Toward the Improvement of Inclusive Education for Students With Disabilities via Online Intervention: Feasibility Randomized Controlled Trial in Australia

ABSTRACT

Training key stakeholders in the design of reasonable adjustments for students with disabilities could promote collaboration and inclusion. We tested the feasibility of a randomized controlled trial to evaluate the effectiveness of online education (intervention) in designing reasonable adjustments by Victorian (Australian) parents and professionals (n = 31). Participants suggested reasonable adjustments for contrived students with varied needs to support their participation in described classroom activities. Data collected pre- and post-intervention, and, at follow-up, were rated across five dimensions: Authenticity, Agency, Real Learning, Strengths Based, and Inclusion. Analyses from 15 intervention and 14 control participants (10 parents, 10 teaching staff, and 9 allied health professionals) showed higher scores for Inclusion post-intervention for the intervention group. This difference was maintained at follow-up, generalized to a new contrived student scenario, and mostly attributable to parents. Online education may improve parents’ contribution in the design of reasonable adjustments that promote inclusion. A larger study is warranted.

KEYWORDS:

• Disability
• inclusion
• inclusive education
• primary school
• reasonable adjustments

A global increase in students with disabilities attending mainstream schools suggests a growing acceptance of their right to an inclusive education (Buchner et al., 2020). Inclusive ideology is reflected in policies around the world, including Australia and other signatories to the UN Convention on the Rights of Persons with Disability (CRPD) (United Nations, 2006). Australia ratified this convention in 2008, including Article 24, which specified the right to education for all children (Iacono et al., 2019). Unfortunately, such policies may not translate to practice (Iacono et al., 2019; Mavropoulou et al., 2021). There is international evidence, including in Australia, of continued segregation of students who may be considered unable to benefit from mainstream teaching (Agran et al., 2020; Mavropoulou et al., 2021). Such challenges in policy-practice translation arguably reflect differences in understanding of inclusion (Wiesel, 2020). Such differences may be the product of limited understanding and skill in knowing how to accommodate the learning needs of students with diverse abilities (Iacono, McKinstry et al., 2020; Organisation for Economic Co-operation Development, 2021). Increasing the ability of teachers and other stakeholders to address the needs of heterogeneous groups of learners may hold the key to authentic inclusion – that is, the education of students with disabilities within mainstream classrooms where they are provided the supports needed to engage in the same curriculum as and alongside their peers without disability within general education classrooms (Amor et al., 2019).

Inclusion in all aspects of society is grounded in a social model of disability, whereby barriers are addressed through reasonable adjustments to systems and processes (Shakespeare, 2018). The requirement for reasonable adjustments has been enshrined in Australian legislation through the Disability Standards Education 2005 (DSE, 2005), with reference to the Disability Discrimination Act 1990 (DDA) (Dickson, 2019). It has been argued, however, that reasonable adjustments lack operational definition, creating difficulties for teachers who must work out how to accommodate the needs of individual learners (Mavropoulou et al., 2021; Poed, 2016). A recent international survey of primary school teachers showed that over a third experienced stress arising from difficulties in modifying lessons for students with diverse needs (Organisation for Economic Co-operation Development, 2021). This stress may be compounded by teachers feeling they have sole responsibility for designing and implementing reasonable adjustments, which, in turn, may underlie practices that serve to exclude students with disabilities within their own classrooms (Poed et al., 2020). In an Australian survey, for example, Poed et al. (2020) found that teachers often did not make reasonable adjustments for these students, and there has been a widespread practice, both in Australia (Poed et al., 2020) and elsewhere (Punch, 2015), of delegating responsibility for designing strategies and implementing them with students with disabilities to teacher assistants, who are not equipped to provide specialist supports.

Models of inclusion that incorporate collaborative approaches have been characterized by the sharing of expertise across stakeholders in the education of students with disabilities. An example is co-teaching, whereby teachers and, often, special educators, work together in designing and delivering instruction within inclusive classrooms (Iacono et al., 2021). In this way, responsibility for designing and implementing lessons and making other adjustments to meet varied learner needs is shared. Unfortunately, the evidence for co-teaching, which occurs in various forms, lacks a strong research base (Iacono et al., 2021), but nonetheless points to potential benefits accrued from combining expertise in a collaborative process (Hyett et al., 2020). This notion was explored by Hyett et al. (2020), who found that teachers, allied health professionals, and parents could develop a team dynamic that engendered shared problem solving around a task of developing supports for mainstream school students with disability. These stakeholders were selected as participants in the co-design process because they are often nominated as members of student support groups in Australian schools (Howes, 2017). Parents were included in alignment with collaborative principles (Friend & Cook, 2017) of contributing expertise about their own child to the professional expertise of teachers and allied health professionals (Hyett et al., 2020). Unfortunately, in practice, parents are often placed in adversarial roles, which Cuskelly (2020) argues is a consequence of privileging the views and input of teachers over parents in relation to the schooling of their children with disabilities.

Although collaboration among teaching and other professionals, and with parents, has been promoted for school education of students with diverse learning needs (Friend & Cook, 2017), differences in approaches to supporting these students may impede attempts to jointly develop supports in the form of reasonable adjustments designed for classroom contexts (de Oliveira Borba et al., 2020). Pfeiffer et al. (2019) noted that in the United States, a collaborative approach across teachers and specialists who provide services in schools is mandated by the Individuals with Disabilities Education Act (IDEA). Similarly, in Australia, student support groups are formed to collaboratively develop individualized education plans (IEP) for students with disabilities, whether in inclusive or segregated settings (Howes, 2017). Such collaboration and sharing of expertise are thwarted, however, by pull-out models of intervention that rely on individualized therapy without integrating skills into classroom activities or sharing expertise with teachers, teacher assistants, or parents. Further, individual intervention is aligned with a medical model of intervention, which can dominate allied health training, with its focus on redressing disability-specific skill deficits (Keller, 2018).

On the other hand, proponents of educational inclusion, in recognition of potential student-specific learning needs, have suggested a tiered model of supports. Known as multi-tiered systems of support (MTSS), Tier 1 of this model has a focus on universal design to cater to the needs of most students, including those with diverse abilities, but keeps them involved in learning and social activities with their peers (Sailor, 2008). More focused supports are indicated through monitoring of students’ response to intervention, which may indicate the need for more intensive supports, such as those provided in small groups (Tier 2) or individually (Tier 3), but always with the aim of enabling the student to perform at grade level in the mainstream class with Tier 1 supports (Sailor, 2008). The MTSS can be argued to shift supports for students with disabilities to a social model, which relies on reasonable adjustments, the language used in the DSE 2005 and DDA in Australia. By focusing on a Tier 1 approach, reasonable adjustments can be designed to cater to students regardless of their needs or whether they are in receipt of funding for specialist services or classroom support by teacher assistants (Iacono et al., 2019), while maintaining the inclusion of students, no matter the extent of their learning challenges. A recent initiative in one Australian school jurisdiction, the Victorian Department of Education, where the current study was situated, has focused funding on a similar tiered model of support for students with disabilities as a way of improving their inclusion in mainstream schools (Victorian State Government, 2022).

Barriers to the development of inclusive reasonable adjustments for students with disabilities include a failure to operationalize them within relevant legislation (Poed, 2016), and a lack of shared understanding of inclusion or approaches to student support across educators and allied health professionals, which Hyett et al. (2020) and Garcia-Melgar et al. (2022) found extended to parents. Professionals, including teachers and allied health professionals, can access formal means of learning, often through varied forms of professional development, which often are unavailable to teacher assistants (Punch, 2015) or parents. Parents who may not know how their child may best be included in school can feel disempowered or that they are viewed negatively by teachers or other school personnel (Cuskelly, 2020). Delivering the same education in models of inclusion and development of reasonable adjustments to various stakeholders offers the potential to support collaborative processes that can serve to further student inclusion in mainstream classes. The focus of this study was to provide three key stakeholder groups with access to the same educational package about reasonable adjustments in terms of what they were and their role in inclusive education. If effective, such education could serve to ensure that stakeholders with different backgrounds and expertise could engage in the task of developing supports from a base of shared understanding of reasonable adjustments.

The question of effectiveness is best addressed by experimental studies. According to Boylan and Demack (2018), in the United Kingdom there has been recent growth in the use of experimental research designs, including randomized controlled trials (RCTs), in an effort to build the evidence for different forms of professional development. In this research, and in broader educational research addressing questions of efficacy, concerns have been raised about costs associated with ensuring rigor in conducting RCTs (Boylan & Demack, 2018; Dawson et al., 2018; Odom, 2021). In particular, Odom (2021) noted challenges faced when trying to maintain experimental rigor in the complex organizational environment of a classroom or school.

Purpose

This study was part of a larger project that built on previous work focused on demonstrating collaborative processes across educators, allied health professionals, and parents (Hyett et al., 2020; Iacono, McKinstry et al., 2020). The research questions tested through the RCT were whether an online education package about inclusive education and reasonable adjustments that embeds the social model of disability (a) would improve the quality of reasonable adjustments generated by parents, teaching staff, and allied health professionals in the education of students with disabilities; and (b) could be sustained and generalized across student scenarios. It was predicted that participants who received the online education package would develop reasonable adjustments that were more aligned with the social model of disability in comparison to those developed by participants who did not receive the online education package. In testing these research questions, we addressed an aim of evaluating the feasibility of conducting a RCT with a relatively small number of participants, with a focus on maintaining random allocation to groups, blinding participants to the hypothesis, and blinding researchers to group allocation.

Methods

Ethical considerations

La Trobe University Human Research Ethics Committee approval was obtained to conduct this study (HREC, Approval number HEC18445) followed by permissions from the state Department of Education and Training and regional Catholic Education Office. All participants provided written consent to participate in this study.

Design

This was a feasibility parallel randomized controlled trial (RCT) with two groups (intervention/control). Participants in the intervention group were provided with an online education package (intervention) and engaged in their regular practice. Participants in the control group engaged in their regular practice only.

There were three data collection points: time 1, conducted pre-intervention to collect baseline and participant demographic data; time 2, conducted post- a four-week intervention exposure period to explore intervention effects; and time 3, follow-up, conducted eight weeks post-intervention to examine sustainability and generalizability of the learning.

Participants

Victorian parents, teaching staff, and allied health professionals with current or recent experience in supporting a child with disability in mainstream or special primary school, sufficient English skills, and who were older than 18 years were eligible for recruitment. Students with disabilities were not included because of ethical and feasibility challenges but were represented through scenarios (see below). Thirteen of 37 primary schools located in regional Victoria agreed to assist with recruitment by distributing participant information and consent forms within their school communities. Notices were distributed via social media, with snowball sampling used once recruitment began. An expression of interest to participate was made by contacting a nominated researcher, who provided additional information. Participants confirmed their agreement by submitting a signed consent form. The target was to obtain consent from at least eight participants from each stakeholder group within the recruitment period of February to July 2019.

Of 31 participants who provided consent, two discontinued (Figure 1) prior to intervention data collection (time 2); data were analyzed for the remaining 29 participants who comprised 10 parents, 10 teaching staff, and 9 allied health professionals (speech pathologists and occupational therapists). Participants were reimbursed financially for their time once completed.

Figure 1. Participant flow through study.

Characteristics for each participant stakeholder group are presented in Table 1. Most participants resided regionally and held a university degree. The distribution of age range across the teaching staff was almost equivalent, with parents mostly older (> = 36 yrs) and allied health professionals mostly younger (< = 35 yrs).

Table 1. Characteristics of each participant stakeholder group.

	Parents (n = 10)	Teaching staff (n = 10)	Allied health (n = 9)
Location
Metro	2 (20%)	1 (10%)	2 (22.22%)
Regional/Rural	8 (80%)	9 (90%)	7 (77.77%)
Age group
Up to 25 yrs	-	1 (10%)	1 (11.11%)
26-35 yrs	2 (20%)	3 (30%)	6 (66.66%)
36-45 yrs	6 (60%)	3 (30%)	1 (11.11%)
45+yrs	2 (20%)	3 (30%)	1 (11.11%)
Highest qualification
Vocational Training*	3 (30%)	2 (20%)	1 (11.11%)
University**	7 (70%)	8 (80%)	8 (88.89%)

N = 29. Counts (frequencies). *Includes pre-university vocational courses, such as Certificate III and Certificate IV in Australia **Includes university degrees (Diploma, Bachelor, Honors) and postgraduate degrees.

All parents had a child with disability, with most having one attending primary school at the time of their participation (six parents with one attending child, three parents with two, and one parent with three attending). The median age of the parents’ oldest child was 9.5 yrs (range: 7-21 yrs). Half of the children were attending government schools and half were attending Catholic schools, with all being taught in mainstream classrooms.

The median time in their current roles for teaching staff was 5 yrs (range: 6mths-18 yrs) and allied health professionals 1.5 yrs (range: 2wks-6 yrs). All had experience with students with disabilities in their professional roles, with three teaching staff and three allied health professionals also having experience in a nonprofessional role.

Randomization and allocation

Assignment of participants to either the intervention or control group was determined by using a computer-generated blocked (block size = 4, balanced for intervention/control group) randomization sequence (Dallal, 2020) prepared by non-blinded researchers not involved in recruitment or data collection. Each participant stakeholder group had a randomization sequence comprising three blocks. Based on the CONSORT participant flow diagram (Moher et al., 2010; Moher et al., 2001; Schulz et al., 2010), Figure 1 presents the participant flow through the study, including allocation of each participant stakeholder group to the intervention and control groups.

Upon receiving participant consent, a researcher involved in recruitment provided the participant’s study number and stakeholder group to the non-blinded researchers. A non-blinded researcher assigned the participant sequentially to the randomized sequence provided for that participant’s stakeholder group. Participants were blind to their group allocation until post-baseline (time 1) data collection.

Allocation concealment was maintained by password protecting the randomization sequence and making it available only to the two non-blinded researchers throughout the duration of the data collection period. Researchers involved in ongoing recruitment and data collection were blind to participant group allocation (intervention or control) to protect from selection and unconscious bias. On completion of data collection, a research team member was provided with access to the randomization sequence to prepare for data analysis, which included coding the participant’s group (intervention or control). Only this research team member was privy to the group allocation code used during data analyses. Code break of the group assignment occurred once data analysis decisions were finalized as determined by the senior researcher who was fully blinded throughout the trial.

Intervention

The intervention was an education package developed for the study and accessed from a website, thereby ensuring it was accessible to all participants. This mode of delivery also allowed for flexibility, in that participants could progress through it at their own pace and at time(s) convenient to them within the four-week intervention exposure period that was nominated by the researchers according to participants’ entry into the study and completion of the time 1 data collection. The package, titled Designing reasonable adjustments to support school inclusion of children with a disability addressed two intended learning outcomes (ILO) in relation to primary school education. These ILOs were to (a) understand the origins of reasonable adjustments in educational and disability policy, the social model of disability, and inclusive education; and (b) apply the social model of disability to the design of reasonable adjustments for students with varied educational needs. The ILOs were provided in the package, with four learning modules. The first module contained key concepts of inclusive education (UNICEF, 2012), the social model of disability contrasted to the medical model of disability, reasonable adjustments explained in the context of MTSS (Sailor, 2008), and a team approach. In this first module theoretical, philosophical and practice underpinnings of the online education were presented (first ILO). Reasonable adjustments were described in terms of five dimensions that had emerged from workshop discussions of a panel of four members, who represented stakeholders experienced in the education of students with disabilities (Iacono, Spong et al., 2020). These dimensions of Real Learning, Authenticity, Agency, Strengths Based, and Inclusion provided the basis for applying these concepts (second ILO) to two students with varied education needs, who were the focus of the next two modules: Sally and Jamal.

Within their respective modules, Sally and Jamal were described in terms of their personalities, family situation, and learning strengths and challenges. Also presented was information about their student support teams (e.g., parent, teacher, teacher assistant, allied health professional), IEP (Arthur-Kelly & Nielands, 2017), and an activity that addressed an Australian curriculum learning area and was suitable for each student’s grade level (Australian Curriculum Assessment and Reporting Authority, n.d.): Grade 1 for Sally (7 years old) and Grade 6 for Jamal (12 years old). Within each student module, six reasonable adjustments were provided. For Sally, each reasonable adjustment was followed by an evaluation according to the five dimensions. Definitions of these dimensions were provided within the module and were also presented in the first module of key concepts. Similar content was provided for Jamal, but participants were asked to choose the best reasonable adjustment from three options for each of six reasonable adjustments, with evaluative feedback provided for their response. Participant responses were not recorded but served as a means for self-evaluation of learning.

The final module contained evidence reviews of inclusive education practices prepared by Iacono, McKinstry, et al. (2020). Topics were collaborative consultation, co-teaching, education support staff (i.e., teaching assistants) in the classroom, peer supports, social skills groups, and other enhancers.

Various presentation formats were included in the online education package, including written text, graphics, short video animations, and information sheets that could be downloaded. During the study, the online education package required a password to access, but once data collection and analysis were complete, was made open access – see https://www.reasonableadjustmentsinschools.org.au.

The La Trobe Reasonable Adjustments for Inclusive Education (RAIE)

The RAIE was developed for the purpose of this study and provided the outcome measure. It comprised three student scenarios with similar details as those included for student scenarios in the education package and a rating scale. Scenario students were identified as Sienna, Parri, and Elang, who had varied characteristics and learning needs. The task was to respond to a scenario by providing three reasonable adjustments that would support the student’s participation in a described classroom activity relating to an Australian curriculum learning area and was appropriate to the target student’s grade level. These reasonable adjustments were then rated by a researcher (not involved in the randomization and allocation process or data collection) according to a 5-point scale, with each dimension and point on the scale operationally defined. Scores indicated alignment to the social model of disability based on the extent to which each of the five dimensions were addressed (1 = not at all to 5 = fully addressed). Iacono, Spong et al. (2022) provides information about the development, trialing, and reliability of the RAIE. The RAIE was shown to have overall good point-to-point reliability with agreement between the researcher and a professional experienced in inclusive education, but independent to all aspects of the study, scoring 75.3% with an Interclass Correlation (ICC) of .778 (95% CI .761–.793). Reliability was moderate ICC <.70 for the dimensions of Agency and Real Learning. Overall reliability for each of the scenarios of Parri and Elang was good, ICC being .82 (95% CI .79–.84) and .77 (95% CI .72–.81), respectively, and moderate for Sienna, ICC being .74 (95% CI .71–.77) (Iacono, Spong et al., 2022).

Data collection

One week prior to each data collection point, participants were emailed the RAIE. Data collection occurred via interview, during which a researcher assigned with the task of data collection (a) read aloud a scenario, the student’s IEP goals, and the class lesson; (b) asked the participant to suggest their reasonable adjustments that would enable the student to participate in the class lesson; (c) wrote down the participant’s response, and prompted, if necessary, to capture details of the activity, where it would occur, and who would be involved. For example, a prompt from the researcher to the participant may have been, “Can you explain how the activity would look like in more detail?,” and (d) read back what they had written to ensure that the participant’s intended reasonable adjustments had been accurately recorded. Digital and written recordings of the participants’ reasonable adjustments were collected. The researchers followed a script to maintain consistency in language. Each interview duration was approximately 45 minutes and there was no time limit for participants to respond.

Each reasonable adjustment was coded such that each code enabled identification of the participant code, data collection time point, student scenario, and number of reasonable adjustment (i.e., to indicate all three were collected). Codes blinded the researcher involved in rating reasonable adjustments to information regarding time point of data collection (Iacono, Spong et al., 2022).

Researchers involved in data collection contacted the non-blinded researchers when a participant completed time 1 data collection (baseline). Using a scripted e-mail, a non-blinded researcher informed the participant of their group allocation (assigned according to the randomization sequence), which for the intervention group, included the online link and password to access the online education package. Participants in the intervention group were informed that completion of the online education would take approximately 2–3 hours and was required before time 2 data collection. All participants were asked to avoid sharing details and to reply to the e-mail sent from the non-blinded researcher. The reply provided a measure of compliance to the four-week intervention exposure period for those allocated to the intervention group. Fourteen participants replied the same day or day after, and two participants (both control group) more than a week later (median reply time = 3 days, range within 1–22 days).

Participants were asked to suggest reasonable adjustments for the Sienna and Parri scenarios at times 1 and 2 to provide data for baseline and intervention effect measures, and also at time 3 to measure sustainability of learning from the online education package. Participants were also asked to provide reasonable adjustments for the Elang scenario, but at time 3 only, to provide data to measure generalizability of learning – that is, whether learning from the intervention could be extended to apply to a different student scenario. At the commencement of times 2 and 3 data collection interviews, participants were reminded to avoid revealing their group allocation to the researchers interviewing them. At these time points, they were asked whether they had engaged in additional education or training about inclusive education in order to record potential confounders. One participant (intervention) revealed their group at time 2. Two participants (one control, one intervention; time 2 for both and time 3 for one) engaged in activities comprising content of direct relevance to the intervention. Three participants (all intervention group; time 2 for all and time 3 for one) engaged in professional development related to disability.

Data analysis

For each participant, six reasonable adjustments were generated at data collection time 1 and time 2, and nine generated at time 3 (i.e., three reasonable adjustments generated for each student scenario presented at the given time point). For each participant, a dimension score and an overall score was calculated for each scenario at each time point. The dimension score was calculated by tallying the rating provided for the given dimension for each of the three reasonable adjustments (score range: 3–15). The overall score was calculated by tallying the dimension scores.

Two overall scores were calculated: the tally of the five dimensions (Overall 5D, score range: 15–75) and the tally of the three dimensions (Overall 3D, score range: 9–45) for which good reliability had been demonstrated: Authenticity, Strengths Based, and Inclusion (Iacono, Spong et al., 2022). Analysis of Overall 3D was conducted to explore whether dimension reliability impacted the sensitivity of detecting a significant intervention effect.

Given this was a feasibility study with limited statistical power associated with the small sample size, separate analyses were conducted for each scenario with a reduction in the number of variables included. To examine whether engagement with the online education package (intervention) improved the quality of reasonable adjustments generated as indicated by the RAIE ratings of the intervention group, a 2 (time: 1 and 2) × 2 (group: control, intervention) mixed design ANOVA was conducted on dimension and overall (3D and 5D) scores. One-way ANOVA was used to explore simple main effects for significant interactions. Familywise error was not accounted for through Bonferroni adjustment of alpha, thereby risking a Type II error to ensure potential effects were detected and could inform a larger study. Sustainability of learning was examined through visual inspection of the mean dimension and overall scores obtained for the Sienna and Parri scenarios at time 3 (follow-up). Generalization of learning to a new student scenario was examined through visual inspection of mean scores obtained for the Elang scenario. Median scores across time and group (control, intervention) for each participant stakeholder group were visually examined through graphical representations where a significant group difference was observed post-intervention. This visual inspection allowed examination of possible differences in scores and thus generation of reasonable adjustments by stakeholder group.

Three missing values occurred where a third reasonable adjustment was not provided for a scenario at a data collection time point. These were imputed with the average of the two ratings provided for the respective dimension for the given scenario at that time point. Missing values also occurred where a participant did not provide any reasonable adjustments for a scenario and these were not replaced: one participant at time 3 for the scenario of Parri, and another participant at time 3 due to discontinuing participation post-time 2 (Figure 1). Data were inspected for extreme outliers (±3 standard deviations from the mean), statistical assumptions were checked, and p-values <0.05 were considered statistically significant. Non-parametric statistics (not shown) confirmed parametric statistical findings where violations of assumptions were suggested (see Appendix for reference to median dimension and overall scores across scenarios).

The measure of central tendency reported throughout was determined by appropriateness as per the statistics applied, sample size, and normality. Statistics were performed using IBM SPSS v26.

Results

Mean (standard deviation) dimension and overall scores for the intervention and control groups at each time point for each scenario are presented in Table 2, along with the 2 × 2 ANOVA interaction findings.

Table 2. Mean (standard deviation) dimension and overall scores for the intervention and control groups at times 1, 2, and 3 for each student scenario, along with 2 (time: 1, 2) × 2 (group: intervention, control) ANOVA interaction findings.

	Sienna scenario			Parri scenario			Elang Scenario
	Control	Intervention	Interaction	Control	Intervention	Interaction	Control	Intervention
Authenticity
Time 1	10.36 (2.17)	9.87 (2.26)	p = .051, ηp^2 = 0.13	10.25 (2.97)	10.87 (2.00)	p = .44
Time 2	10.07 (3.00)	11.40 (1.64)		9.50 (2.71)	10.90 (2.42)
Time 3	9.77 (3.30)	10.40 (2.41)		10.62 (2.60)	11.14 (2.11)		9.92 (1.85)	9.40 (3.04)
Agency
Time 1	4.57 (1.40)	5.00 (1.77)	p = .32	4.39 (1.33)	5.20 (1.90)	p = .89
Time 2	4.36 (1.34)	5.33 (1.99)		4.64 (0.93)	5.57 (2.11)
Time 3	4.77 (1.64)	6.13 (2.59)		4.77 (1.30)	5.64 (1.55)		4.77 (1.48)	5.07 (2.15)
Real Learning
Time 1	8.43 (2.21)	7.60 (2.10)	p = .97	7.39 (2.22)	8.40 (1.76)	p = .056
Time 2	8.86 (2.80)	8.07 (2.15)		7.86 (2.38)	7.33 (1.72)
Time 3	8.00 (2.68)	8.10 (2.25)		7.08 (1.71)	7.64 (2.41)		8.00 (1.92)	7.33 (2.02)
Strengths Based
Time 1	9.79 (2.08)	8.93 (2.02)	p = .46	8.57 (2.44)	8.87 (2.92)	p = .14
Time 2	10.14 (2.74)	8.53 (2.17)		8.57 (1.91)	7.30 (2.23)
Time 3	8.77 (2.74)	9.50 (1.78)		8.31 (2.18)	8.29 (2.70)		8.62 (1.98)	8.40 (1.88)
Inclusion
Time 1	7.43 (2.56)	5.80 (1.94)	p < .001, ηp^2 = 0.44	7.11 (2.75)	7.20 (2.80)	p = .018, ηp^2 = 0.19
Time 2	6.86 (2.21)	8.67 (2.38)		6.64 (2.98)	9.03 (2.98)
Time 3	7.00 (2.61)	8.63 (3.39)		6.54 (3.18)	9.07 (2.95)		5.54 (1.33)	7.20 (3.17)
Overall 3D
Time 1	27.57 (5.88)	24.60 (4.64)	p = .04, ηp^2 = 0.14	25.93 (7.03)	26.93 (5.89)	p = .49
Time 2	27.07 (6.88)	28.60 (5.17)		24.71 (6.09)	27.30 (6.23)
Time 3	25.54 (7.72)	28.53 (5.96)		25.46 (5.94)	28.50 (6.06)		24.08 (4.07)	25.00 (6.59)
Overall 5D
Time 1	40.57 (8.52)	37.20 (6.97)	p = .12	37.71 (9.83)	40.53 (8.24)	p = .96
Time 2	40.29 (10.59)	42.00 (7.23)		37.21 (8.69)	40.20 (8.69)
Time 3	38.31 (11.52)	42.77 (9.25)		37.31 (7.55)	41.79 (9.17)		36.85 (6.12)	37.40 (9.97)

Time 1 = Pre-intervention; Time 2 = Post-intervention period (~4wks); Time 3 = follow-up (~8wks post intervention). Overall 3D = tally of dimension scores for Authenticity, Strengths Based, and Inclusion; Overall 5D = tally of all five dimension scores. Control n = 14 (time 1 and 2), n = 13 (time 3). Intervention n = 15 all times except n = 14 for Parri time 3.

Sienna scenario

As shown in Table 2, for Sienna, a significant time by group interaction effect was observed for the Inclusion dimension (only) and for Overall 3D.

Inclusion dimension

Between the intervention and control groups, scores for Inclusion did not differ significantly at time 1, baseline (p = .06). A significant group difference (p = .04, η_p² = 0.14) was observed at time 2, post-intervention, with the intervention group scoring higher than the control group. Across times 1 and 2, scores for Inclusion did not differ significantly for the control group (p = .12); for the intervention group, they were significantly higher at time 2 than time 1 (p = .001, η_p² = 0.58).

Trends at time 3, follow-up, were similar to results at time 2, with similar mean scores for Inclusion; the intervention group scoring higher than the control group.

Overall 3D

No significant group difference was observed in Overall 3D scores obtained at time 1 (p = .14) or at time 2 (p = .50). Across times 1 and 2, Overall 3D scores did not differ significantly for the control group (p = .70), but were significantly higher at time 2, post- intervention, than time 1, baseline, for the intervention group (p = .03, η_p² = 0.29).

The mean Overall 3D score obtained by the intervention group at time 3, follow-up, showed little change from time 2. There was a trend suggestive of a group difference at time 3, with scores for the intervention group being higher than the control group.

Parri scenario

For Parri, a significant time by group interaction effect was observed for the Inclusion dimension only (Table 2).

Inclusion dimension

Scores for Inclusion did not differ significantly between groups at time 1 (p = .93), but the intervention group scored significantly higher (p = .04, η_p² = 0.15) than the control group at time 2, post-intervention. Scores for Inclusion did not differ significantly across times 1 and 2 for the control group (p = .39), but were significantly higher at time 2 than time 1 (p = .03, η_p² = 0.31) for the intervention group.

The findings at time 2 were reflected in the trends at time 3, follow up, with minimal change in mean scores and maintenance of the intervention group scoring higher than the control group.

Participant stakeholder groups

Median Inclusion scores obtained for each time point for the participant stakeholder groups (parents, teaching staff, allied health professionals) are shown in Figure 2a for Sienna and Figure 2b for Parri. Both figures show that higher scores for the intervention group than the control group at time 2, post-intervention, were largely accounted for by parents. In contrast, there was no consistency and no clear evidence of group differences at time 2 across scenarios for the teaching and allied health professionals respectively. Higher scores obtained at time 2 compared to time 1 for the intervention group were accounted for by the parents and allied health professionals but were inconsistent across scenarios for teachers. Compared to teaching staff, changes in scores for the parents and allied health professionals of the intervention group were less marked from time 2 to time 3.

Figure 2. (a) Median inclusion dimension scores obtained by each participant stakeholder group across each data collection time point for the Sienna scenario. Parent intervention n = 6 and control n = 4 all times. Teaching staff intervention n = 5 and control n = 5 all times. Allied health intervention n = 4 all times and control n = 5(time 1 and 2), n = 4 (time 3). (b) Median inclusion dimension scores obtained by each participant stakeholder group across each data collection time point for the Parri scenario. Parent intervention n = 6 (time 1 and 2), n = 5 (time 3) and control n = 4 all times. Teaching staff intervention n = 5 and control n = 5 all times. Allied health intervention n = 4 all times and control n = 5(time 1 and 2), n = 4 (time 3).

Elang scenario – generalization

Mean dimension and overall scores for Elang (Table 2) trended toward a group difference for the Inclusion dimension only, where a higher mean score was observed for the intervention than the control group.

Summary of between-group intervention effect findings

The intervention group obtained significantly higher scores for the Inclusion dimension than the control group at time 2, post-intervention, for both the Sienna and Parri scenarios, which appeared to be accounted for by parents. Trends at time 3, follow-up, for the Sienna and Parri scenarios suggested maintenance of scores and group difference from time 2. The group difference was also evident for the scenario of Elang.

Discussion

In this study, by implementing an RCT, we addressed the research question of whether an online education package would result in improvements in the reasonable adjustments for primary school-age students with disabilities designed by teaching staff, allied health professionals, and parents. We further evaluated whether any changes seen would be sustained over time and generalized using a descriptive approach. The number of participants was kept smaller than conventionally required for an RCT to test the feasibility of maintaining other key features of the design.

The finding regarding the efficacy of the online education package (RCT) was of a significant intervention/control group difference post-intervention at time 2 for both Sienna and Parri scenarios for only one of five dimensions on the RAIE: Inclusion. According to visual inspection of group medians post-intervention, this difference was accounted for by one group of stakeholders only: parents. Parents, in particular, would have differed from teachers and allied health professionals in their understanding of inclusive education, reasonable adjustments, and how to support academic learning and social participation of students with varied needs by virtue of their different professional training and experience, and/or lived experiences of a child with disability. We found that parents were particularly likely to benefit from the online education package, which, perhaps, could serve to ensure the shared understanding they need to fully participate as equal partners (Cuskelly, 2020) in collaborative processes to design educational supports for their own children (Hyett et al., 2020). Changes seen for the allied health professionals might be accounted for by increasing their ability to consider adjustments suitable for a classroom learning task rather than withdrawing the child, particularly through the online education, but possibly, at least to some extent, also through repeating the task. This explanation may account for improvements seen for the intervention group (Keller, 2018). That is, the online education package may have served to enhance common understanding and skill across the participant groups as a starting point to collaborative processes.

The finding that the dimension of Inclusion was most amenable to change for the Intervention group could be attributed to a bias in the online education package (i.e., an emphasis on this over other dimensions). Arguments against this explanation include that the online modules addressed each of the five dimensions through explanation and example, as well as in self-assessment quizzes (participant responses were not captured or sent to the researchers). Still, Iacono, Spong et al. (2022) found that of the five dimensions, Inclusion, along with Authenticity and Strengths Based, resulted in good agreement between independent coders of reasonable adjustments generated across all participants (i.e., education staff, allied health professionals, and parents) and from all data collection time points. These dimensions may have been more easily grasped by participants post-intervention, than were the dimensions of Agency and Real Learning, which Iacono, Spong et al. (2022) argued may require further contextual support than could be conveyed in the brief scenarios and curriculum objective descriptions.

The improved ability in designing reasonable adjustments across student scenarios was encouraging. The scenarios of students used for the outcome measure, the RAIE, differed in terms of their characteristics and learning challenges to students who were included in the online education package. Characteristics of students depicted for the RAIE may have evoked specific disability types, in particular intellectual disability and complex communication needs for Sienna, and communication and social-emotional difficulties suggestive of autism for Parri, and a mixture of these characteristics for Elang. Although there were some similarities with students depicted in the online education intervention in terms of learning challenges, the profile of one student, Sally, was suggestive of fetal alcohol spectrum disorder (FASD) and that of Jamal suggested cerebral palsy and complex communication needs. These disabilities were not named in any of the RAIE or online education package scenarios, but differences in the characteristics depicted, as well as the curriculum area that provided the contexts for the reasonable adjustments, may have contributed to differences in the reasonable adjustment scores seen across participant groups. As an example, previous researchers have noted that children with more severe intellectual disabilities are less likely to attend mainstream schools or classes than those with mild intellectual disabilities or on the autism spectrum (Agran et al., 2020; Mavropoulou et al., 2021). The potential for these varied profiles to influence the success of the online education package (with Sally and Jamal as scenarios on which to base the package) or the ease with which participants could generate quality reasonable adjustments (for Sienna, Parri, or Elang) is difficult to discern from the results. The changes seen could reflect the potential for the education to have greater benefits when considering students who are less likely to be seen in mainstream classrooms, as compared to those for whom inclusion is more often seen. Further, the difference between intervention and control groups was sustained at follow-up, and for a third scenario – Elang, who could also be described as having a severe disability.

The rigor of experimental studies is premised on controlling potential confounds that can otherwise reduce confidence in attributing any effects seen between groups to the intervention (Odom, 2021), which for this study was delivered online. Using this medium and restricting access of the intervention website to the intervention group ensured that what was provided to participants of this group was consistent. What was not controlled, however, was how closely individual participants worked through all elements, including the time they spent completing each and when they completed the education package in relation to the second data collection interview date. The occurrences of possible differences across individuals within groups are well catered for, however, in large sample RCTs, which reduce their impact by distributing chance error, and thereby, increase the power of statistical tests to detect any existing intervention effects (Dawson et al., 2018). In conducting a small sample feasibility study, we demonstrated the feasibility and worth of conducting an RCT in an area in which there is concern about the costs involved and whether they can yield useful information (Boylan & Demack, 2018; Dawson et al., 2018), but not the impact of differences within or across groups caused by factors other than the intervention.

To counter the lack of power from a small sample, a relatively liberal p-value was used in that we did not adjust for familywise error, thereby increasing the chance of failing to accept a true null hypothesis: that is, finding an effect that may not in fact be real or be only a likely effect (Dawson et al., 2018). The consequence has been that the study was open to potential confounds, while, arguably, demonstrating the value in further research into this or similar education to improve the quality of reasonable adjustments.

Limitations and future research directions

The development of an educational package to teach about the design of reasonable adjustments that embraces the social model of disability has been an innovative approach to addressing the professional development gap in this area, which we have extended to target parents, rather than professionals alone. Further innovation has been through the continued, free online availability of the educational package past the duration of this project, to increase understanding of inclusion and promote collaboration between key stakeholders. There are limitations to this study, however, that require noting for the benefit of future research.

Although we obtained small effects that reached significance, they do require interpretation given our adoption of a liberal p-value (i.e., by not adjusting for multiple comparisons), a strategy used to avoid rejecting an intervention whose promise might become more apparent with a larger sample size. In doing so, the need for rigor took precedence over power as a key feasibility consideration. According to Odom (2021), small effect sizes in educational research are common, as is the rejection of interventions, which may in fact prove effective in better powered studies or those in which accommodations are made for the complex contexts of schools and classrooms. The modest findings from this study, therefore, provide rationale for further research to evaluate the same online education to determine whether changes could also be effected in teachers and allied health professionals across all scenarios through a larger, well-powered study. A further argument for continuing this research is that those changes that were detected were sustained at follow-up and the learning could be generalized to an additional, different scenario (Elang).

Of concern in conducting a larger RCT is the difficulty of maintaining rigor through the data collection period (Odom, 2021). Maintaining the online and self-paced features of the intervention is recommended to avoid the vagaries caused by organizational complexities, including scheduling, multiple activities, and competing demands on participant time (Odom, 2021). It also allows recruitment to extend beyond single classrooms and schools, thereby enabling sample sizes that are often difficult to achieve in education studies. In future research, collection of qualitative data from participants regarding their engagement with the online education package is encouraged to inform the useability of the online education package.

Conclusions

This feasibility study provides evidence that engagement in online education regarding the social model of disability, reasonable adjustments, and inclusive education can improve the quality of reasonable adjustments developed by stakeholders in the primary school education of a child with disability. The improvement in quality was observed for the dimension of Inclusion only. This evidence suggests that further research examining the impact of this online education package is warranted, and, further, that this study design could be used to inform a larger scaled RCT that will have sufficient statistical power to detect other possible effects of the intervention.

Acknowledgments

We would like to acknowledge Ms Emily Greaves for her assistance with rating the reasonable adjustments and Dr. Oriane Landry for her contribution in the earlier phases of our research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This project was supported by La Trobe University Building Healthy Communities Research Focus Area funding.

Appendix

Median (Range) dimension and overall scores for the intervention and control groups at times 1, 2 and 3 for each student scenario.

Table

	Sienna scenario		Parri scenario		Elang scenario
	Control	Intervention	Control	Intervention	Control	Intervention
Authenticity
Time 1	11 (4–12)	10 (6–14)	11 (4.5–14)	11 (7–13)
Time 2	10 (4–15)	12 (8–14)	10 (4–13)	11 (5–14)
Time 3	11 (3–15)	11 (6–14)	11 (6–14)	11.5 (8–15)	10 (7–13)	10 (3–13)
Agency
Time 1	4 (3–7)	5 (3–9)	4 (3–8)	5 (3–10)
Time 2	4 (3–7)	4 (3–9)	4.5 (3–6)	5 (3–9)
Time 3	4 (3–7)	5 (3–11)	4 (3–7)	5 (3–8)	5 (3–9)	4 (3–9)
Real Learning
Time 1	9 (4–11)	8 (3–12)	7.5 (3–11)	9 (5–11)
Time 2	9.5 (4–13)	8 (4–11)	8.5 (4–11)	7 (5–11)
Time 3	9 (4–12)	7 (5–11)	7 (4–10)	7.5 (4–12)	8 (6–12)	8 (3–10)
Strengths Based
Time 1	10.5 (5–12)	9 (6–12)	8.5 (3–12)	9 (3–12)
Time 2	11 (4–14)	8 (5–12)	8.5 (5–12)	7 (3–12)
Time 3	9 (5–13)	10 (6–12)	8 (5–12)	9 (3–12)	8 (6–12)	9 (4–11)
Inclusion
Time 1	7.5 (3–11)	5 (3–11)	6.5 (4–11)	7 (3–14)
Time 2	7 (3–10)	9 (4–13)	6.5 (3–12)	10 (4–13)
Time 3	7 (3–11)	8 (5–13)	5 (3–13)	9.5 (5–13)	5 (4–8)	7 (3–13)
Overall 3D
Time 1	29 (14–34)	23 (19–37)	26.5 (12–35)	26 (14–38)
Time 2	28.5 (12–38)	28 (18–38)	27 (12–33)	28.5 (16–37)
Time 3	24 (11–37)	29 (18–38)	25 (16–35)	28.5 (19–39)	24 (18–31)	26 (12–36)
Overall 5D
Time 1	43.5 (21–50)	35 (30–55)	38 (21–54)	42 (22–55)
Time 2	43 (19–58)	42 (28–54)	40 (19–49)	42 (25–54)
Time 3	37 (18–54)	46 (27–57)	37 (24–51)	42.5 (29–58)	35 (27–48)	38 (18–55)

Time 1 = Pre-intervention, Time 2 = Post-intervention period (~4wks), Time 3 = follow-up (~8wks post intervention). Overall 3D = tally of dimension scores for Authenticity, Strengths Based, and Inclusion, Overall 5D = tally of all five dimension scores. Control n = 14 (time 1 and 2), n = 13 (time 3). Intervention n = 15 all times except n = 14 for Parri time 3.