a mean of 5.83 (range = 4.80–6), strongly endorsed these interventions, underlining their overall acceptability and effectiveness. Similarly, student teachers, who had the opportunity to observe and interact with the participating children, provided very high ratings for ELOs, with an average score of 5.95 (range = 4.90–6). This consensus suggests that both experienced teachers and those in training regarded ELOs as effective and valuable tools for enhancing the development and engagement of young children with ASD and ID in inclusive preschool settings.

4. Discussion

In this study, the implementation of ELOs during centre‐time/free‐play activities had a substantial positive impact on the engagement and learning of children diagnosed with both ASD and ID in inclusive preschool classrooms. The findings indicated a clear and immediate increase in engagement levels when ELOs were introduced for all four participants, alongside notable progress in children's unprompted correct responses for individualised goals. The data showed that children experienced consistent improvements in engagement as the intervention continued. This is consistent with prior research demonstrating the effectiveness of ELOs (Malmskog and McDonnell 1999; VanDerHeyden et al. 2005) or other systematic approaches (Barton et al. 2018; Bellini et al. 2007; Koegel et al. 2012) in promoting engagement and supporting individualised learning goals. The consistency and generalisation of the results across the four participants suggest that ELOs are a robust approach that can be effectively employed with preschool‐aged children diagnosed with ASD and ID. The intervention not only increased overall engagement but also had a substantial impact on active engagement, an essential aspect of participation and learning. One notable feature of this study is the individualisation of embedded instruction for each child. By aligning the embedded instruction with the specific goals of each child's IEP, the researchers addressed the unique needs and abilities of each participant, resulting in measurable learning progress in addition to enhanced engagement. This research contributes to the broader literature by highlighting the importance of individualisation in intervention strategies, particularly for children with dual diagnoses of ASD and ID.

Generalisation, in the context of skill development, is a fundamental aspect that holds immense significance for the comprehensive development of children (Chandler et al. 1992). It is essentially the ability to apply and transfer acquired skills from one context to another, ensuring that skills learned in a particular setting or situation are not limited to that specific environment (Brown and Odom 1994). Instead, they become part of a child's versatile toolkit, allowing them to participate meaningfully and adapt to various contexts and activities. In this study, the children's capacity to transfer the engagement and learning skills cultivated through ELOs to diverse classroom activities, such as mealtimes and small group activities, is a testament to the efficacy of embedded interventions. Mealtimes and small group activities are distinct from the centre‐time/free‐play activities in their structure, expectations and social dynamics. The fact that children extended their improved engagement and learning behaviours to these different activities suggests that the acquired skills were not tied to one specific context but had a broader relevance.

In addition to the immediate and sustained increases in child engagement observed during the intervention phase, this study provides evidence of the maintenance effect of ELOs. Following the intervention, data collection continued during maintenance sessions conducted at various intervals. The results clearly indicated that the gains made during the intervention were not fleeting. Children who had shown significant improvements in engagement continued to maintain those heightened levels over an extended period, with learning gains also persisting beyond the intervention phase. This maintenance of gains is a crucial finding, demonstrating the durability and lasting impact of embedded interventions. It signifies that the increased engagement levels and learning behaviours achieved through this approach were not merely temporary but had the potential to become enduring improvements, reinforcing the long‐term benefits of embedded instruction in inclusive preschool classrooms. Such maintenance is especially promising for the development of critical social and learning skills in young children with dual diagnoses of ASD and ID and suggests the potential for sustainable progress over time (Gunning et al. 2019).

The high ratings provided by both preschool teachers and student teachers on the social validity questionnaires are indicative of the acceptability and effectiveness of ELOs within inclusive preschool settings. This alignment in perceptions between experienced teachers and those in training suggests that embedded instruction is a feasible and valuable approach that can be adopted by educators in various stages of their careers. These results align with previous research, which also emphasised the practicality and feasibility of embedded instruction in inclusive settings (Malmskog and McDonnell 1999; Rakap and Balikci 2023; VanDerHeyden et al. 2005).

4.1. Limitations and Future Directions for Research

While the study's results are promising, several limitations should be acknowledged. First, this study focused on children with mild to moderate ASD and ID. While this was a valuable starting point, it is essential to recognise that the impact of ELOs may differ for children with more severe ASD and ID. Future research should expand its scope to include children with a broader range of disabilities, ensuring a more comprehensive understanding of how embedded interventions can be tailored to the diverse needs of this population.

Second, while this study took place in inclusive classrooms with a favourable student–teacher ratio of 12–16 children per class with one teacher and one student teacher, this setting may not reflect typical class sizes found in other educational contexts, such as larger mainstream classrooms common in the United Kingdom (Foley 2023). The smaller class sizes in this study likely provided more individualised attention and support during embedded instruction, which may have contributed to the observed engagement and learning outcomes. As such, the findings may not fully generalise to settings with larger student‐to‐teacher ratios, where individualised support may be more challenging to implement. Future research should explore the effectiveness of ELOs in larger classroom environments to examine how class size might impact the feasibility and outcomes of this approach.

Third, this study utilised a non‐concurrent multiple‐baseline design, which, while valuable for examining individualised interventions in complex social settings, may have limitations in terms of experimental control. A concurrent multiple‐baseline design, for example, could enhance control by implementing interventions for multiple participants simultaneously. This design would enable a clearer comparison of intervention effects across participants. However, it is important to acknowledge that the choice of design was likely influenced by practical constraints, given the complex, real‐world educational settings in which the study took place. Future research could consider employing a concurrent multiple‐baseline design where feasible while still recognising the practical challenges of conducting such research in inclusive preschool settings.

Furthermore, this study predominantly relied on quantitative measures to assess engagement and learning. While this approach provides valuable data, incorporating qualitative data would offer a more comprehensive understanding of the effects of ELOs. Qualitative data, such as teacher and caregiver observations and feedback, could provide insights into the nuanced aspects of child engagement and learning, shedding light on how ELOs impact not only the engagement behaviours themselves but also the quality of interactions, the child's emotional responses and other qualitative dimensions.

5. Conclusion

This study contributes to the existing literature on embedded interventions and the development of effective strategies for promoting engagement and learning among children with dual diagnoses of ASD and ID. The positive impact of ELOs on child engagement and learning, the generalisation and maintenance of skills, and the strong social validity reported by teachers and student teachers collectively support the notion that embedded instruction is a valuable tool for enhancing the development and participation of young children with complex needs in inclusive preschool settings. Further research in this area has the potential to refine and expand the use of ELOs as a tailored and individualised approach to improving engagement and learning among children with dual diagnoses.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data available on request from the first author.