Neurodivergence as a predictor of school exclusion: findings from a large school-based survey

Original Research Article Neurodiversity Volume 4: 1–16 © The Author(s) 2026 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/27546330261505142

Neurodivergence as a Predictor of School Exclusion: Findings From a Large School-Based Survey

journals.sagepub.com/home/ndy

Lili Bokor[footnote 1] , Alice M.G. Quinton2,3, Simona Skripkauskaite2,4, OxWell Study Team2,5, Mina Fazel[footnote 5] and Francesca Happé[footnote 3]

Abstract Background: School exclusion is linked to poor educational, social, and health outcomes. This study explored self- reported school exclusion among self-identified neurodivergent (e.g. autistic and attention deficit hyperactivity disorder [ADHD]) versus neurotypical peers. Methods: Cross-sectional data came from the 2023 OxWell Student Survey of 15,012 secondary school pupils (aged 11–19 years) in England. Participants self-identified as neurodivergent (n = 3,081) or neurotypical (n = 11,931) and reported an exclusion history and special educational needs (SEN) support. Chi-square tests assessed group differences, and logistic regressions tested whether neurodivergence predicted exclusion beyond other risk factors, with an explora- tory analysis of permanent exclusion. Results: Neurodivergent students were nearly four times more likely to report exclusion (14.6% vs. 4.1%), and more likely toreport permanent exclusionspecifically(2.3%vs.0.4%)than neurotypical peers. Elevated riskpersisted across genders and school years. SEN support was not associated with reduced exclusion among neurodivergent students. Neurodivergence remained an independent predictor of exclusion after accounting for gender, ethnicity, age, and socio-economic deprivation; this estimated effect was even higher for permanent exclusion specifically (odds ratio = 4.75). Conclusions: Neurodivergent young people experience disproportionately high exclusion rates, representing a particu- larly vulnerable group. Findings highlight the need to identify modifiable factors that could inform policy and practice to reduce exclusion.

Lay Abstract Previous research has shown that autistic students and students with ADHD are more likely to experience difficulties at school, including a higher risk of school exclusion. However, this work has largely focused on specific diagnoses, meaning that many young people who think and learn differently, particularly those without a formal diagnosis or who do not use diagnostic labels, are often overlooked. This study examines school exclusion in relation to self-identified neurodiver- gence using a large national survey of secondary school students in England. Instead of asking about specific conditions, students were asked whether they considered themselves neurodivergent, reflecting current thinking in the neurodiver- sity field and allowing a broader range of experiences to be included. The study explored both factors linked to a higher risk of exclusion and factors that might help protect against it, while also considering differences in students’ backgrounds,

Corresponding Author: Lili Bokor, Department of Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, 16 De Crespigny Park, London, SE5 8AF, UK. Email: [email removed]

Creative Commons Cc By: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https:// creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage).

2 Neurodiversity 4(0)

such as gender, ethnicity, and experiences of special educational needs (SEN) support at school. The findings show that students who identify as neurodivergent are more likely to be excluded from school, highlighting the need for change across education systems. For schools, this emphasises the importance of recognising and responding to students’ needs early, treating disruptive behaviour as a sign that a student needs more support rather than grounds for punishment, and ensuring that SEN support is timely, appropriate, and tailored to individual needs – rather than relying solely on formal diagnoses to determine who receives that support. For researchers, the results point to looking beyond specific diagno- ses to better understand the wider range of neurodivergent experiences and the factors that increase or reduce the risk of exclusion. For policymakers, the findings support the development of policies that prioritise early support and effective school-based practices to reduce preventable school exclusion and improve outcomes for neurodivergent young people.

Keywords Neurodivergence, autism, ADHD, school exclusion, adolescents, special educational needs

Received: March 18, 2026; accepted: October 22, 2026 policy. The Department for Education guidelines outline legal frameworks for exclusion, with specific protections for students with special educational needs (SEN) and dis- abilities (Department for Education, 2024) and other pro- tected characteristics – such as race and sex – identified by The Equality Act (2010) (Government Equalities Office & Equality and Human Rights Commission, 2013). Yet, there remains a persistent gap between policy inten- tions and actual school practices (Palikara et al., 2018). Despite progressive policies aiming for inclusive education, certain student groups, including boys, students from ethnic minority backgrounds, and those from socio-economically disadvantaged families, are disproportionately excluded (Social Finance, 2020; Timpson, 2019). Within this broader context, neurodivergent (ND) students – such as those with autism, attention deficit hyperactivity disorder (ADHD), dyslexia, and others (Walker, 2021) – are increasingly being recognised as an especially vulnerable group.

Introduction

School exclusion refers to the formal or informal removal of a student from the regular educational environment, tem- porarily or permanently (Department for Education, 2025). In the UK, the two most common types of official exclusion are permanent exclusions (removal from the school roll) and fixed-term exclusions or suspensions (tem- porary bars from attendance) (Department for Education, 2025). Informal exclusions include prolonged absence from school, moving or exiting schools early. Alongside these are unrecorded, often unregulated exclusions, which occur when pupils are asked by staff to stay home or experi- ence school environments in which they do not feel wel- come or able to attend (Martin-Denham, 2020; National Autistic Society, 2023).

The Department for Education statistics indicate a steady increase in both suspensions and permanent exclusions over the past decade (Department for Education, 2025). In 2023– 2024, there were over 900,000 suspensions and 10,000 per- manent exclusions, indicating a 21% and 16% increase from the previous year, respectively.

Studies show that pupils with SEN are seven to eight times more likely to be excluded than their peers (Gill et al., 2017; Social Finance, 2020), with neurodevelopmental con- ditions such as ADHD and autism particularly associated with school exclusion (John et al., 2022). For instance, chil- dren with ADHD are reported to be two to five times as likely to be excluded (Paget et al., 2015; Parker et al., 2015), and autistic students are nearly twice as likely to face permanent exclusion compared to neurotypical (NT) peers (Blacher & Eisenhower, 2022; Guldberg et al., 2021; John et al., 2022; Totsika et al., 2020). Social com- munication difficulties – often associated with autism – have also been linked to a nearly threefold increase in exclu- sion risk (Paget et al., 2015).

School exclusion is linked to a wide range of negative short- and long-term outcomes. Exclusions are associated with poorer academic achievement and employability, as well as physical and mental health problems (Ford et al., 2017; Martin-Denham, 2020; McLean & McVicar, 2025; Obsuth et al., 2022; Sutherland & Eisner, 2014; Tejerina-Arreal et al., 2020). At the community level, school exclusion is associated with increased antisocial behaviour, substance use, and involvement with the criminal justice sys- tem, contributing to the so-called ‘school-to-prison pipeline’ (Gerlinger et al., 2021; Hemphill et al., 2012; Henry et al., 2011; Sanders et al., 2018). The impact of exclusion also extends to families, placing additional financial, emotional, and logistical strain on parents/caregivers and their wider families (Cleary et al., 2024).

Qualitative research provides deeper insight into how ND students are often excluded in response to behaviours stemming from school-related anxieties, unmet needs, or misunderstood communication styles (Day, 2025). ND pupils, and autistic pupils in particular, have described school as a source of chronic distress driven by sensory demands, social expectations, and a lack of staff

Given these significant consequences, school exclusion is a pressing social issue and a key concern for education

reflects the experiences of all ND students, not only those with formal diagnoses.

understanding, contributing to school-based anxiety and attendance difficulties (Connolly et al., 2023; Fielding et al., 2025). Recent quantitative research further supports these findings, indicating that ND students report higher levels of school-based stress, emotional burden, lack of institu- tional support, and social isolation compared to their NT peers (Lukito et al., 2025).

This study aims to address these gaps by investigating school exclusion among self-identified ND students within a nationally representative sample, comparing self-reported exclusion rates between ND and NT students, and examin- ing the roles of demographic factors and SEN support.

Some research suggests that having formal support for SEN may offer a protective buffer against exclusion. In England, students with additional needs may receive SEN support, including school-funded provision such as targeted teaching-assistant support, adapted teaching strategies, or in-class accommodations. Where needs are more complex, an Education, Health and Care Plan (EHCP), a legally bind- ing document that guarantees specified educational, health, and social care provision, is issued by the local authority. An EHCP can include additional funding, therapies (e.g. speech and language support), or specialist school place- ment (Department for Education, 2024). Critically, eligibil- ity for both forms of support is needs-based rather than diagnosis-based, meaning that students without a formal clinical diagnosis, including many self-identified ND stu- dents in the present sample, can, in principle, still access support. A review of school exclusion data in England found that a protective effect of EHCPs is specific to per- manent exclusions and varies significantly depending on the child’s main needs (Timpson, 2019). This suggests that access to protective support is not uniform across all ND students but is instead contingent on the type of need that is formally recognised and documented, raising the question of what happens to those children whose neurodi- vergence goes unidentified or undiagnosed.

We hypothesised that: (1) ND students will report higher rates of school exclusion and permanent school exclusion than NT peers, across year groups and gender categories; (2) receiving support for SEN will have a protective effect against school exclusion in the ND group; and (3) neurodi- vergence will predict school exclusion above and beyond other risk factors including gender, ethnicity, age, and socio-economic deprivation.

Method

Study Design and Participants

This study used cross-sectional data from the 2023 ‘OxWell Student Survey’ study, a bi-annual non-identifiable survey of schools and further education colleges (institutions in England providing education and vocational training to stu- dents typically aged 16 and over, outside the compulsory school system) across four regions in England: Berkshire, Liverpool, Milton Keynes, and Oxfordshire. Pupils self- reported on their mental health, wellbeing, and school life (Mansfield et al., 2021). In 2023, students completed one of two age-appropriate surveys, either for Years 5–6 (ages 9–11), or 7–13 (ages 11–19+). Data collection took place during school time in February–March 2023, with volun- tary, anonymous participation and no monetary incentive. Pupils under 16 years provided assent (with parental opt-out available), while those 16 and over provided informed con- sent. After watching an information video, pupils were pro- vided with a unique login, instructions, and links for the 15– 40-min survey. Pupils could skip specific questions and end the survey at any time without compromising their participation.

While this body of work clearly demonstrates that ND students face disproportionate exclusion risks, most exist- ing studies rely on clinical diagnoses, school records, or parent and teacher reports, with few using self-identification despite its growing recognition within the neurodiversity movement (Ardeleanu et al., 2024; Van Rijswijk & Curșeu, 2025). Self-identified ND status has been shown to capture cognitive differences and predict outcomes above and beyond formal diagnosis in adult samples (Van Rijswijk & Curșeu, 2025). Many individuals who identify as ND may lack access to formal diagnoses due to systemic barriers such as long waiting lists, yet still face exclusion- related challenges – a concern echoed in England’s 2026 Schools White Paper, which proposes shifting support towards a needs-based model available regardless of diag- nosis (Department for Education, 2026). Existing studies also often focus on specific diagnoses rather than ND self- identification, potentially overlooking under- or misdiag- nosed groups such as girls, gender-diverse students, and ethnic minorities (Gellert et al., 2025; Hisle-Gorman et al., 2019; Roman-Urrestarazu et al., 2021). Investigating self- identified neurodivergence and school exclusion is there- fore essential for ensuring research and educational policy

In total, 43,735 pupils from 182 schools and 25 school districts participated in the OxWell 2023 Student Survey.

Participants were sequentially removed from the analytic sample if they completed the primary-school version of the survey (Years 5–6), which did not include the neurodiver- gence item (n = 9,250), completed the survey in under 10 min (n = 3,694), or did not provide consent or assent (n = 1,202). A further 14,335 pupils were not retained due to missing responses on key analytic variables (e.g. age, gender, ethnicity, socio-economic deprivation, neurodiver- gence, school exclusion, or SEN support). Responses were treated as missing if participants did not provide answers or selected ‘Prefer not to say’ for these variables. Two hundred and forty-two pupils aged 19 years or older were not included in the analyses. The final analytic sample

4 Neurodiversity 4(0)

comprised N = 15,012 pupils from 78 secondary schools across 21 school districts, including 3,081 pupils who self- identified as ND and 11,931 who identified as NT.

Socio-Economic Deprivation. Eight items about poverty (see Appendix B), such as ‘My family uses food banks’, were used as a proxy measure for socio-economic deprivation. The first seven items were adapted from the Wales’s Young People’s Survey on Child & Family Poverty 2019 (McFarlane, 2021), and the eighth was created by the OxWell Study Team. Response options were ‘Never or hardly ever’, ‘Some of the time’ and ‘Often’. A binary indi- cator was created by coding responses ‘Never or hardly ever’ and ‘Some of the time’ as 0, and ‘Often’ as 1. A count variable was computed by summing the responses (range = 0–8), with higher scores indicating greater socio-economic deprivation.

A sensitivity analysis was conducted in G*Power 3.1.9.7 to determine the smallest effects that could be detected with adequate power given the final analytic sample (N = 15,012). For the primary group comparison of neurodiver- gence by exclusion status, a sensitivity analysis for a χ[footnote 2]

test of independence (df = 1; α = .05; power = .80) indicated that the study was sufficiently powered to detect effects as small as w = 0.023. A further sensitivity analysis was con- ducted for the multivariable logistic regression examining predictors of school exclusion. Using a two-tailed α = .05, power = .80, total sample size N = 15,012, an exclusion prevalence of 0.063, and a neurodivergence prevalence of 0.205, the minimum detectable effect for a binary predictor corresponded to an odds ratio (Or) = 0.77, equivalent in magnitude to Or = 1.30 in the risk-increasing direction, assuming R2 other X = 0.05.

Neurodivergence. Participants answered the question, ‘Do you consider yourself to be dyslexic/dyspraxic, and/or aut- istic, and/or have ADHD (i.e. neurodivergent)?’ Response options were ‘Yes’, ‘No’, ‘Not Sure’, and ‘Prefer not to say’. After excluding ‘Not Sure’ and ‘Prefer not to say’ responses, participants were grouped into categories ND if answering ‘Yes’ or NT if answering ‘No’.

See Table 1 for the characteristics of the ND and NT groups. Group differences were observed across all demo- graphic characteristics. Most had negligible to small effect sizes (age, year group, and gender), while differences in eth- nicity, socio-economic deprivation, and SEN support were moderate to large.

School Exclusion. Participants were asked whether they had ever been excluded from school or college. Response options were ‘Yes’, ‘No’, and ‘Not Sure’. After excluding ‘Not Sure’ responses, participants were categorised as excluded or not excluded.

Materials

Demographic Characteristics. Participants provided demo- graphic information including age, school year, gender, eth- nicity, and questions relevant to socio-economic deprivation.

Permanent School Exclusion. Participants who reported hav- ing been excluded were asked whether any of those exclu- sions were permanent. Response options were ‘Yes’, ‘No’ and ‘Don’t know what this means’. After excluding ‘Don’t know what this means’ responses, participants were categorised as permanently excluded or not perman- ently excluded.

Age and School Year. These responses were given through dropdown options, with ages ranging from ‘11’ to ‘over 18’, and school years 7–13.

Support for SEN. Participants self-reported whether they receive support for SEN, including an EHCP. Although the survey item itself did not list specific types of provision, SEN support in England typically encompasses measures such as extra time in exams, one-to-one or small-group teaching assistance, individualised learning plans, speech and language therapy, or classroom adjustments, while an EHCP additionally guarantees legally specified provision and, in some cases, specialist placement. Response options were ‘Yes’, ‘No’, ‘Not Sure’ and ‘Prefer not to say’. After excluding ‘Not Sure’ and ‘Prefer not to say’ responses, par- ticipants were categorised as receiving SEN support or not receiving SEN support.

Gender. Participants were asked to report their gender iden- tity with options ‘Female’, ‘Male’, ‘Other’ and ‘Prefer not to say’, with an additional free-text option to self-identify. Free-text responses were reviewed and classified as ‘Gender diverse’ or coded as missing data following OxWell guidelines. After excluding missing data, partici- pants were grouped into gender categories: Girl, Boy, and Gender diverse.

Ethnicity. Pupils reported their ethnic group with 18 options on a dropdown list. The options were aggregated into five high-level ethnic groups (see Appendix A), consistent with that used by the (England and Wales) Office for National Statistics census (Office for National Statistics, 2022): White, Mixed/Multiple ethnic groups, Asian/Asian British, Black/African/Caribbean/Black British, and Other ethnic group.

Data Analysis

All data processing and statistical analyses were performed with R version 4.3.1. Bonferroni corrections were applied

Note. Effect sizes calculated using Cohen’s d or Cramer’s v. *p < .05, **p < .01, ***p < .001. Percentages represent within-group proportions, except Year Group, which is between groups. SEN: special educational needs.

Table 1. Demographic Characteristics of the Neurodivergent and Neurotypical Groups.

6 Neurodiversity 4(0)

2.8% of NT girls reported exclusion, χ2 = 249.40, p < .001, v = 0.18. Among gender-diverse students, exclusion rates did not differ significantly between ND (9.6%) and NT (6.4%) groups, χ2 = 1.46, p = .227, v = 0.06.

across all multiple comparisons, and adjusted p-values <.05 were interpreted as statistically significant.

To examine group differences between ND and NT par- ticipants, Welch two-sample t-tests and chi-squared tests (χ2) were used for continuous and categorical variables, respectively. Effect sizes (Cohen’s d for t-tests; Cramer’s v for χ2 tests) were also calculated to assess the magnitude of group differences. Additional χ2 and Fisher’s exact tests examined group differences in school exclusion rates, per- manent exclusions, and exclusion rates by year groups and gender. Group differences in school exclusion rates between students reporting SEN support and those not receiving support were examined using χ2 tests across the full sample and within ND and NT groups.

Year Groups. As shown in Table 2, ND students reported significantly higher exclusion rates than NT peers across all year groups, χ2 ≥12.72, p < .001. All p-values remained significant after Bonferroni correction, with small effect sizes (0.13 ≤v ≤0.21).

Year Groups and Gender. ND girls were significantly more likely than NT girls to be excluded in most year groups, holding for Years 7–12 (χ2 ≥8.91, p < .01; 0.12 ≤v ≤ 0.27; all significant after Bonferroni correction; Table 2). ND boys were also significantly more likely than NT boys to be excluded across all year groups (χ2 ≥16.87, p < .001; 0.15 ≤v ≤0.23; all significant after Bonferroni cor- rection; Table 2). Due to small group sizes, Fisher’s exact tests were conducted for girls in Year 13, and for all year groups among gender-diverse students; these comparisons showed no significant differences in exclusion rates. Frequencies for each neurodiversity and gender group are shown in Table 2.

To test whether neurodivergence predicted school exclu- sion beyond other risk factors, two multivariable logistic regressions were compared. The first model included four predictors (gender, ethnicity, age, and socio-economic deprivation), while the second model additionally included neurodiversity. Reference levels for the categorical pre- dictor variables were ‘Neurotypical’, ‘Girl’ and ‘White’. ORs and corresponding 95% CIs were reported. Model fit was evaluated using the Akaike information criterion (AIC), and likelihood ratio tests.

As an exploratory sensitivity analysis, the primary two- model structure was re-estimated with the outcome restricted to permanent exclusion rather than any school exclusion, to examine whether predictors of exclusion dif- fered for this more extreme subgroup. Given the small num- ber of permanent exclusion cases for some ethnicity categories, Model 2 was additionally re-estimated using Firth penalised logistic regression as a robustness check against rare-event bias.

Support for SEN and School Exclusion

Overall, students who reported receiving support for SEN were significantly more likely to report exclusion than those without SEN support, χ2 (1, N = 15,012) = 99.89, p < .001, v = .08 (negligible effect). Specifically, 12.1% of students with support had been excluded compared with 5.6% of those without support. Among ND students, the exclusion rate did not differ significantly between those with SEN support (14.9%) and those without (14.5%), χ2 (1, N = 3,081) = .043, p = 0.837, v = .005 (negligible effect). In contrast, NT students who received SEN support were sig- nificantly more likely to report exclusion (6.1%) than NT students without support (4.0%), χ2 (1, N = 11,931) = 4.60, p = .032, v = .02 (negligible effect).

Results

Neurodivergence and School Exclusion

The ND group reported significantly higher rates of school exclusion than the NT group (14.6% vs. 4.1%), χ2 (1, N = 15,012) = 462.81, p < .001, v = .18 (small effect). Nearly half (48.0%) of all excluded students identified as ND com- pared with 20.5% of the overall sample.

Risk Factors and School Exclusion

In Model 1 (Table 3), gender, age, and socio-economic deprivation significantly predicted school exclusion, along with several ethnicities, after holding other factors constant. With neurodiversity added in Model 2, neurodiversity, age, and socio-economic deprivation remained significant pre- dictors, along with some gender identities and ethnicities. A comparison of the two models is presented in Appendix C.

Permanent Exclusions. Significantly more ND students (2.3% of all ND students and 15.7% of excluded ND stu- dents) reported having been permanently excluded from school compared with their NT peers (0.4% of all NT stu- dents and 9.4% of excluded NT students), χ2 (1, N = 891) = 477.78, p < .001, v = .18 (small effect).

Gender. Both ND boys and ND girls reported significantly higher exclusion rates than their NT peers: 17.8% of ND boys versus 5.5% of NT boys were excluded, χ2 = 227.16, p < .001, v = 0.18; and 12.5% of ND girls versus

Model Comparisons. Model 1 had an AIC value of 6755.62, while Model 2 had a lower AIC value of 6442.46, indicat- ing improved model fit. The likelihood ratio test comparing

Table 2. Number and Percentage of Excluded Neurodivergent and Neurotypical Students, by Year Group and Gender, With Group Differences.

(continued) 8 Neurodiversity 4(0) Table 2. Continued.

Note. Effect size calculated using Cramer’s v. *p < .05, **p < .01, ***p < .001. #Cells with counts <5 were suppressed to protect participant anonymity. Percentages indicate the proportion of excluded students within each neurodiversity and year group.

the two models was significant, χ2 (1, N = 14,997) = 315.17, p < .001. Together, these tests suggest that includ- ing neurodiversity in Model 2 improved overall fit beyond the demographic and socio-economic predictors in Model 1.

exclusion (n = 117) specifically (Appendix D). ND students had 4.75 times the odds of permanent exclusion compared with NT peers, 95% CI [3.19, 7.05], a higher estimated effect than for exclusion overall (Or = 3.84). Boys had 1.81 times the odds of permanent exclusion compared with girls, 95% CI [1.22, 2.69]. Older students, 95% CI [1.05, 1.27], and those with greater socio-economic depriv- ation, 95% CI [1.07, 1.29], also showed elevated odds. Compared with White students, those from Mixed/ Multiple ethnic groups had 1.80 times the odds of perman- ent exclusion, 95% CI [1.03, 3.15], while Asian/Asian British students had lower odds, 95% CI [0.18, 0.98]. Effects for gender-diverse students and for Black/African/ Caribbean/Black British and Other ethnic groups were not significant in this subgroup. Firth penalised logistic regres- sion produced consistent estimates (Appendix D).

Predictors. Results from the multivariable logistic regres- sion (Model 2) are shown in Table 3.

Neurodiversity. ND students had 3.84 times the odds of reporting school exclusion compared with NT peers, 95% confidence interval (CI) [3.32, 4.44].

Gender. Boys had 1.82 times the odds of exclusion com- pared to girls, 95% CI [1.58, 2.10]. Specifically, 8.0% of boys and 4.6% of girls reported exclusion. The effect for gender-diverse students was not significant.

Discussion

Ethnicity. Compared with White students, those from Mixed/Multiple ethnic groups had 1.35 times the odds of exclusion, 95% CI [1.05, 1.72]; students from Black/ African/Caribbean/Black British backgrounds had 1.79 times the odds, 95% CI [1.34, 2.29]; and students from Other ethnic groups had 1.92 times the odds, 95% CI [1.42, 2.50] of exclusion.

This study examined school exclusion and associated risk factors in a nationally representative sample of 3,081 self- identified ND secondary school students and 11,931 NT peers. As hypothesised, ND students reported significantly higher rates of both overall and permanent school exclu- sion. This disparity persisted across genders and most school year groups. Receiving SEN support was associated with higher school exclusion risk among NT students but showed no association with exclusion rates for ND students. Self-identified neurodivergence remained a strong inde- pendent predictor of exclusion, after adjusting for gender, ethnicity, age, and socio-economic deprivation. Together, these findings suggest that ND students are particularly vul- nerable to exclusion regardless of formal recognition or support status, pointing less to a need for earlier diagnosis, and more to a need for schools to treat behaviour that cur- rently leads to exclusion as an opportunity to identify and provide support, rather than grounds for punishment.

Age. For each additional year of age, the odds of exclusion increased by a factor of 1.18, 95% CI [1.13, 1.22].

Socio-Economic Deprivation. For each one-unit increase on the socio-economic deprivation measure, the odds of exclu- sion increased by a factor of 1.10, 95% CI [1.06, 1.15].

Sensitivity Analysis: Permanent Exclusions

As an exploratory sensitivity analysis, Models 1 and 2 were re-estimated with the outcome restricted to permanent

Table 3. Model 2: Logistic Regression Predicting School Exclusion.

Note. Reference categories are Neurotypical, Girl, and White. B = unstandardized coefficient; SE = standard error of the coefficient; β = standardised coefficient. *p < .05, **p < .01, ***p < .001.

Neurodivergence and Exclusion

exclusion processes may not consistently identify or address ND students’ needs. In some cases, exclusion may function as a mechanism for accessing additional sup- port or alternative provision; however, persistently ele- vated exclusion rates point to the importance of earlier identification and proactive support within mainstream school settings.

Consistent with Hypothesis 1, ND students reported sig- nificantly higher rates of school exclusion (14.6%) and permanent exclusions (2.3%) compared to NT students (4.1% and 0.4%, respectively), representing a more than threefold increase. An exploratory sensitivity analysis confirmed this pattern: consistent with the overall exclu- sion finding, permanent exclusion disproportionately occurred among ND relative to NT students, although this affected relatively fewer students overall. These find- ings align with prior research documenting heightened exclusion risk among students with ADHD, autism, and communication difficulties (John et al., 2022; Paget et al., 2015; Parker et al., 2015), while extending this evidence to students who self-identify as ND. This pattern was evident across gender categories and school years, when stratify- ing by both. Although small subgroup sizes for gender- diverse students and Year 13 girls limited reliable infer- ence, the overall trend suggests school disciplinary and

Gender-diverse students showed a notable intersection of high neurodivergence rates and elevated exclusion risk. In Model 1, gender diversity predicted increased exclusion, but this effect was no longer significant once neurodiver- gence was included. This aligns with research showing ele- vated rates of autism and other neurodevelopmental conditions among transgender and gender-diverse indivi- duals (Warrier et al., 2020), suggesting the exclusion risk here may partly reflect overlapping neurodivergence. Given the particular vulnerabilities of gender-diverse youth, further research should explore their exclusion experiences in greater depth.

10 Neurodiversity 4(0)

SEN Support

Neurodivergence as an Independent Predictor

The second key finding concerns SEN support. Contrary to expectations and some previous literature (Timpson, 2019), SEN support was associated with higher exclu- sion rates across the overall sample and among NT stu- dents, although the effect size was small. Among ND students, exclusion rates did not differ by support status, suggesting SEN provision did not mitigate their heigh- tened risk.

The final key finding, consistent with Hypothesis 3, is that neurodivergence predicts school exclusion rates beyond other established risk factors, including gender, ethnicity, age, and socio-economic deprivation. Even after controlling for these variables, ND students had almost four times the odds of being excluded compared to their NT peers, consist- ent with previous estimates for specific diagnostic groups of three- to five-fold increases (Guldberg et al., 2021; Paget et al., 2015). These prior estimates, however, were derived from smaller, often self-selected samples. By contrast, the present study drew on a large, regionally diverse, population-based sample in which neurodivergence was self-identified rather than clinically confirmed. The similar- ity of risk estimates across such different samples and mea- sures is itself notable. It suggests the higher exclusion risk linked to neurodivergence may not just be a by-product of who gets referred for or receives diagnosis but reflects a broader pattern affecting ND young people in general, including the many who have never received, or used a for- mal diagnostic label. This reinforces the need for schools to move beyond diagnosis-based eligibility criteria, focusing instead on need, and fostering environments where ND stu- dents are understood, supported, and able to succeed.

Several factors may help explain these patterns. First, the SEN support measure was self-reported and likely cap- tured a range of informal or reactive provision rather than formally structured EHCPs. Second, students receiving support may represent those with the most complex aca- demic, social, or behavioural needs – already at elevated risk of exclusion. Third, because SEN support was assessed independently of neurodivergence status, some group overlap or misclassification was possible: students with ND-relevant traits may have been categorised as NT, which could attenuate rather than inflate observed group differences. Some students may receive SEN sup- port for reasons unrelated to neurodivergence (e.g. phys- ical or sensory disabilities or mental health needs; Department for Education & Department of Health, 2015). Future research linking self-identification to formal SEN categorisations would be informative. Fourth, exist- ing support may focus on academic or learning needs rather than the emotional and environmental factors that often precede exclusion, and interventions may be imple- mented only after difficulties escalate, limiting their pre- ventative effectiveness.

Strengths and Limitations

This study has several notable strengths, including strong statistical power and a focus on self-identified neurodiver- gence – an approach aligned with the neurodiversity para- digm and inclusive of underdiagnosed populations (Walker, 2021). However, several limitations should be acknowledged. First, SEN support measures lacked detail on the type or intensity of provision. Second, missing data reduced subgroup sizes, particularly among gender- diverse students, limiting the reliability of some compari- sons. Third, students not attending mainstream school due to exclusion, avoidance, or other factors were not included in the survey, likely excluding some of the most vulnerable young people. Fourth, the survey did not collect data on for- mal clinical diagnosis, medication use, or specific type of support received. Because it was not possible to determine how many self-identified ND students also had a formal diagnosis, this study cannot establish the extent to which self-identification, independent of diagnosis, is associated with exclusion; our findings may therefore overlap with a diagnosis-based approach. Relatedly, self-identification may involve measurement constraints in both directions: some students, particularly younger respondents, may not recognise clinical terms such as ‘ADHD’, ‘autistic’ or ‘dys- lexic/dyspraxic’ despite exhibiting elevated traits, while others with a formal diagnosis may decline to self-identify as ND. Fifth, socio-economic deprivation was measured using a self-report poverty index rather than a standardised,

The protective effects reported in the Timpson Review (2019) – which included both state-funded mainstream and special schools (i.e. schools specifically organised to provide education for pupils with SEN) – may partly reflect the stronger safeguards, more structured provision, and closer monitoring typically found in specialist set- tings. Notably, formal protections in the Review reduced permanent exclusions specifically but were not associated with fixed-term exclusions or day-to-day disciplinary response, suggesting formal recognition may shield stu- dents from the most severe sanction without changing how disruptive behaviour is addressed more broadly. This reinforces the argument that reducing exclusion risk may depend less on diagnostic recognition and more on reforming how schools respond to perceived disruptive behaviour. However, the present cross-sectional design limits inferences about the timing and direction of these associations. Longitudinal research is needed to clarify whether SEN support operates preventively or is imple- mented reactively, distinguishing formal from informal support (e.g. EHCPs vs. ad hoc assistance) and assessing whether provision meets ND learners’ academic, social, and emotional needs.

area-based measure (e.g. index of multiple deprivation [IMD]), as individual postcode data were not available; a unit increase is therefore harder to interpret. Notably, self- reported indicators of financial hardship have been found to be more strongly associated with health outcomes than area-based IMD measures in other UK samples (Homer et al., 2023), suggesting the self-report approach used here may carry sensitivity advantages alongside its limita- tions. Finally, the cross-sectional design limits causal infer- ence, and longitudinal research is needed to clarify the temporal relationships between neurodivergence, support, and exclusion outcomes. Given these measurement limita- tions and the specific regional sampling frame, the preva- lence rates reported here should not be interpreted as epidemiological estimates of neurodivergence in the broader population.

for other established risk factors. While previous research has often focused on specific diagnostic groups such as aut- ism or ADHD, these results highlight that self-identified ND young people more broadly represent a particularly vul- nerable group, with this vulnerability most pronounced for the most severe disciplinary outcome, permanent exclusion. Given the well-documented educational, social, and health consequences of exclusion, there is an urgent need to move beyond a diagnosis-driven model of recognition and instead reform behaviour policies so that ND students are offered support rather than punishment when their behav- iour currently places them at disproportionate risk of exclu- sion – promoting meaningful inclusion for ND young people.

Acknowledgements We would like to express our gratitude to all the students who took part in the OxWell Student Survey, as well as to the staff at partici- pating schools and further education colleges for the significant time and effort they devoted to organising and delivering the survey.

Implications and Future Directions

These findings carry important implications for policy and practice. Existing evidence suggests pupils awaiting an autism diagnosis face heightened exclusion risk, yet diagnosis alone does not guarantee that schools can meet their needs (Chapman, 2023) – underscoring that recognition alone is not sufficient for change. Instead, these findings point to the need for schools to reconsider how behaviour policies respond to ND students, so that behaviour currently met with exclu- sion is treated as a signal for additional support rather than grounds for punishment, regardless of diagnostic or support status. A needs-led approach, empowering schools to recog- nise and respond to support needs earlier, may reduce reliance on clinical services and lower exclusion rates.

ORCID iDs Lili Bokor https://orcid.org/0009-0002-4160-4401 Francesca Happé https://orcid.org/0000-0001-9226-4000

Ethical Considerations and Informed Consent Statements Participating schools were mainly recruited through local author- ities, which enrol students using either (1) a parental opt-out approach with student assent for those under 16, or (2) informed consent from students aged 16 and above. All procedures involv- ing human participants were approved on 19 January 2023 by the University of Oxford’s Medical Sciences Interdivisional Research Ethics Committee (Ms Idrec; R62366/RE0014).

Research consistently shows school distress is closely tied to neurodivergence, with autistic children disproportionately affected (Connolly et al., 2023). ND children are more likely to encounter adverse school experiences such as bullying or unaccommodating environments, which may be experienced as distressing or traumatic (Fielding et al., 2025). Given estab- lished links between trauma, chronic stress, and school exclu- sion, trauma-informed, family-based approaches show promise for improving attendance, reducing exclusion, and lowering later youth justice involvement (Day, 2025). Although school-based interventions can reduce exclusion, many appear to have only temporary effects, and further evaluation is needed to identify strategies with lasting impact (Valdebenito et al., 2019). Future research could extend beyond diagnostic categories to capture the broader spectrum of neurodevelopmental needs, emotional distress, and adequacy of school support. It is essential to examine which interventions produce sustainable reduction in exclusions, particularly for especially vulnerable groups such as gender- diverse and ND students.

Funding The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The OxWell Student Survey was funded by the National Institute for Health and Care Research (NIHR) Applied Research Collaboration Oxford and Thames Valley at Oxford Health National Health Service (NHS) Foundation Trust (M.F.) and the Westminster Foundation (E.S.) with support from Liverpool City Council, Milton Keynes City Council, NHS Frimley Integrated Care Board (ICB), NHS Buckinghamshire, Oxfordshire and Berkshire West ICB and Oxfordshire City Council.

Declaration of Conflicting Interests This study demonstrates that self-identified ND students have almost four times the odds of being excluded from school compared to their NT peers, even after accounting The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

12 Neurodiversity 4(0)

Fielding, C., Streeter, A., Riby, D. M., & Hanley, M. (2025).

Data Availability Statement The data supporting these findings are available upon reasonable request. Researchers can access the dataset by applying through the BrainWaves Data Portal, where all applications are reviewed to ensure appropriate use. Additional information – including the full questionnaire, study protocol, and supporting materials – can be found on the OxWell project’s Open Science Framework page (https://osf.io/sekhr/).

Neurodivergent pupils’ experiences of school distress and attendance difficulties. Neurodiversity, 3, Article 2754633025 1327056. https://doi.org/10.1177/27546330251327056 Ford, T., Parker, C., Salim, J., Goodman, R., Logan, S., & Henley,

W. (2017). The relationship between exclusion from school and mental health: A secondary analysis of the British child and adolescent mental health surveys 2004 and 2007. Psychological Medicine, 48(4), 629–641. https://doi.org/10. 1017/s003329171700215x Gellert, B., Ostrowski, J., Pinkas, J., & Religioni, U. (2025).

References Ardeleanu, K., Steinberg, H., Garfield, T., Voltaire, S., Shea, L.,

Underdiagnosed and misunderstood: Clinical challenges and educational needs of healthcare professionals in identifying autism spectrum disorder in women. Behavioral Sciences, 15(8), 1073. https://doi.org/10.3390/bs15081073 Gerlinger, J., Viano, S., Gardella, J. H., Fisher, B. W., Curran, C.,

Brown, M., Chvasta, K., & Tan, C. D. (2024). Self-identification of autism: Why someautistic adultslacka clin- ical diagnosis and why this matters for inclusion. Autism, 29(9), 2344–2355. https://doi.org/10.1177/13623613241297222 Blacher, J., & Eisenhower, A. (2022). Preschool and child-care

Higgins, F., & Higgins, E. M. (2021). Exclusionary school dis- cipline and delinquent outcomes: A meta-analysis. Journal of Youth and Adolescence, 50(8), 1493–1509. https://doi.org/10. 1007/s10964-021-01459-3 Gill, K., Quilter-Pinner, H., & Swift, D. (2017). Making the differ-

expulsion: Is it elevated for autistic children? Exceptional Children, 89(2), 178–196. https://doi.org/10.1177/ 00144029221109234 Chapman, E. (2023). Preventing unmet need from leading to

ence: Breaking the link between school exclusion and social exclusion. Institute for Public Policy Research. https://ippr- org.files.svdcdn.com/production/Downloads/making-the- difference-report-october-2017.pdf Government Equalities Office, & Equality and Human Rights

school exclusion: Empowering schools to identify neurodiver- sity earlier [Doctoral dissertation, University of Leeds]. https:// etheses.whiterose.ac.uk/id/eprint/33429/ Cleary, M., West, S., Johnston-Devin, C., Kornhaber, R., McLean,

L., & Hungerford, C. (2024). Collateral damage: The impacts of school exclusion on the mental health of parents caring for autistic children. Issues in Mental Health Nursing, 45(1), 3–8. https://doi.org/10.1080/01612840.2023.2280718 Connolly, S. E., Constable, H. L., & Mullally, S. L. (2023). School

Commission. (2013, February 27). Equality Act 2010: Guidance. Gov.UK. https://www.gov.uk/guidance/equality- act-2010-guidance Guldberg, K., Wallace, S., Bradley, R., Perepa, P., Ellis, L., &

MacLeod, A. (2021). Investigation of the causes and implica- tions of exclusion for autistic children and young people in England. University of Birmingham. https://research. birmingham.ac.uk/en/publications/investigation-of-the-causes- and-implications-of-exclusion-for-aut Hemphill, S. A., Heerde, J. A., Herrenkohl, T. I., Toumbourou, J.

distress and the school attendance crisis: A story dominated by neurodivergence and unmet need. Frontiers in Psychiatry, 14, 1237052. https://doi.org/10.3389/fpsyt.2023.1237052 Day, A. (2025). Kids ‘at risk’ of school exclusion and youth justice

involvement? Or neurodivergent children and families in need of trauma-informed support? Emotional and Behavioural Difficulties, 30(3), 196–211. https://doi.org/10.1080/13632752. 2025.2499788 Department for Education. (2024). Suspension and permanent

W., & Catalano, R. F. (2012). The impact of school suspension on student tobacco use. Health Education & Behavior, 39(1), 45–56. https://doi.org/10.1177/1090198111406724 Henry, K. L., Knight, K. E., & Thornberry, T. P. (2011). School

exclusion from maintained schools, academies and pupil refer- ral units in England, including pupil movement: Guidance for maintained schools, academies, and pupil referral units in England. Gov.UK. https://www.gov.uk/government/ publications/school-exclusion Department for Education. (2025, July 10). Suspensions and per-

disengagement as a predictor of dropout, delinquency, and problem substance use during adolescence and early adult- hood. Journal of Youth and Adolescence, 41(2), 156–166. https://doi.org/10.1007/s10964-011-9665-3 Hisle-Gorman, E., Landis, C. A., Susi, A., Schvey, N. A., Gorman,

G. H., Nylund, C. M., & Klein, D. A. (2019). Gender dysphoria and autism spectrum disorder: A narrative review. LGBT Health, 6(3), 95–100. https://doi.org/10.1089/lgbt.2018.0252 Homer, K., Taylor, J., Miller, A., Pickett, K., Wilson, L., &

manent exclusions in England, academic year 2023/24. Explore our statistics and data – Gov.UK. https://explore- education-statistics.service.gov.uk/find-statistics/suspensions- and-permanent-exclusions-in-england/2023-24 Department for Education. (2026). Every child achieving and

Robson, J. (2023). Making ends meet – relating a self-reported indicator of financial hardship to health status. Journal of Public Health, 45(4), 888–893. https://doi.org/10.1093/ pubmed/fdad161 John, A., Friedmann, Y., DelPozo-Banos, M., Frizzati, A., Ford,

thriving (CP 1508-I). Gov.UK. Department for Education & Department of Health. (2015). SEND

code of practice: 0 to 25 years. https://assets.publishing. service.gov.uk/media/5a7dcb85ed915d2ac884d995/SEND_ Code_of_Practice_January_2015.pdf

T., & Thapar, A. (2022). Association of school absence and exclusion in children with psychiatric disorder or impairing psychopathology: A systematic review. Emotional and Behavioural Difficulties, 20(3), 229–251. https://doi.org/10. 1080/13632752.2014.945741 Roman-Urrestarazu, A., Van Kessel, R., Allison, C., Matthews, exclusion with recorded neurodevelopmental disorders, mental disorders, or self-harm: A nationwide, retrospective, electronic cohort study of children and young people in Wales, UK. The Lancet Psychiatry, 9(1), 23–34. https://doi.org/10.1016/s2215- 0366(21)00367-9 Lukito, S., Chandler, S., Kakoulidou, M., Griffiths, K., Wyatt, A., F. E., Brayne, C., & Baron-Cohen, S. (2021). Association of race/ethnicity and social disadvantage with autism prevalence in 7 million school children in England. JAMA Pediatrics, 175(6), e210054. https://doi.org/10.1001/jamapediatrics.2021. 0054 Sanders, J., Liebenberg, L., & Munford, R. (2018). The impact of Funnell, E., Pavlopoulou, G., Baker, S., Stahl, D., & Sonuga-Barke, E. (2025). Emotional burden in school as a source of mental health problems associated with ADHD and/or autism: Development and validation of a new co-produced self-report measure. Journal of Child Psychology and Psychiatry, 66(10), 1577–1592. https://doi. org/10.1111/jcpp.70003 Mansfield, K. L., Puntis, S., Soneson, E., Cipriani, A., Geulayov, school exclusion on later justice system involvement: Investigating the experiences of male and female students. Educational Review, 72(3), 386–403. https://doi.org/10.1080/ 00131911.2018.1513909 Social Finance. (2020). Who’s at risk of exclusion? An analysis in G., & Fazel, M. (2021). Study protocol: The OxWell school survey investigating social, emotional and behavioural factors associated with mental health and well-being. BMJ Open, 11(12), e052717. https://doi.org/10.1136/bmjopen-2021- 052717 Martin-Denham, S. (2020). A review of school exclusion on the Cheshire West and Chester. https://www.socialfinance.org.uk/ insights/whos-at-risk-of-exclusion Sutherland, A., & Eisner, M. (2014). The treatment effect of school exclusion on unemployment. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2380956 Tejerina-Arreal, M., Parker, C., Paget, A., Henley, W., Logan, S., mental health, well-being of children and young people in the City of Sunderland. University of Sunderland. https://sure.- sunderland.ac.uk/id/eprint/11940/ McFarlane, K. (2021). Report on the 5th Annual Child and Family Emond, A., & Ford, T. (2020). Child and adolescent mental health trajectories in relation to exclusion from school from the avon longitudinal study of parents and children. Child and Adolescent Mental Health, 25(4), 217–223. https://doi. org/10.1111/camh.12367 Timpson, E. (2019). Timpson review of school exclusion.

Poverty Surveys 2021. Children in Wales. https://www. childreninwales.org.uk/application/files/7416/3162/4537/ Child_Poverty_Survey_Report_2021_English_v7.pdf McLean, A., & McVicar, D. (2025). Does being excluded from

school harm student achievement? Evidence from siblings in English population data. https://doi.org/10.2139/ssrn.5385305 National Autistic Society. (2023, May 30). Education report 2023.

Gov.UK. https://assets.publishing.service.gov.uk/govern- ment/uploads/system/uploads/attachment_data/file/807862/ Timpson_review.pdf Totsika, V., Hastings, R. P., Dutton, Y., Worsley, A., Melvin, G.,

https://www.autism.org.uk/what-we-do/news/education- report-2023 Obsuth, I., Madia, J. E., Daniels, H., Thompson, I., & Murray, A.

Gray, K., Tonge, B., & Heyne, D. (2020). Types and correlates of school non-attendance in students with autism spectrum dis- orders. Autism, 24(7), 1639–1649. https://doi.org/10.1177/ 1362361320916967 Valdebenito, S., Eisner, M., Farrington, D. P., Ttofi, M. M., &

L. (2022). The impact of school exclusion in childhood on health outcomes in adulthood: Estimating causal effects using inverse probability of treatment weighting. https://doi.org/10. 31234/osf.io/v9zm5 Office for National Statistics. (2022, November 29). Ethnic group,

Sutherland, A. (2019). What can we do to reduce disciplinary school exclusion? A systematic review and meta-analysis. Journal of Experimental Criminology, 15(3), 253–287. https://doi.org/10.1007/s11292-018-09351-0 Van Rijswijk, J., & Curșeu, P. L. (2025). Formal versus self-

England and Wales. Home – Office for National Statistics. https://www.ons.gov.uk/peoplepopulationandcommunity/ culturalidentity/ethnicity/bulletins/ethnicgroupenglandandwales/ census2021#ethnic-groups-in-england-and-wales Paget, A., Parker, C., Henley, W., Heron, J., Ford, T., & Emond,

identified neurodivergence: A comparative study in work environments. Behavioral Sciences, 15(4), 420. https://doi. org/10.3390/bs15040420 Walker, N. (2021). Neuroqueer heresies: Notes on the Neurodiversity paradigm, autistic empowerment, and Postnormal possibilities. Autonomous Press. https://ebookcen- tral.proquest.com/lib/ed/detail.action?docID=6870963 Warrier, V., Greenberg, D. M., Weir, E., Buckingham, C., Smith,

A. (2015). Which children and young people are excluded from school? Findings from the avon longitudinal study of parents and children (ALSPAC). The Lancet, 385, S76. https://doi. org/10.1016/s0140-6736(15)60391-9 Palikara, O., Castro, S., Gaona, C., & Eirinaki, V. (2018).

Professionals’ views on the new policy for special educational needs in England: Ideology versus implementation. European Journal of Special Needs Education, 34(1), 83–97. https:// doi.org/10.1080/08856257.2018.1451310 Parker, C., Whear, R., Ukoumunne, O. C., Bethel, A., Thompson-Coon, J., Stein, K., & Ford, T. (2015). School

P., Lai, M., Allison, C., & Baron-Cohen, S. (2020). Elevated rates of autism, other neurodevelopmental and psychiatric diagnoses, and autistic traits in transgender and gender-diverse individuals. Nature Communications, 11(1), 3959. https://doi. org/10.1038/s41467-020-17794-1

14 Neurodiversity 4(0)

Appendix A. Original Options and Aggregated Groups of Ethnicities.

Survey Ethnicity Options Aggregated Ethnicity Groups

White – English/Welsh/Scottish/Northern Irish/British White

White – Irish White – Gypsy or Irish traveller White – Other Mixed/Multiple ethnic groups – White and Black Caribbean Mixed/Multiple ethnic groups Mixed/Multiple ethnic groups – White and Black African Mixed/Multiple ethnic groups – White and Asian Mixed/Multiple ethnic groups – Other Asian/Asian British – Indian Asian/Asian British Asian/Asian British – Pakistani Asian/Asian British – Bangladeshi Asian/Asian British – Chinese Asian/Asian British – Other Black/African/Caribbean/Black British – African Black/African/Caribbean/Black British Black/African/Caribbean/Black British – Caribbean Black/African/Caribbean/Black British – Other Arab Other ethnic group Other ethnic group Appendix B. Original Poverty and Socio-Economic Deprivation Survey Items, Frequencies and Percentages of Their Answers.

(continued) Appendix B. Continued.

Never or Hardly Ever

Some of the Time Often

Item N (%) At school, I am unable to afford to eat. 14,082 (93.1%) 740 (4.9%) 297 (2.0%) At home, I do not have enough space to do things like homework or chill out. 13,339 (88.2%) 1,274 (8.4%) 506 (3.3%) At home, I have no internet access or poor internet access. 14,108 (93.3%) 809 (5.4%) 202 (1.3%) At home, I go to bed hungry because there is not enough food in the house. 14,567 (96.4%) 429 (2.8%) 123 (0.8%) Appendix C. Independent Predictors of School Exclusion in Models 1 and 2.

Note. Reference categories are Neurotypical, Girl, and White. Or = odds ratio; CI = confidence interval. *p < .05, **p < .01, ***p < .001.

16 Neurodiversity 4(0)

Appendix D. Exploratory Sensitivity Analysis: Independent Predictors of Permanent School Exclusion.

Model 1 Model 2 Firth Regression

Variable Or [95% Ci] p value Or [95% Ci] p value Firth Or [95% Ci] Firth p Neurodiversity Neurotypical – – 1 Neurodivergent 4.75 [3.19, 7.05] <.001 4.72 [3.20, 6.96] <.001 Gender Girl 1 1 Boy 1.85 [1.25, 2.73] .002 1.81 [1.22, 2.69] .003 1.81 [1.23, 2.66] .003 Gender-diverse 2.74 [1.30, 5.77] .008 1.66 [0.79, 3.52] .183 1.73 [0.84, 3.58] .139 Ethnicity White 1 1 Mixed/Multiple ethnic groups 1.69 [0.97, 2.96] .064 1.80 [1.03, 3.15] .041 1.84 [1.06, 3.19] .029 Asian/Asian British 0.27 [0.12, 0.62] .002 0.42 [0.18, 0.98] .046 0.45 [0.20, 1.02] .055 Black/Black British/African/Caribbean 0.75 [0.33, 1.72] .496 1.08 [0.47, 2.52] .854 1.16 [0.52, 2.60] .715 Other ethnic group 1.28 [0.61, 2.68] .508 1.78 [0.84, 3.75] .131 1.87 [0.91, 3.85] .088 Age 1.17 [1.06, 1.29] .002 1.16 [1.05, 1.27] .003 1.16 [1.05, 1.27] .003 Socio-economic deprivation 1.26 [1.16, 1.38] < .001 1.17 [1.07, 1.29] .001 1.18 [1.07, 1.29] < .001 Note. Reference categories are Neurotypical, Girl, and White. Or = odds ratio; CI = confidence interval. *p < .05, **p < .01, ***p < .001.