Abstract
To date, reports of empirically validated comprehensive intervention programs for children with autism spectrum disorder (ASD) have been limited to preschool-age children. We examined the effects of a model inclusive kindergarten program for children with ASD. Forty-one children received instruction in an inclusive kindergarten program with their peers for 28 hrs a week. A comparison group (n = 21) received an eclectic intervention in public, general education, kindergarten classrooms. Examiners administered standardized tests of cognitive, language, and adaptive behavior skills to children in both groups at the beginning and end of the school year. There were no differences in test scores between the two groups at baseline. Following intervention, the model program group had higher mean standard scores in all skill domains. The differences were statistically significant for all domains except adaptive behavior and spoken language. These findings are consistent with reports of success for inclusive programs for preschool children with ASD.
Introduction
With the incidence of autism spectrum disorder (ASD) increasing to one in 68 U.S. children (Centers for Disease Control and Prevention, 2014), there is mounting urgency for the validation of interventions and treatment programs grounded in empirically based practices. Furthermore, given the impetus for educating students with special needs in the least restrictive environment, the importance of investigating comprehensive intervention inclusion models is vital (Crosland & Dunlap, 2012; Odom, Buysse, & Soukakou, 2011; Strain & Bovey, 2011).
Odom, Boyd, Hall, and Hume (2010) suggest two classifications of intervention, “focused intervention practices” and “comprehensive treatment models.” Focused intervention practices implemented with individual children with ASD emphasize specific targeted skills, such as language (Goldstein, Schneider, & Theiman, 2007; Kaiser & Roberts, 2011) or behavior (Conroy, Asmus, Sellers, & Ladwig, 2005; Dunlap & Fox, 2011), and occur over a limited period of time (National Research Council [NRC], 2001; Odom, Collet-Klingenberg, Rogers, & Hatton, 2010; Reihow & Volkmar, 2010). Comprehensive treatment models are described as a set of practices “designed to achieve broader learning or a developmental impact on the core deficits of ASD” (Odom, Boyd, et al., 2010, p. 426) over an extended period of time (Rogers & Vismara, 2008; Strain, Schwartz, & Barton, 2011).
Researchers suggest that comprehensive treatment programs provide crucial guidelines for best practice for children with ASD (NRC, 2001; Odom et al., 2011; Strain et al., 2011). These programs range from one-to-one, home-based interventions to interventions delivered in typical educational settings such as child care and schools. Along with the disparity with regard to where intervention is delivered are the radical differences in the conceptual approaches to intervention. In developmentally segregated programs that focus on one-to-one adult-directed intervention, such as Early Intensive Behavior Intervention (EIBI) and TEACCH, the impetus is on modifying the environment to accommodate the child (Eldevik, Hastings, Jahr, & Hughes, 2012, Lovaas, 1987; Panerai, Zingale, & Trubia, 2009). Other programs, such as the Learning Experiences and Alternative Program for Preschoolers (LEAP) and Developmentally Appropriate Treatment for Autism (Project DATA), emphasize building the child’s repertoire so he or she can successfully participate in environments alongside typically developing peers (Schwartz, Sandall, McBride, & Boulware, 2004; Strain & Bovey, 2011). The Early Start Denver Model combines both schools of thought and divides the child’s intervention program accordingly (Rogers & Vismara, 2008).
Traditionally, most programs are dedicated to the remediation of the core deficits of ASD to enhance the child’s social and academic repertoire (Odom, Boyd, et al., 2010). In these types of programs, it is only after the child has mastered a predetermined level of skills that he or she is given the opportunity to enter age-typical learning and social settings (Lovaas, 1987; Panerai et al., 2009). Some children never master the skills required to enter age-typical settings, whereas others enter these settings with a paraeducator who compensates for the child’s inadequate, social repertoire (Rispoli, Neely, Lang, & Ganz, 2011; Strain & Bovey, 2011).
Although many studies report the results of interventions targeting the remediation of core deficit skills of preschool children with ASD, to date, no comprehensive treatment models have demonstrated educational outcomes for kindergarten students with ASD placed in fully inclusive classrooms without an individually assigned paraeducator (Odom, Collet-Klingenberg, et al., 2010; Strain et al., 2011). The NRC (2001) recommends systematically implemented and developmentally planful intervention occurring in social settings with children without disabilities as often as possible. These recommendations remain relevant in literature today and are further supported by subsequent research targeting early childhood best practices on a broader scope (Odom, Collet-Klingenberg, et al., 2010). An emphasis on multi-dimensional interventions delivered in inclusive programs forms the empirical platform for children with ASD to develop functionally relevant skills across all domains (Bagnato, McLean, Macy, & Neisworth, 2011; Brown & Conroy, 2011; Stahmer, Akshoomoff, & Cunningham, 2011). More specifically, it is critical to develop an evidence base for children with ASD to access inclusive kindergarten programs bridging their transition from preschool to school-age programs. Inclusive kindergarten programs must allow for social participation, the support of children’s academic growth, and access to the general education curriculum as well as foster successful participation in assessment in preparation for first grade.
Research conducted with children with and without disabilities in the same classroom provides a foundation for understanding the benefits of inclusion for all children (Chandler-Olcott & Kluth, 2009; Grindle et al., 2012; Odom et al., 2011; Rafferty, Piscitelli, & Boettcher, 2003; Stahmer et al., 2011; Strain & Bovey, 2011). Inclusive classrooms provide direct access to general education curricula and promote the development of social competency (Wolery & Hemmeter, 2011). Also documented are the types of environmental arrangements, such as defined learning areas, organized learning materials, predictable routines, and visual prompts and schedules supporting the inclusion of children with disabilities (Brown & Conroy, 2011; Dettmer, Simpson, Myles, & Ganz, 2000; Heflin & Alberto, 2001). Although language is a core deficit of ASD, incidental teaching and the use of verbal behavior strategies have been found to be effective for children with ASD (Kaiser & Roberts, 2011; Yoder & Stone, 2006). Furthermore, research on self-monitoring and positive behavior support has demonstrated wide effectiveness in natural environments for children with ASD who exhibit inappropriate behavior (Crosland & Dunlap, 2012).
Once the environment is addressed and empirically based strategies are identified, the extent of the child’s academic achievement may be evaluated. Many empirically validated classroom strategies strengthened academic achievement for children with ASD. Two such examples are active student responding and explicit and differentiated instruction (Flores & Ganz, 2009; Wolery & Hemmeter, 2011). Other important elements for success in inclusive settings include assessment-driven intervention (Greenwood, Carta, & McConnell, 2010), parent involvement and support (Sainato & Morrison, 2001), and teacher preparedness (Snyder, Hemmeter, & McLaughlin, 2011). The true challenge of designing successful inclusive programs becomes one of accomplishing social immersion while achieving academic progress for all children in the classroom (Crosland & Dunlap, 2012; Odom et al., 2011; Strain & Bovey, 2011).
In this demonstration model, empirically based, complex intervention practices were used within a comprehensive treatment model. The purpose of our study was to compare the outcomes of a model inclusive kindergarten program for children with ASD on their cognition, academic achievement, language, and adaptive behavior with a comparison group of children with ASD attending public kindergarten programs.
Method
Participants
This study was conducted over four school calendar years. Participants were recruited each year and were at least five years old by the beginning of the school year.
Participants in the model and comparison classrooms had an independent diagnosis of ASD or pervasive developmental disorder–not otherwise specified (PDD-NOS). Participants received a diagnosis from a licensed psychologist not affiliated with the study. Participants’ diagnoses of ASD or PDD-NOS were further confirmed by a second licensed school psychologist who served as a member of students’ public school evaluation teams. Children receiving only discrete trial training in either group or one-to-one settings in school were excluded as were children who were repeating their kindergarten year.
Sampling Procedures
Samples of convenience were used in this study to recruit model and comparison participants. Children in the model program were recruited based on their ranking on an open waiting list held by a chartered, nonpublic school emphasizing educational inclusion for children with ASD. Children meeting the criteria for the study were accepted as their name moved to the top of the waiting list and space became available in the model classrooms. Tuition was paid for by the child’s local school district. Other options available to parents in this community were public school classrooms, private programs, and clinical 1:1 services in the area.
Children in the comparison kindergarten classrooms were recruited via the special education coordinators at four public school districts. Local special education coordinators were asked to identify classrooms in their buildings that included kindergarten age children meeting the criteria for the study. Samples of convenience were used due to the fact that children in comparison classrooms were already assigned to a particular location and type of classroom by district personnel prior to entering the study.
Experimental Conditions
The experimental sites for the study were kindergarten classrooms in a Midwest suburban area. The model classrooms consisted of six to seven children with ASD and six to nine typically developing children. In each of the model classrooms, there was an equal or greater number of typically developing children than children with ASD. There was one model classroom in the first year and two model classrooms in subsequent years for a total of seven model classrooms. Staff in the model classrooms consisted of one licensed general education teacher and one assistant with a related bachelor’s degree or an associate degree, and one other trained assistant, except for Year 2 when one of the teachers held a dual degree in general and special education. The classrooms were arranged in clearly defined learning areas such as circle, reading, drawing, dramatic play, blocks and games, science, and computer. The children attended from 9:00 a.m. to 3:00 p.m. on Monday through Thursday and from 9:00 a.m. to 1:00 p.m. on Friday. The daily schedule consisted of arrival, circle, reading and handwriting, lunch and recess, learning centers, recess and snack, and math. In addition, Spanish instruction, art, music, and physical education occurred once or twice a week. Classrooms reflected general education kindergarten rooms in terms of environmental arrangement and materials, curricula, routines, and activities. Children participated in large and small group activities throughout the day.
Model
Teachers and assistants were trained and supported by researchers and mentor teachers before and during the school year. The model classrooms implemented the following intervention elements throughout the daily routine:
Full Day Programming: Children were offered a full 6-hr day of kindergarten, except on Fridays when children attended for 4 hrs. Although the optimal number of hours per week for intervention remains controversial, researchers agree that increased engagement and meaningful opportunities to respond combined with functionally appropriate goals are beneficial for children with ASD (NRC, 2001; Strain & Bovey, 2011).
Inclusion: Children participated in the general education classroom for all activities from the beginning of school until the end of the year. Pull out activities, such as speech or occupational therapy, were held to a minimum to avoid isolating children from their peers. An emphasis was placed on providing therapy services within the classroom during the daily routine when possible. Children with ASD were expected to engage in all classroom activities with typically developing peers in small and large group activities. No assistants were assigned to individual children. Full membership of children with ASD in all aspects of the general education classroom was encouraged (Odom et al., 2011; Strain et al., 2011).
General Education Curriculum: Children were taught using the general education curriculum throughout the school day alongside their typically developing peers. Instruction was aligned with state standards and mapped out from the beginning of the school year to the end. Accommodations and modifications were provided along with differentiated instruction as a matter of routine in the classrooms. The curriculum addressed core deficit areas of children with ASD (i.e., social, communication, and restricted interests, and obsessive behavior) as well as academic skills. The curriculum for core academics included Reading Mastery (Engelmann & Bruner, 2003), MacMillan-McGraw Hill’s Math (2005), Everyday Counts (Gillespie, 2006), and Handwriting Without Tears (Olsen, 1998).
Evidence-Based Strategies: The environment was organized and arranged to support a wide range of diverse learners (Wolery & Hemmeter, 2011). Visually defined learning areas, as well as individual workspaces, supported children with ASD in general education classrooms. Model classrooms were structured to provide predictability and security for all children. To promote children’s learning and independence, visual supports such as schedules, rules, and expectations were provided for the classroom and some individual students. Instructional strategies were used by teachers to promote direct and intensive intervention to meet the needs of children with ASD and included naturalistic instruction, direct instruction, prompting strategies, visual supports, self-monitoring, peer-mediated instruction, reinforcement, and fluency building activities (Flores & Ganz, 2009; NRC, 2001; Odom, Collet-Klingenberg, et al., 2010). See Treatment Fidelity Checklist in the appendix.
Behavior Management: An emphasis was placed on positive behavior support and assisting children in the development of their self-management skills (Dunlap & Fox, 2011). Teachers were trained to use a classwide behavior management system, while providing individual behavior support and strategies for those who needed it.
Teacher Training: Structured, ongoing support and feedback were provided for classroom staff as it pertained to the items on the Treatment Fidelity Checklist. This checklist can be found in the appendix. Before the school year, teachers attended workshops to plan their classrooms. During the school year, teachers had a monthly meeting with research staff to share information, and to discuss concerns. Teachers were asked to assess their delivery of intervention using the Treatment Fidelity Checklist. If needed, research staff observed children to develop an individual behavior plan. In addition, peer coaching was used to enhance and expand teachers’ instructional and behavior management skills (Kohler, Crilley, Shearer, & Good, 1997; Snyder et al., 2011). Teachers modeled specific teaching behaviors for other model classroom teachers, observed each other’s classrooms, and provided feedback.
Comparison Classrooms
There were a total of 17 general education kindergarten comparison classrooms throughout the course of the study. The classrooms were located in four school districts and comprised of one or two children with ASD and 12 to 22 typically developing children. The comparison classrooms were staffed in the following manner: Eight classrooms had one licensed general education teacher and a paraeducator assigned to one or two children with ASD; four classrooms had a licensed general education teacher and an instructional assistant for the entire classroom; three classrooms had a licensed general education teacher, an instructional assistant, and an assigned paraeducator to one child with ASD; and two classrooms were staffed with a single licensed general education teacher. There were two children in the comparison group from self-contained classrooms located in two different districts, but their data were excluded from the analysis as there were no typically developing peers in the classroom.
Education as Usual
The participants in the comparison classrooms received “education as usual” as defined within the context of their public school curriculum (Grindle et al., 2012). All comparison classrooms were organized into learning centers such as reading, math, science, and handwriting. Daily schedules included common activities such as arrival, circle, reading, recess, lunch and/or snack, math, science, learning centers, and weekly occurrences of music, art, and physical education. Comparison classrooms used a variety of eclectic approaches to teach reading and language arts, mathematics, and handwriting. These included a combination of informal and formal general education and remedial published curricula, such as Literacy Collaborative (Marshall & Davis, 1999), THE LETTER PEOPLE® (Lehr, Wertheim, & Johnson, 2002), Touch Math (2002), Everyday Mathematics (University of Chicago School Mathematics Project, 2007). The following strategies and accommodations were noted by the teachers to be embedded in their curricula for children with ASD: visual prompts; picture cues; activity schedules; modeling; direct instruction and guided practice; small group instruction; prompting strategies, including diminishing prompts; and behavioral checklists.
Children in the general education comparison classrooms attended half days 5 days per week or full days 3 days per week. Children with ASD attended these classrooms during the same times and also participated in pull out activities such as speech, occupational therapy, or reading instruction. Participants’ Individual Education Plans were developed and remained the responsibility of the children’s intervention specialist.
Treatment Fidelity
A measure of treatment fidelity was conducted to ensure that the intervention was accurately and consistently implemented in the model classrooms. In addition, the same treatment fidelity measure was used in the comparison sites to assist in the description of the educational experiences of the children with ASD (see the appendix). This procedure follows the recommendations of Gersten, Fuchs, Coyne, Greenwood, and Innocenti (2005) and Wolery (2011), who suggest that treatment integrity be conducted in the comparison sites for descriptive purposes. Prior to the beginning of the study, research staff members reviewed the literature for empirically based intervention practices for children with ASD. The Treatment Fidelity Checklist was developed from this review and used to determine whether these practices were being implemented in the classroom. Observers were trained to use the Treatment Fidelity Checklist by reviewing the component definitions and identifying characteristics of the interventions on-site before using the checklists in the study. Research staff members completed the Treatment Fidelity Checklist in the model and comparison classrooms six times during the school year. The checklist consisted of categories detailing classroom management, general instructional strategies, and strategies for facilitating active engagement. In addition, categories targeting academic readiness, communication and social interaction, independent performance and generalization, and behavior management were included. Intervention was assessed by marking each component as either observed (+), not observed (−), or not applicable (NA). To ensure the believability of the data, interobserver agreement (IOA) for the Treatment Fidelity Checklist was measured for 20% of the sessions. IOA was calculated using the formula (Agreements / Agreements + Disagreements) × 100%. Researchers viewed videotaped sessions during arrival, circle time, and reading to assess IOA of Treatment Fidelity measurements.
Outcome Measures
All participants in the model and comparison classrooms were individually assessed at the beginning and end of the school year using standardized tests. These tests were used to assess child outcomes in cognition, academic achievement, communication, and adaptive behavior.
To assess cognitive ability, a nonverbal intelligence test, the Leiter International Performance Scale–Revised (Leiter-R; Roid, & Miller, 2002) was implemented and a performance score intelligence quotient was obtained. This assessment was chosen as it was devised for a wide variety of user groups and yields a measure of intelligence independent of language ability, a confounding factor for children with ASD. Administration instructions are delivered nonverbally using gestures, and children’s nonverbal intelligence was assessed using pictures, figural illustrations, and coded symbols. Using Fisher’s z-value, correlation metric average reliability coefficients for the Leiter-R ranged from .89 to .91. Test–retest reliability correlations (with interval not reported) for brief and full IQ were .91. The Leiter-R presents consistent evidence of validity from criterion-related studies demonstrating an excellent level of classification accuracy with an overall hit rate of 96% for children with cognitive delays using a cut point of 70 on the full IQ scale (Roid & Miller, 1997).
Measurement of academic achievement was made using the Kaufman Test of Educational Achievement, Second Edition (KTEA-II; Kaufman & Kaufman, 2004). Using the KTEA-II, participants were assessed individually in the areas of reading (letter and word recognition), mathematics (math concepts, applications, and computation), written language (written expression), and oral language (listening comprehension and oral expression). A composite score for academic achievement is derived from this measure as well as a composite score for Oral Language and subtest scores for Oral Expression and Listening Comprehension. This test was chosen due to high internal-consistency reliabilities in age norms (ages 4½ through 25) and grade norms (K-12). Reliability coefficients for kindergarten for subtests ranged from .80 to .97, with a composite of .95 (Kaufman & Kaufman, 2004).
The Test of Language Development (TOLD-P:3; Newcomer & Hammill, 1997) was conducted and a Spoken Language Quotient was obtained. With regard to the reliability of the TOLD-P:3, the test–retest coefficients for the subtests used in this study were greater than .80 and for the composites rounded to or exceeded .90 (Newcomer & Hammill, 1997). The authors suggest that the outcomes of the TOLD-P:3 yielded an acceptable degree of reliability and contain little or no time sampling error.
In addition, the Vineland Adaptive Behavior Scales–Classroom Edition (Sparrow, Balla, & Cicchetti, 1985) was completed by each child’s classroom teacher to measure adaptive behavior. This Classroom Edition provides an Adaptive Behavior Composite score obtained from the domains Communication, Daily Living Skills, Socialization, and Motor Skills. For the Classroom Edition, strong reliability scores are reported ranging from .80 to .95 with the adaptive behavior composite yielding a .98 coefficient alpha means (Harrison, 1985).
All pre- and post-assessments, except for the Vineland Adaptive Behavior Scales, were individually conducted by trained research staff in both the model program and comparison sites. The research staff consisted of two full-time individuals who held PhDs in special education, two doctoral students in special education, and a licensed teacher with a master’s degree in special education. Researchers read and studied testing and assessment manuals and practiced conducting tests and assessments with typically developing children until they were proficient with delivery of testing items. When training to implement tests and assessments, staff observed each other practicing and provided feedback. In addition, 25% of the assessments in model and comparison classrooms were reviewed by another test administrator, and any scoring errors were corrected.
Data Analysis
This study used a quasi-experimental design using a group comparison between kindergarten children attending inclusive model program classrooms and those in public school classrooms. Statistical analyses were performed using SPSS 22.0. Descriptive statistics were used to examine measures of central tendency and dispersion, and nonparametric statistics were deemed appropriate for the majority of variables based on their nonnormal distributions determined from visual inspection and confirmed by the Shapiro-Wilk test of normality. For ease of presentation, variables are expressed as means ± SD (minimum, maximum), and bar graphs are presented for visual comparison. Change scores were calculated by subtracting the pre-intervention from post-intervention score. Between-group (model vs. comparison) differences in change scores were compared using independent t tests for normally distributed variables, and Mann–Whitney U tests for variables with Shapiro-Wilk p values < .05. Effect sizes were calculated for between-group differences in change scores by dividing the mean difference by the pooled SD (for normally distributed variables), or by dividing the Mann–Whitney U statistic z score by the square root of the total N (for nonnormally distributed variables). Chi-square analysis was used to compare proportions between groups for Leiter test scores falling ≥ or < 70, with Fisher’s exact test being used when expected cell frequencies were less than 5. A p value < .05 was considered to be statistically significant.
Results
Participants and School Environment
A total of 62 children with ASD participated in the study. Forty-one children (85% male) participated in the model program, and 19 children (90% male) from public school programs served as participants for comparison. Data of 2 participants in self-contained classrooms were excluded because there were no typically developing peers in the classroom. The race/ethnicity background of the model program group was 93% Caucasian and 7% Other, whereas the comparison group was comprised of 79% Caucasian, 11% Hispanic, 5% Asian, 5% African American, and 5% Other.
The age of the two groups did not differ, with M ± SD values of 75.7 ± 5.6 months and 74.1 ± 4.3 months for the model and comparison groups, respectively. As designed, the children in the model program were exposed to more school hours per week (28 ± 0) than the children in the comparison group (14.3 ± 4.6). The mean number of years of teaching was 4.7 ± 3.1 years for the model classroom teachers and 19.1 ± 9.5 years for the comparison classroom teachers.
Treatment Fidelity
Treatment fidelity was compared between the model and comparison classrooms. The mean percent for “Yes” responses was significantly higher in the model program classrooms (70.6%) than in the comparison classrooms (22.2%). Conversely, the mean percent for “No” responses was significantly higher in the comparison classrooms (65.0%) compared with the model program (18.8%). Specific Treatment Fidelity Checklist results are located in Table 1. The mean IOA was 97% with a range from 94% to 100%.
Summary of Treatment Fidelity Results for Comparison and Model Classrooms.
Child Outcomes
The baseline (pre) and after intervention (post) test scores, as well as effect sizes for between-group comparisons for the pre- to post-intervention change scores, are presented in Table 2. The changes with intervention (post–pre) are presented in Figures 2 and 3.
Pre- and Post-Intervention Test Scores for the Model and Comparison Groups Presented as M ± SD (Minimum, Maximum).
Note. Leiter-R = Leiter International Performance Scale–Revised; KTEA-II = Kaufman Test of Educational Achievement, Second Edition; TOLD-P:3 = Test of Language Development.
Effect size determined from the mean difference in change score (post- minus pre-intervention) divided by the pooled SD.
Effect size determined from group differences in change scores for Mann–Whitney U statistic (Z score divided by the square root of the total n).
CONSORT diagram.
Change scores (post- minus –pre intervention) for Performance Intelligence Quotient, KTEA-II Comprehensive Achievement, and Vineland Adaptive Behavior for the comparison and model program groups.
Change scores (Post – pre-intervention) for spoken language (TOLD P:3) and KTEA-II oral language and expression, and listening comprehension for the Comparison and Model Program Groups. Bars represent means with error bars showing the 95% confidence interval.
Leiter-R
At baseline, the performance IQ did not differ between the model program and comparison groups. However, by the end of intervention, the change in performance IQ was significantly higher in the model program group (13.7 ± 12.6 points) compared with the comparison group (2.5 ± 11.3 points) (Figure 2). The number of children in the model program group with a performance IQ < 70 decreased from 11 to six with intervention compared with the number in the comparison group (from 2 to 1), although the proportions <70 were not significantly different between groups.
KTEA-II comprehensive achievement composite
At baseline, there were no group differences in comprehensive achievement composite scores. Following intervention, the model program group increased by a mean of 4.7 ± 12.2 points. In contrast, the mean score of the comparison group decreased by −2.6 ± 7.5 points, resulting in significant between-group differences for change from pre- to post-intervention, as shown in Figure 2.
KTEA-II oral language composite score
At baseline, the two groups did not differ in oral language composite scores. The change from pre- to post-intervention was significantly different between groups (Figure 3), with the model program group improving by a mean of 5.4 ± 9.3 points, whereas the comparison group score decreased by a mean of −3.6 ± 5.8 points.
KTEA-II oral expression standard score
Oral expression standard scores (Figure 3) did not differ between model program and comparison groups at baseline, but the changes from pre- to post-intervention were significantly different (3.3 ± 9.8 vs. −7.1 ± 8.8 points, respectively).
KTEA-II listening comprehension standard score
At baseline, the listening comprehension scores did not differ between groups (Figure 3). Following intervention, the change in the model program group’s scores were significantly greater (6.0 ± 10.4) compared with the comparison group’s scores (0.2 ± 7.5).
TOLD-P:3 spoken language quotient
There was no difference between groups in TOLD-P:3 spoken language quotient (Figure 3) at the beginning of the school year. Similarly, the change with intervention did not differ between groups (3.3 ± 10.4 vs. −1.5 ± 7.1 for model and comparison groups, respectively).
Vineland composite
As shown in Table 2, the Vineland composite scores (Figure 2) did not differ between groups at baseline. With intervention, both groups improved (5.4 ± 13.0 vs. 2.4 ± 4.3 for the model and comparison groups, respectively), but the changes were not statistically different between groups.
Discussion
The children participating in the model kindergarten classrooms made statistically significant gains in nonverbal intelligence, academic achievement, and language scores over children’s scores in the comparison classrooms. At the onset of the study, the model program and comparison groups’ pre-test scores on standardized assessments were not statistically different. However, post-intervention, the children with ASD in the model program had significantly greater improvement in nonverbal IQ, academic achievement, oral expression, as well as oral and listening comprehension scores than the children with ASD in the comparison groups. Moderate effect sizes were observed for between-group differences in change from pre- to post-intervention for these test scores. In fact, children in the comparison classrooms exhibited either no improvement or even decreases in their standardized test scores. Both model and comparison groups demonstrated similar improvement in pre- and post-test outcomes on the Vineland Adaptive Behavior Scales–Classroom Edition (Sparrow et al., 1985).
The children with ASD in the model classrooms made statistically significant growth in critical areas of development with minimal or no pull out services. Researchers agree that research conducted in natural educational settings is worthwhile, but plagued with challenges (Eldevik et al., 2012; Grindle et al., 2012; Strain & Bovey, 2011). The current study identified the operational variables supporting the academic growth of children with ASD in general education classrooms, including implementation of the model by the teachers. Expectations in the model classrooms focused on equal classroom membership and defined the independent performance required to negotiate a general education classroom. Children with ASD in the model classrooms were accepted without meeting “kindergarten readiness” or a set of predetermined skill levels (Sainato & Morrison, 2001). With the core deficits of ASD being comprised of significant language and social skill deficits, the use of a nonverbal test of intelligence allowed researchers to gather more accurate pre- and post-intervention performance IQ scores regardless of the participant’s ability to speak or write.
Empirically based practices embedded systematically within the structure of the model classrooms allowed access to the general education curriculum throughout the school day. Direct and explicit instruction, differentiated instruction, individualized behavior management systems, fluency building strategies, and an empirically based reading curriculum were vital to the progress of the children in the model classrooms. Exposure to the general education curriculum promoted improved cognitive, academic, and language skills. Intervention focusing on the development of self-management skills, following directions, and attending to the task further encouraged model participants’ ability to take the post-tests.
There was a significant difference in treatment fidelity between practices observed in the model and comparison classrooms. Treatment fidelity data for the model classrooms demonstrated significantly higher implementation of the intervention components, suggesting the model classrooms were impacted by the high quality and consistent delivery of identified intervention strategies. Data for the comparison classrooms showed that although “recommended practices” strategies were identified as being used by the teachers, there was a lack of implementation in the classroom. Interestingly, this is a similar finding to that of Boyd et al. (2013) in their comparative study of LEAP, TEACCH, and Non-Model-Specific comprehensive treatment preschool special education programs.
Teachers and classroom assistants in the model classrooms were provided systematic training that focused on informing, modeling, coaching, and feedback of teacher behaviors. A critical component of the model program was staff training as it prompted continuity across the model classrooms regardless of the individual characteristics of the children with ASD and the number of years of experience of the staff. With a minimum of six children with ASD in each model classroom, high levels of instructional control were in evidence during large group instruction. Quite surprising was the lack of improvement between pre- and post-intervention for comparison classroom participants. Inherent in the model program is the fact that all children, including those with ASD, are the primary responsibility of the teacher. When children are placed in general education classrooms with an individually assigned paraeducator, as was the case in some comparison classrooms, it often fell on the paraeducator to provide individualized assistance and instruction, effectively creating a “classroom of one.” Participants in the model classrooms were provided ample opportunities to follow teacher directions, receive teacher feedback, and participate equally as members of small and large group instruction.
Additional differences between the model and comparison classrooms were the number of hours participants were in their classrooms and the ratios between children with ASD and their typically developing peers. The children in the model classrooms attended school more hours per week than those in the comparison classrooms. Studies suggest the number of hours of intervention is critical (Howard, Stanislaw, Green, Sparkman, & Cohen, 2014). Strain et al. (2011), however, propose that the number of hours is not as important as the quality of intervention. Variables such as skilled staff, access to peers, specialized learning opportunities, comprehensive curricula, addressing behavior problems, and family involvement are essential to improving child outcomes (Eldevik et al., 2012; Strain & Bovey, 2011; Strain et al., 2011).
The smaller class sizes and staff and children ratios of the model classrooms may have allowed for unique cohorts to develop. The low ratio of teacher-to-children along with the high ratio of children with ASD to the typically developing children in the model constructed classroom dynamics that appeared to support both groups of learners. Model classroom staff were prepared to differentiate instruction and engage children across a broad continuum of learning levels. Low teacher-to-student ratios appeared to contribute to additional practice of academic tasks and individual feedback. Model classroom staff may have adjusted their prompting as the children with ASD developed more complex and independent responding. Typically developing peers may have indirectly provided additional instruction and appropriate responding to classroom expectations. Future studies should investigate the typically developing children and the impact on their academic achievement and development of prosocial behaviors.
The researchers in our study acknowledge a number of limitations. First, it was not possible to randomly assign children in this study due to the necessity to respond to the reality of the public school district’s responsibilities (Ong-Dean, Hofstetter, & Strick, 2011). However, given the preliminary outcomes, we believe it provides a strong foundation for future randomized studies. This study also used samples of convenience. A common criticism of convenience sampling is sampling bias. This refers to the limitation in generalization and inferences about the entire population, resulting in low external validity (Ong-Dean et al., 2011). A second limitation was the lack of a blind study. Researchers were aware of whether the children were assigned to the model or comparison classrooms; however, researchers did conduct testing and observations with the same children throughout the study (Gersten et al., 2005). The absence of random selection and assignment make it difficult to rule out confounding variables and suggest the results of this study should be viewed cautiously. Also concerning was the fact that there were fewer comparison participants than the number of model participants. Future research should attempt to explore and replicate inclusive comprehensive treatment models that include higher number of participants and randomized experimental designs where possible (Ong-Dean et al., 2011). Effect size estimates may be beneficial in determining the sample size needed for adequate power in a randomized controlled design.
In our study, the concept of inclusion was embraced with an emphasis of designing a comprehensive treatment model supporting academic growth of children with ASD in general education kindergarten classrooms. This study indicates that the educational needs of children with ASD may be met in a kindergarten setting with one’s peers. Future studies should evaluate the social growth of children with ASD in general education settings. To our knowledge, this is the first empirical report of an inclusive program for kindergarten age students with ASD. Currently, “education as usual” for children with ASD offers fewer inclusive educational options and more often participation in self-contained classrooms, 1:1 clinical treatment approaches, or general education classrooms with 1:1 adult assistance (Rispoli et al., 2011; Strain et al., 2011). Regardless of the expertise of the professionals implementing these options, the absence of a “critical mass” of same-age peers removes a vital variable required to build sustainable social skills leading to the development of one’s social competency and future academic progress (Harper, Symon, & Frea, 2008; Strain & Bovey, 2011). The statistically significant increases in model participants’ performance IQ, academic achievement, and language provide evidence that children with ASD should have options. These options should include same-age typical educational environments that provide systematic, empirically based intervention to promote growth for all members.
Footnotes
Appendix
Authors’ Note
Sunhwa Jung is now at Kongju National University, Chungcheongnam-Do, Republic of Korea. Judah Axe is now at the Simmons College, Boston, Massachusetts.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a Model Demonstration Grant H324MO030199 from the U.S. Department of Education.