Stakeholder groups included autistic adults, caregivers of autistic persons, psychologists, speech-language therapists (SLTs) and occupational therapists (OTs). Caregivers of autistic persons who had completed at least Grade 4 of a mainstream curriculum qualified for inclusion, as did autistic adults who had completed at least Grade 4 in South Africa. The criteria for selection to professional groups (educators, psychologists, SLTs and OTs) included experience with autistic learners (Grades 4–12) or experience in adjudication of assessment accommodations at district or provincial education offices. All respondents needed to be over 18 years of age and be sufficiently proficient in the English language to complete the questionnaire.

One hundred and eighteen individuals consented to participate. However, 16 questionnaires were largely incomplete and were discarded, leaving 102 respondents and demonstrating an 86.4% completion rate. Respondents had to self-identify their stakeholder groups. A difficult decision was made to exclude 10 individuals identified as belonging to two stakeholder groups. Excluding these 10 participants was considered necessary as one aim of the study was to compare different stakeholder groups, using the two-proportion z-test for independent groups. Including these respondents in both of the stakeholder groups they identified would result in a small percentage of the sample being dependent observations, invalidating the independent groups. Additionally, such a move would over-represent the voice of these 10 participants as their responses would be included in multiple groups. The authors acknowledge that these hybrid groups can offer valuable insights that bridge personal and professional experiences, highlighting the complexities of educational support. The final sample size was therefore 92 respondents.

Using G*Power software (Faul et al. 2007), we calculated the minimum sample size required for a two-proportion z-test at a significance level of 0.05, with a statistical power of at least 0.80 and a medium effect size. The analysis indicated that a minimum of 64 responses was necessary. These parameters were chosen because detecting medium effects with adequate power is generally considered sufficient for meaningful interpretation, whereas minimal effects may reach statistical significance (e.g. p < 0.05), although often lacking practical relevance (Peeters 2016). The achieved sample size of 92 participants exceeded the recommended minimum, ensuring that the statistical tests had the required power.

When respondents selected ‘other’ when qualifying their involvement in autism, they were included in the stakeholder group closest to their stated involvement: ‘Child psychiatrist’ and ‘social worker’ were grouped with psychologists and ‘sibling with autism’ was grouped with caregivers. The final sample had four autistic adults (n = 4/92; 4.3%), 14 caregivers (n = 14/92; 15.2%), 15 educators (n = 15/92; 16.3%), 27 OTs (n = 27/92; 29.3%), 10 psychologists (n = 10/92; 10.9%), 18 SLTs (n = 18/92; 19.6%), and three ‘other’ (n = 3/92; 3.3%) responses, in which the involvement was not qualified, and one respondent did not select a group (n = 1/92; 1.1%).

The majority of respondents were English-speaking (n = 59/92; 64.1%), from Gauteng (n = 34/92; 37.0%) and the Western Cape provinces (n = 24/92; 28.3%), the two most densely populated provinces within South Africa. Many respondents were noted to work in more than one setting, but predominantly in private practice (25/70; 35.7%), private centre or school (23/70; 32.9%) and state special school (15/70; 21.4%). Almost a third (19/70; 27.1%) of the respondents in the professional groups (educators, psychologists, SLTs and OTs) had more than 10 years’ experience with autistic learners; 22.9% (16/70) had 6–9 years’ experience; 22.9% (16/70) had 3–5 years’ experience; 21.4% (15/70) had less than 3 years’ experience and 5.7% (4/70) did not indicate their experience level.

Questions in the questionnaire were compiled by reviewing policy documents related to assessment accommodations within South Africa, including Annexure C1 of the National Policy Pertaining to the Conduct, Administration and Management of the National Senior Certificate Examination (DBE 2014b; DoE 2001). The table summarising accommodations from this document was included in the questionnaire for participant perusal. Additionally, the literature referencing accommodations for various barriers to assessment was consulted (Elliot, Kratochwill & Schulte 1999; Larson et al. 2020). Autism-associated characteristics that could impact assessment performance at the school level were considered (Ashburner et al. 2010; Tamm et al. 2020).

The questionnaire, designed for completion within 15–20 min, consisted of dichotomous, nominal multiple-choice and Likert-type questions that were closed-ended, as well as open-ended questions to obtain rich in-depth information on the elicited perspectives.

A mixed-method survey design was used. Stakeholders’ perspectives were collected through a researcher-designed, self-administered, web-based questionnaire distributed by Qualtrics software (Version: June 2021). Links to the web-based questionnaire were posted on the Facebook™ pages of South African professional organisations and South African therapist groups, teacher groups and autism groups. All had granted permission to have the link shared on their pages. When the link was clicked, the informed consent letter was displayed, requiring consent before proceeding to the questionnaire. Organisations (South African Speech-Language-Hearing Association, Occupational Therapy Association of South Africa, South African Association for Child and Adolescent Psychiatry and Allied Professions, and Autism South Africa) were also approached to distribute the link to their member databases.

All data were analysed by using SPSS Statistics software, version 27. The two-proportion z-test was used to make pair-wise comparisons among stakeholder groups. A p-value less than 0.05 indicated that the responses differ significantly. The 5-point Likert scales were collapsed into three categories: disagreement, neutrality and agreement.

Qualitative information gathered from open-ended questions was initially coded by the first author, and a consensus of themes was then achieved among the authors. In the qualitative component, thematic saturation was approached as no substantially new themes emerged after coding the final set of responses. However, we acknowledge that the limited representation of educators and autistic adults may have constrained the diversity of perspectives.

Ethical clearance to conduct this study was obtained from the Research Ethics Committee of the Faculty of Humanities, University of Pretoria (No. HUM022/0720).

Findings from this study concur with previously documented inconsistencies in educator knowledge regarding assessment accommodations (Hodgson, Lazarus & Thurlow 2011). This finding was reiterated by one of the respondents’ comments:

Autistic learners are therefore at risk, as educators and other support staff may not be aware of the support that can be offered, as documented in policy. Similarly, a lack of knowledge of education policies hinders lobbyists for change and the translation of policy to practice. Consequently, training and collaboration are needed regarding assessment accommodations for autistic learners across inclusive and special schools (Larson et al. 2020; Wilkinson & Twist 2010). This policy is vitally important as the incidence of autism and the enrolment of autistic learners are rising, as shown by studies in the Western Cape province of South Africa and elsewhere in the world (Lindsay et al. 2014; Pillay et al. 2021, 2022).

Almost half of the respondents felt that the accommodations summarised in Annexure C1 (DBE 2014b) accommodated the needs of autistic learners during assessment, and yet the majority indicated that additional accommodations would be valuable. This view may be related to the fact that respondents felt that existing accommodations have a clear academic focus but do not support the barriers affecting participation in assessments, such as anxiety and sensory regulation needs.

This South African study has demonstrated the need for collaboration and training on assessment accommodations across the education system and for relevant stakeholders to adequately support autistic learners. Policy implementation should evolve, and research findings, such as this study, should inform policy and practice (Lazarus et al. 2009).

Previous research has indicated a need to recognise the ‘more individualised, unique and autism-specific needs’ of autistic learners (Saggers et al. 2019), rather than grouping autism with other conditions. Respondents emphasised that autistic learners should be considered individuals with different needs and that assessment accommodations must reflect this view:

Respondents noted that current accommodations do not sufficiently support barriers related to anxiety and sensory regulation. In the United States, for example, over 90% of postsecondary institutions used academically focused accommodations, but only a third provided sensory accommodations (Brown & Coomes 2016). Global research has indicated the need to address sensory and other nonacademic support for autistic learners in both classroom and assessment settings (Ashburner et al. 2010; Brown & Coomes 2016).

The most-selected additional assessment accommodation in this study was ‘specialised settings: low arousal; special lighting and/or acoustics’, despite respondents simultaneously indicating uncertainty about the effectiveness of environmental changes for reducing anxiety. Sensory-friendly environments as assessment accommodations are not widely studies in the literature (Lai & Berkley 2012; Leifler et al. 2021). However, at least one study demonstrated that low-distration environments can positively affect concentration for students with sensory defensiveness (Lewis & Nolan 2013), and the autism community continues to advocate for sensory-related accommodations (Sarrett 2018). Respondent comments further highlighted the need for sensory-friendly environments and personalised regulatory supports:

Provision of additional time has been recommended as reasonable accommodation for exam stress and anxiety (Jansen et al. 2017). Just over half of the respondents agreed that additional time would alleviate anxiety for autistic learners.

The pre-assessment training suggestions by respondents are similar to previous (Simpson, Griswold & Smith Myles 1999; Songlee et al. 2008; Tamm et al. 2020; Tyrrell & Woods 2018), showing that preparation strategies reduce anxiety and support executive functioning. One autistic respondent highlighted the need for materials that are not based solely on nonautistic assumptions, emphasising risks such as miscommunication or gaslighting:

The variability in the options selected for assessment accommodation (866 total responses) was large. These differences may stem from the unique and varied needs of autistic learners as well as the diverse stakeholders with whom they interact (Keen et al. 2016; Witmer & Ferreri 2014). Accommodations should therefore be selected in a bespoke fashion to accommodate individual needs.

A small number of statistically significant differences emerged across stakeholder groups. Although limited in number, these differences are meaningful. Psychologists and caregivers differed significantly in their views on the value of specialised settings. This divergence likely reflects the distinct experiential and professional vantage points from which these groups understand learning environments. Caregivers routinely adapt home contexts to support daily functioning, often using intuitive strategies to minimise sensory and environmental barriers (Schiavone et al. 2018). Hence, they may be less exposed to the kinds of environmental constraints arising within classroom settings, in which sensory, social and organisational demands are less routinely modified. Psychologists, by contrast, are trained to identify how environmental stressors exacerbate emotional, behavioural and self-regulatory challenges, which may heighten their recognition of the potential pitfalls with stressors related to assessment. These divergent insights illustrate how varying degrees of proximity to educational environments and differences in professional frameworks shape stakeholders’ perspectives of what supports are most needed.

Autistic adults and caregivers selected the use of an encourager or motivator significantly more often than OTs did. Although specific literature supporting the use of motivators is limited, it is plausible that lived experience offers unique insight into the importance of emotional co-regulation and motivational support during demanding scholastic tasks. The comment from an autistic respondent in this study underscores the heterogeneity of autistic profiles and highlights the role that a regulation partner or motivational presence may play in reducing stress, enhancing task engagement and supporting performance. Research on autistic self-advocacy similarly notes that autistic individuals often emphasise relational and regulatory supports that may be under-recognised in clinical or educational frameworks (Milton 2019; Pellicano et al. 2014). Thus, different endorsement rates for encouragers and/or motivators may reflect the value of integrating lived expertise, which can illuminate needs that professionals may overlook.

Psychologists also selected ‘no two exams on one day’ more frequently than OTs did. This choice may relate to psychologists’ training in emotional and behavioural well-being, including their attention to factors such as cognitive load, anxiety, fatigue and stress dysregulation (Kellems et al. 2016). The assessment spacing aligns with principles of reducing cumulative stressors, especially for learners with heightened vulnerabilities in emotional or executive functioning. OTs, while knowledgeable about environmental and sensory barriers, may place comparatively less emphasis on emotional load across multiple assessments.

Taken together, these group differences illustrate the importance of incorporating diverse stakeholder perspectives into research on assessment accommodations. Each group brings distinct forms of knowledge (professional expertise, contextual familiarity or lived experience) that shape their perceptions of what supports are necessary. The differences observed in this study highlight the risk of relying too heavily on any single stakeholder group, as doing so may inadvertently privilege certain types of expertise over others. Multistakeholder engagement, therefore, remains essential for developing assessment practices that are equitable, ecologically valid and responsive to the wide variability within the autistic population.

Education White Paper 6 notes that the curriculum must be flexible enough to accommodate different learning needs and styles (DoE 2001). The study findings align with this, emphasising the need to expand the available assessment accommodations and ensure flexibility across the education system. Such flexibility supports learners with executive function, communication difficulties and sensory needs, consistent with UDL.

The frameworks for inclusive education around the world have called for UDL and assessment (Capp 2020; Dolan & Hall 2001), but as yet, the change in assessment practices has been limited (O’Neill & Padden 2022). Universal design elements prioritised by respondents in this study included: ‘all directions and prompts in simplified language’, ‘redundancy (questions presented auditorily and in written form)’ and ‘all questions starting on a new page’. These accommodations are not currently available within South African policy (DBE 2014b), except redundancy, which is offered to learners with specific learning disabilities related to reading and writing. Respondents also called for pre-assessment accommodations as a specific training approved by provincial or district education offices, to be provided as a matter of course. Including this option as an element of universal design may benefit all learners and offer flexibility from the planning stage, and not a later adaptation. The need for strengthening the education system to support autistic learners is also obvious (Pillay et al. 2022), as materials should be available for teachers to use in support of learners with barriers to learning, including autistic ones. The pre-assessment training suggestions by the respondents in this study correlate with those proposed in earlier research (Simpson et al. 1999) to support autistic learners in preparation for assessment; by reducing anxiety and supporting executive functioning, as one respondent mentioned “as in the principal of social stories” (Respondent 90, Speech-language therapist, Western Cape). These suggestions support learners in ways similar to social stories, bu reducing anxiety, increasing familiarity and supporting executive functioning (Songlee et al. 2008; Tamm et al. 2020; Tyrrell & Woods 2018). Respondents prioritised universal design elements such as simplified language in directions, redundancy (written and auditory formats), and the starting of questions on new pages. These are not yet included in South African policy (DBE 2014b), except redundancy, for specific learning disabilities. Respondents also recommended universal pre-assessment training, which could benefit all learners and embed flexibility from the planning stage.

Although some limitations in this study have been highlighted, it is important to consider how this study, with future studies, could create concrete opportunities for policy development and expansion. Further training for educators on autism is needed as a result of the increasing numbers and lack of knowledge about the specific barriers facing autistic individuals. This approach will, in turn, develop classroom-based accommodations that all educators can employ. Policy-makers should also consider what universal design aspects can be ‘built-in’ to assessment accommodations processes, such as providing redundancy should a learner want this, or presentation of exam papers with keywords explained and questions presented on separate pages. The ‘allowable’ assessment accommodations should also be broadened, allowing educators to select individual accommodations to suit every learning barrier, including those perceived as ‘nonacademic barriers’ that clearly impact academic performance. The importance of policy-makers consulting with varied stakeholders in developing these accommodations is important, and likewise, the importance of transverse teams at the district and provincial level is emphasised.

The strength of this study was its incorporation of perspectives from a range of stakeholders. This type of participatory research is valued as it can lead to contextually appropriate suggestions for the easy translation of research into policy and practice, thereby maximising success (Pellicano 2020; Pillay et al. 2022; Saggers et al. 2019). The study findings can be used to support the assessment of policy evolution in line with Education White Paper 6 (DoE 2001) and the Screening, Identification, Assessment and Support (SIAS) policy (DBE 2014a) in support of autistic learners in assessment.

As a result of the recruitment strategy, which relied on distributing survey links through Facebook™ groups, a sampling bias is likely. Respondents were already engaged in autism-related online communities, which may have skewed the sample towards individuals with heightened interest, awareness or access to autism-related resources. Consequently, teachers from rural or under-resourced districts were under-represented. This factor aligns with the findings of the study, which show that most respondents were situated in urban areas and were predominantly English or Afrikaans first-language speakers. These limitations have implications for both the generalisability of the results and the equity of the study, as the perspectives of educators in linguistically diverse, rural or resource-constrained contexts may not be adequately reflected. Perspectives of stakeholders from varied geographic settings, languages, cultures and socio-economic settings are needed to gather a comprehensive view.

Similarly, autistic adults (4.3% of respondents), caregivers (15.2%), educators (16.3%) and psychologists (10.9%) were notably under-represented, while OTs formed the largest stakeholder group (29.3%). This imbalance has important implications, as it may skew the findings towards a predominantly clinical or therapeutic perspective, thereby limiting the extent to which the lived experiences of autistic adults and the day-to-day insights of caregivers and educators are reflected in the data. Consequently, the perspectives most relevant to learner- and family-centred approaches may not be fully captured.

Future research should broaden stakeholder representation to ensure more balanced and inclusive multistakeholder perspectives, particularly by more deeply engaging autistic adults and caregivers whose lived experiences are central to learner- and family-centred practice. Studies should also examine assessment accommodations across both mainstream and special school settings, as well as among professionals with varying levels of experience, to understand how contextual and expertise-related factors shape accommodation practices.

Further work is needed to evaluate the effectiveness and differential benefits of specific accommodations for autistic learners, thereby contributing to evidence-based practice. Additionally, research should investigate the support needs of educators in implementing accommodations within everyday teaching and learning activities, as well as formal assessments.

Finally, the opportunity for developing and empirically testing universal design strategies in curriculum and assessment to more proactively meet the needs of autistic learners is immense.