Suspensions in QLD state schools, 2016–2020: overrepresentation, intersectionality and disproportionate risk

Abstract

Overrepresentation in exclusionary school discipline is extensively documented for certain groups of students, particularly students of colour and those with disability, yet while these groups may overlap, policies that aim to address disproportionality tend to consider equity factors in isolation. The majority of research on overrepresentation, intersectionality and disproportionate risk has been conducted in the United States, yet there has been limited research within an Australian context despite reports of the overrepresentation of students in priority equity groups: students with a disability, Indigenous students and those living in out-of-home care. To disentangle the intersectionalities between these three priority equity groups, we created seven independent groups in which one or more of these factors was present. We then compare each of these seven groups to one referent group (not Indigenous, no disability, not in care) to examine trends in Queensland state school suspensions from 2016 to 2020. Findings show that students in the seven groups were issued suspensions at rates disproportionate to enrolments, in contrast to the referent group, with the risk of suspension increasing with greater intersectionality. Further analyses highlight disability as a common underlying factor, both in terms of increasing risk of suspension and in its repeated use, particularly for students recorded as receiving adjustments in the Social-Emotional category. These findings underscore the need for more nuanced reform and support strategies that can better account for the intersectionalities between groups.

Introduction

Sharp increases in the use of exclusionary school discipline in Queensland (QLD) state schools have been periodically reported in the media since 2014, when the QLD government weakened legislative thresholds and increased principals’ powers to suspend (Graham et al., 2020). Media reports typically use Student Disciplinary Absence (SDA)¹ data that has been made publicly available on the QLD Department of Education website; however, these data are limited. Only the number of suspension, exclusion and enrolment cancellation incidents is reported and not the number of students associated with those incidents. Increases in the number of incidents reported may give the impression that there are very many ‘badly behaved students’ who are increasing in number, when the reality could instead be that there is a much smaller number of very vulnerable students receiving very many SDAs. The latter scenario is very different to the former and requires careful analysis and a multidimensional evidence-based response. For any response to be effective, however, its development needs to be informed by deep understanding of who is most at risk of exclusionary discipline in QLD state schools, for which behaviours it is being used and how often. This approach will also reliably indicate whether a disciplinary response is what those students really need or whether there are other deep-seated problems that need to be addressed first.

For example, large increases in the number of preparatory (Prep) year children being suspended have been reported since the number of prep suspensions doubled in 2015 (Bruce, 2015). Previous research using publicly available data pointed to two possible factors: the 2014 change in legislation which weakened safeguards, such as parents’ right to appeal (Carden, 2018), in combination with lowering of the school starting age (Graham et al., 2020). The QLD government initially responded to the concerns raised by encouraging the use of developmentally appropriate pedagogies in Prep, but these previous analyses using publicly available data could not identify which prep children are being suspended, and whether developmentally appropriate pedagogies are the right solution for them. Developmentally appropriate pedagogies, whilst worthwhile and important, will not address problems created by mainstream school environments that are inaccessible to students with high-incidence disabilities like Attention-Deficit Hyperactivity Disorder (Graham & Tancredi, 2019). Nor will they address the needs of an Indigenous student with a disability, for example, who is living in care and experiencing ongoing trauma. Such students require nuanced policy responses that have been crafted to prevent the siloing of intersecting groups.

The question, therefore, should not be how many prep children are receiving SDAs, but which children across all year groups are receiving SDAs and what proportion of them is Indigenous, has a disability and/or is living in out-of-home care? This information is important; first, because it will highlight which types of supports are needed, and second, it will provide much needed context on the use of school suspension. This is necessary in a political discourse environment that frames increases in suspension² as the result of rising student indiscipline coupled with inadequate teacher preparation, effacing the possible contributions of policy and practice, as well as social, racial and political injustices (Rudolph & Thomas, 2023).

Questions about the risk of suspension for members of different groups and whether these risks are being adequately addressed through departmental policies are difficult to answer without access to longitudinal data disaggregated by equity group like that made available in the Inquiry into Suspension, Exclusion and Expulsion processes in South Australian government schools (hereafter referred to as ‘SA Inquiry’, Graham et al., 2020). However, answers to questions such as these are essential if we really want to achieve excellence and equity in Australian school education. In an effort to discover the real state of affairs for children and young people attending QLD state schools, Queensland Advocacy for Inclusion (QAI) submitted two Right to Information requests to the QLD Department of Education for the overall numbers of short-term suspensions, long-term suspensions, exclusions and cancellations of enrolment for the five years from 2016 to 2020. These data were requested for students in three priority equity groups: students with disability, students living in out-of-home care (OOHC) and students who identify as Aboriginal and/or Torres Strait Islander (hereafter Indigenous students). Preliminary analyses revealed that Indigenous students and students with a disability are ‘three times more likely to be suspended than the general population’ in QLD state schools, while students in care ‘were six times more likely’ (Marszalek, 2022, np). These rates are higher than those identified in South Australian government schools (Graham et al., 2020) and are deeply concerning given the QLD government’s stated commitment to systemic inclusive education reform since 2018. Importantly, this commitment was sparked by a series of improper disciplinary responses in QLD state schools almost a decade ago, which led to a review of education for students with disability in that state and the subsequent release of a new inclusive education policy and reform strategy by the QLD Department of Education. Understanding this background is critical for identifying gaps in the policy and reform apparatus that may be failing students in priority equity groups.

Background to the QLD 2018 inclusive education policy

In October 2015, the story of a 9-year-old boy on the autism spectrum being subjected to time-out in a ‘prison-like cell’ 20 times in one year hit the national print and television media (Vonow, 2015). In response, the then QLD Education Minister, The Hon. Kate Jones, immediately ordered the school to stop using the room and launched an investigation. Before that investigation had concluded, the Minister announced establishment of a new $2 million Autism and Reading Hub to provide ‘evidence-based support and advice to help state and non-state schools engage and retain students with autism so they are academically and socially successful’ (Queensland Government, 2016, np). Four months later, the Minister announced that an independent review into education policy for students with disability would be conducted. In September 2016, Deloitte Access Economics was announced as the successful tenderer, and their final report was released in March 2017. The 193-page report spoke to ‘the need to adopt a unique approach to school performance for students with disability’ (p. 36) and advocated for the ‘establishment of measurable outcomes at the school and system level … similar to what is currently undertaken for reporting on outcomes for Indigenous students’ (p. 68).

One indicator explicitly recommended for disaggregated reporting in the Deloitte Disability Review was SDAs. The Review report (p. 104) noted that

• students with disability in Queensland state schools are more likely to be subject to SDAs than students without disability,

• the average use of SDA among students with disability in Queensland has been consistently growing since 2011 and

• SDA rates may have been exacerbated by the change to legislation which expanded principals’ powers to suspend and took effect in 2014.

However, the Review team did not undertake any further analysis to understand which students with disability were being suspended or excluded and why, nor did the final report offer much in the way of methods to reduce the use of SDAs, other than for the department to continue implementing Positive Behaviour for Learning (PBL). This suggestion was not an explicit recommendation and was made despite acknowledgment in the report that there had been no evaluation of PBL impact or implementation fidelity in the 460 Queensland state schools that had received training in PBL by the time of the review. Nor was the appropriateness of PBL for Indigenous students or those with disability considered.

The Review team also only examined SDAs of students with a verified disability, a classification that covers just those students with disability receiving substantial individual support through the recently superseded Education Adjustment Program (EAP) and not those students with disability ineligible for EAP support. Therefore, while students with a disability were described as disproportionately impacted by the use of SDAs in the Deloitte report, the true magnitude of their overrepresentation was masked, for there are many students with a disability—like, for example, those with Attention-Deficit Hyperactivity Disorder (ADHD)—who experience significant barriers to learning but who were ineligible for EAP and who often do not receive the reasonable adjustments to which they are entitled under the Disability Standards for Education (Graham & Tancredi, 2019). Their difficulties in school and with learning also tend to be interpreted through the prism of disruptive and non-compliant behaviour and responded to accordingly (Graham & Tancredi, 2019).

While the connection between learning and behaviour was noted in the Deloitte report and the development of a needs-based (as opposed to categorical) funding mechanism was recommended, the Review team failed to examine known intersectionalities between disability, Indigeneity and OOHC, and nor did the team examine associations with exclusionary discipline, with the result that students with disability were treated as a homogenous group when they are not. Nor were SDAs disaggregated by disability category, which gives the impression that all students with disability are overrepresented, when only some may be, and potentially grossly so. Patterns like this are important because they give insight into what is not working for who, which is knowledge essential for effective inclusive school reform (Graham et al., 2023). The short reporting timeframe required, as well as access to institutional data and the expertise necessary to make meaningful sense of it may have limited the Review team. These limitations may also be why there were only 17 (predominantly system performance) recommendations that did not speak to the legislative changes, support structures, or specific professional learning necessary to enable the required reform of practice both in and outside classrooms. While governments understandably wish to address problems quickly, shallow investigative depth leads to a lack in the critical detail needed to develop effective policy and response frameworks.

All 17 recommendations (see Table 4, Appendix) made in the Deloitte Review final report were accepted by the Queensland Government for implementation by the department. One of the department’s first actions was to appoint an Assistant Director-General to lead a dedicated ‘Disability and Inclusion’ branch within the department to support reform implementation and to ensure a senior voice at the strategy and policy table for the inclusion of students with disability. However, due to a restructure in 2022, this position no longer exists. Another early piece of work, in response to Recommendation 4–2, involved the development of an inclusive education policy statement and implementation strategy.

Queensland’s inclusive education policy

The Queensland Government’s inclusive education policy statement was developed in consultation with key stakeholders and was the first in Australia to be informed by General Comment No. 4 (GC4) (United Nations, 2016) on Article 24 of the Convention on the Rights of Persons with Disabilities (CRPD) (United Nations, 2008). At 24-pages, GC4 is ‘the most comprehensive and authoritative international instrument explaining the human right to inclusive education’ (Graham et al., 2020, p. 2). It ‘explicitly defines inclusive education, as well as models of provision that are incompatible with it, and outlines the processes and practices necessary to ensure [its] realisation’ (p. 2). The QLD Inclusive Education Statement draws on GC4 to explicitly define inclusive education, distinguishing it from exclusion, segregation and integration. It similarly refers to the nine core features of inclusive education outlined in GC4, and unlike the Victorian government’s Inclusive Education Policy of the time, the QLD policy is broad and not restricted to students with disability, which was wise given the tendency for all things disability-related to be siloed.

However, rather than speaking of all students, the Statement named nine groups to be ‘included’: Aboriginal and Torres Strait Islander students, culturally and linguistically diverse students, gifted and talented students, students identifying as LGBTQI+, refugee students, rural and remote students, students with disability, students in out-of-home care and students with mental health needs. By ‘pointing up a difference’ (Graham & Slee, 2008), the Statement also normalises groups that are not named (CIS-gendered, heterosexual, non-disabled, non-Indigenous, Australian born, native English-speaking, urban students who live with their biological parents and who do not have mental health needs or gifts and talents) and inadvertently marginalises those that are named. It also portrays these groups as distinct, denying the numerous intersectionalities between them. The subsequent development of separate strategies—as has occurred in QLD with the ‘Every student with disability succeeding’ strategy, and the ‘Every Aboriginal and Torres Strait Islander student succeeding’ strategy—carries with it the danger that students in these groups will be viewed through the one lens. Simply speaking, an Aboriginal student with hearing impairment will need reasonable adjustments to access and participate in classroom activities, but their need for adjustments could be missed if cultural responsiveness is the only focus for Indigenous students and not the possibility of disability.

The department’s final Inclusive Education Policy was released in March 2018 and later received international recognition through an award from The Zero Project, an initiative of the Essl Foundation aimed at recognising and supporting implementation of the CRPD (Zero Project, 2020). The success of the policy was to be evaluated using four outcome measures (Table 1), which were expanded to eight in a review of the policy in 2021.

Table 1 Outcome measures for Every Student with Disability Succeeding Strategy, comparing 2018–2021 and 2021–2025

A commitment to reducing SDAs for students with disability was evident in both iterations of the policy; however, there were subtle changes over time. For example, the 2018–2021 strategy committed to decreasing the proportion of students with disability receiving an SDA and the 2021–2025 strategy committed to decreasing the number of students receiving SDAs, without specifying disability. Number is vulnerable to enrolment growth which may be why the Department made the switch; however, in neither iteration is there a commitment to reduce the number of incidents. The problem is that this target could be met simply by reducing the number of students receiving a suspension, but not reducing the overall number of suspensions, which could also mean some students receive many more (repeat) suspensions, yet the target will ostensibly still be achieved and can be reported as such. A progress report for the ‘Every Student with Disability Succeeding Strategy’ was published in 2021; however, it has since been removed from the Department’s website. At that time, the only improvement was in A-E reporting.

Comparing outcome measures

Recall that the Deloitte Review recommendations were that measurable outcomes at the school and system level be developed for students with a disability like those for Indigenous students. Despite this recommendation and the Department’s subsequent development of outcome measures for disability, there is a clear difference between the two sets of measures in terms of their measurability.

The ‘Every Aboriginal and Torres Strait Islander Student Succeeding’ strategy was updated in 2019 (Department of Education Queensland, 2019). It also had four outcome measures; however, rather than vague statements like ‘improving the A-E performance’—which could be achieved with very minor and relatively meaningless improvements, say from E to D, and is highly vulnerable to ‘soft’ marking—the targets for Aboriginal and Torres Strait Islander students are more explicit. They are to

• Increase the proportion of students achieving C or above in English to 80% in 2022.

• Improve the proportion of educators who strongly agree they are confident in embedding cultural perspectives in learning to 30% by 2022.

• Increase the student retention rate from Years 10 to 12 to 90% by 2022.

• Reduce the number of students receiving multiple school disciplinary absences by 2022.

Although these outcome measures are more explicit, the wording around SDAs is again interesting. In this case, the government’s commitment is to reduce the number of Aboriginal and Torres Strait Islander students receiving multiple SDAs. But, in neither the case of students with disability nor Aboriginal and Torres Strait Islander students, is there a commitment to reduce the total number of incidents. It is, therefore, entirely possible to meet this target by reducing the number of Aboriginal and Torres Strait Islander students receiving multiple SDAs (e.g. from 1000 to 950 students) but to not reduce the total number of incidents, nor the number of multiple SDAs overall. In fact, the 950 Indigenous students in the example above who are receiving multiple SDAs could receive even more SDAs than they did in previous years, increasing SDAs overall, but this would not violate the target for it will still be 50 fewer students receiving multiple SDAs. Further, while there is a commitment to increasing the proportion of educators who strongly agree they are confident in embedding cultural perspectives, there is no such reference to inclusive or trauma-informed practice, despite higher proportions of Indigenous students with a disability and/or living in out-of-home care than non-Indigenous students. Further, there is no explicit strategy, outcome measures or progress reports for students in OOHC, despite known intersectionalities between disability, Indigeneity and being in-care.

As the Queensland government publishes only limited SDA data, which includes only the number of incidents (and not the number and type of students associated with those incidents), it is currently impossible for the government to be held to account for the success of either strategy or for inequitable disciplinary outcomes for students in priority equity groups. This is a critical failure of the Deloitte Review which recommended only that the department disaggregate data for students with disability but not that those data be made publicly available or even that the patterns in and associations in the data should inform the development of reduction targets to be achieved by thorough legislative, policy and practice reform. While the case for such reform has recently been made in research detailing increases in the suspension and exclusion of Indigenous students (Graham et al., 2020), that research was again using publicly available data and could not disentangle groups beyond Indigenous and non-Indigenous. While necessary to identify differential experiences, such analyses may inadvertently reinforce deficit views of Indigenous children and young people and/or prompt fast-tracking an increase in the proportion of teachers confident in embedding Indigenous perspectives. This, however, will not help an Indigenous student with a disability and, if disability is the ‘common denominator’ in QLD, as it was in South Australia (Graham et al., 2020), the response needs to take disability into account. Similarly, if disability is entangled with being in OOHC, then this too must necessarily feed into the response. In this paper, we build on the approach taken in the SA Inquiry to examine whether Indigenous students, students with disability and students in OOHC are overrepresented in suspensions, compared to all students. We then apply the eight groupings employed in that study to disentangle the intersectionalities occurring between these three groups. This approach can better indicate factors that may be at play, and therefore, where education reform efforts should concentrate.

Method

Data source

A Right to Information (RtI) request was submitted by Queensland Advocacy for Inclusion (QAI) in December 2021 to obtain enrolment and student disciplinary absence (SDA) data from the Queensland Department of Education (DoE). This project was granted an ethics exemption by Queensland University of Technology (QUT) Human Research Ethics Committee (application no. 6376), given the use of non-identifiable secondary data, in accordance with the National Statement on Ethical Conduct in Human Research (2007, updated 2018).

The RtI request included all four categories of SDA (short suspensions, long suspensions, exclusions and enrolment cancelations), for calendar years 2012–2020. In this paper, we present analyses on short and long suspensions data. Two forms of data were requested for each category of SDA: student-level data (representing how many students were issued an SDA during the calendar year) and incident-level data (total SDAs issued during the calendar year). These data were obtained for all students and further disaggregated according to whether students were (i) Indigenous, (ii) recorded as having a disability for the purposes of the Nationally Consistent Collection of Data on School Students with a Disability (NCCD) (provided from 2016 onwards), (iii) living in out-of-home care (OOHC) or (iv) any combination of these groups. However, as these three priority equity groups are not independent, interpreting these rates without accounting for intersectionality can lead to an inaccurate view of the underlying factors driving these high rates. Moreover, as each group is also subsumed under the category of ‘all students’, higher rates for priority equity groups artificially inflate the overall student rates for this group, rendering any comparison inappropriate. Instead, a more meaningful reference point for the rates of priority equity groups would consist of students who are non-Indigenous, do not have a disability and do not live in OOHC.

To quantify the unique contribution of risk or overrepresentation associated with each factor (Indigeneity, disability, OOHC), the following independent groups were determined for all enrolment and exclusionary discipline data provided, from 2016 onwards (to allow inclusion of NCCD data):

• 1. Indigenous only

• 2. NCCD only

• 3. OOHC only

• 4. NCCD and Indigenous only

• 5. OOHC and Indigenous only

• 6. OOHC and NCCD only

• 7. Indigenous, NCCD and OOHC

• 8. Not in any of the above priority equity groups

While the final group excludes students who are Indigenous, have an identified disability, or live in out-of-home care, it is worth noting there may be additional underlying factors which place some students in this group at risk of suspension, such as socioeconomic status, gender, and/or unidentified disability; however, such an analysis is outside the scope of the present study given the available data. Finally, we drew on NCCD adjustment categories as a way of gauging which students with disability may be at more risk of suspension. There are four adjustment categories in NCCD: Physical, Sensory, Cognitive and Social-Emotional. Inferences can be drawn from these categories in that an adjustment in the physical category is likely to have been made to provide access to a student with a mobility impairment, whereas an adjustment in the Social-Emotional category is more likely to be made for a neurodivergent student.

Total enrolments and a breakdown of the enrolments by group is provided in Table 2. We also include the state school enrolments as a proportion of all Queensland students (government and non-government), which hovered around 67% (67.2–66.9%) during the years under investigation (ABS, 2023).

Table 2 Enrolments in each priority equity group

Analytic plan

In each of the analyses described below, suspension data were disaggregated by calendar year. Short and long suspension incidents were combined wherever possible, given the absence of any specific research hypotheses regarding differences in short and long suspensions, and due to the lower instances of long suspension incidents, necessitating caution when it comes to interpretation of patterns. However, combining short and long incidents is not tenable for analyses in which single incidents are compared to repeats or where students must not be represented more than once within an analysis (i.e. risk ratios, Raw Differential Representation). In those instances, we report short and long suspensions separately or short suspensions only.

Overall trends

Firstly, to contextualise subsequent analyses, overall trends in suspensions over time were presented for the three original priority equity groups and all students combined. This descriptive analysis employed the rate of suspensions (short and long incidents) per 1000 students, in each group, from 2012 to 2020.

Overrepresentation

Next, to examine overrepresentation, the distribution of enrolments and SDAs by priority equity group was examined using incident-level SDA (short and long incidents). For both enrolments and SDAs, the percentage of incidents issued to each of the above eight groups was determined. While an exact cutoff to indicate overrepresentation does not exist, these percentages were contextualised using the ± 10% criterion of disproportionality employed by Chinn and Hughes (1987) in their work on overrepresentation in enrolment data.

Risk ratios

Risk ratios were calculated to compare the risk of suspension for students in each priority equity group, or group combination, to the group of students who did not fall into any priority equity group, as articulated in the below formula, where PEG represents the Priority Equity Group of interest:

$$\frac{(No.\, students\, in \,PEG \,suspended\div No. \,students \,in \,PEG \,enrolled)}{(No.\, students\, not\, in \,any \,PEG \,suspended \div No.\, students \,not\, in\, any \,PEG \,enrolled)}$$

A ratio of 1 indicates that the risk of suspension is equivalent for both groups. Where the ratio exceeds 1, the risk is greater for the priority equity group; where it is less than 1, the risk is greater for students not in a priority equity group. In this instance, calculation was based on student- rather than incident-level data, for short and long suspensions separately, and for each calendar year separately. Analyses were conducted using the R package epitools (Aragon et al., 2022). For short suspensions, unconditional maximum likelihood estimation (Wald) was used, and for long suspensions a small sample adjustment was applied, as expected cell sizes were below five in all calendar years. Confidence intervals (95%) were calculated to quantify variability around each risk ratio. Where these intervals do not encompass the null value of 1, a statistically significant difference between the priority equity group and the reference group may be inferred. To account for the number of comparisons (seven groups compared to a reference, within each calendar year), we also generated p values from tests associated with the risk ratios (chi-square tests of independence for short suspensions; Fisher’s exact tests for long suspensions, due to small cell sizes) and applied a Bonferroni correction (α = 0.05/7 = 0.007) to comparisons made in each calendar year. Risk ratios and the associated confidence intervals are depicted graphically using a logarithmic scale to aid interpretability (Levine et al., 2010).

Raw differential representation (RDR)

RDR for each group in each calendar year was calculated, using the number of students issued a short suspension within each calendar year. This metric reflects the number of additional students issued with a short suspension within each priority equity group, who would not have been suspended if the rates were commensurate with the suspension rates of students not in a priority group (Girvan et al., 2019). RDR is calculated by firstly determining the difference in risk of suspension for students in a particular priority equity group and students not in any of the groups and then multiplying this risk difference by the total number of students in the priority equity group. The formula for this metric is presented below:

$$RDR= Number\, of\, students \,in\, PEG \,enrolled \times Risk\, Difference$$

where Risk Difference is calculated by:

$$\frac{No. students in PEG suspended}{No. students in PEG enrolled}- \frac{No. students NOT in PEG suspended}{No. students NOT in PEG enrolled}$$

Disciplinary rate

Overall disciplinary rate per 100 students per school day was calculated per the formula below, for short and long suspensions combined, for each of the priority equity groups. This metric provided a means of equalising discipline rate according to the number of school days (Girvan et al., 2019), which is especially important due to the impacts of the COVID-19 pandemic in 2020 and the number of days students were able to attend school.

$$\frac{Number \,of\,incidents}{\left[\left(Number \,of\,enrolments \div 100\right) \times Number \,of \,school \,days\right]}$$

Single and repeat suspensions

Comparing student-level to repeat incident data enables an examination of the extent to which certain groups are issued multiple SDAs. Hence, for each group, the number of students issued with an SDA and the number of repeat incidents are represented separately in terms of the rate per 1000 students, for each SDA type, using the formula below:

$$\frac{Number\, of\,incidents}{Number \,of \,enrolments} \times 1000$$

For the year 2019 only, SDA and enrolment data were obtained for students identified with a disability according to NCCD, disaggregated by to their category of adjustment (Physical, Social-Emotional, Cognitive, Sensory). Using student-level and incident-level data, single and repeat rates of short suspension incidents per 1000 students were calculated for each category.

Results

Overall trends over time for all students and priority equity groups

Suspension incidents proportionate to enrolments are displayed in Fig. 1 for calendar years 2012 through to 2020. Separate lines depict rates for all enrolled students, Indigenous students, students living in OOHC and students with a disability (from the commencement of NCCD data collection in 2016). In all calendar years, rates are highest for students living in OOHC, followed by those who are Indigenous or have a disability. While rates were marginally lower in 2020, likely due to impacts of the COVID-19 pandemic, they appear to have increased over time for students in OOHC and, to a lesser degree, for Indigenous students and students with a disability.

Fig. 1

Suspensions (short and long combined) per 1000 students according to priority equity group

Distribution of enrolments and suspensions

Figure 2 displays the total number of students enrolled from 2016 to 2020 and the distribution of these students within unique and intersecting priority equity groups. When considering the combined percentage of priority equity groups, approximately a quarter of enrolments are accounted for by students who are Indigenous, have a disability, live in out-of-home care, or some combination thereof. The majority of students enrolled do not fall into any of these categories.

Fig. 2

Percentage of enrolled students and suspension incidents (combined short and long) according to priority equity group

The distribution of suspension incidents (short and long combined) across priority equity groups are also displayed for the years 2016 to 2020. The percentage of suspensions issued to students who are not in a priority equity group averaged 39% (ranging 38.15–40.45% throughout the five calendar years), although this group makes up about 73% of enrolments. In contrast, students who fall into one or more of the three priority equity groups received approximately 61% of total suspensions (ranging 59.55–61.85%), despite collectively representing just over a quarter of enrolments (27%). This percentage, 61%, well exceeds the range that would be expected based on enrolment data (24.3–29.7%) using the ± 10% criterion of disproportionality (Chinn & Hughes, 1987).

To examine these trends in more detail, we examined the distribution of suspensions for the calendar year 2020 using a method employed in the SA Inquiry which helps to highlight commonalities between groups, as well as illustrate the dominance of particular characteristics. Figure 3 lays bare the ‘common denominator’ status of disability. Just over one third of all students suspended in 2020 had a disability but were not Indigenous and not living in OOHC. A further 15.5% of suspensions went to students in two or more priority equity groups and almost all of those (97.16%) had a disability, with the lowest proportion of suspensions in this group (2.84%) going to those without disability. As in the SA Inquiry, most affected among students in two or more priority equity groups are Indigenous students with a disability.

Fig. 3

Percentage of enrolled students and suspension incidents (combined short and long) in 2020 according to priority equity group

Risk ratios

Risk ratios were calculated to compare, separately for each priority equity group or combination, the risk of being issued a suspension among students who are in the group with the risk among students who are not in any of the groups (the reference group). To ensure students were not represented more than once within these calculations, single suspensions data were used for these calculations, and short and long suspensions were analysed separately. As shown in Fig. 4, the risk of short suspension was greater for every priority equity group in comparison to the reference group, and these risks increased in line with greater intersectionality among groups.

Fig. 4

Risk of short suspension for students in each priority equity group, as compared with students who are not in any of the priority equity groups

Fig. 5

Risk of long suspension for students in each priority equity group, as compared with students who are not in any of the priority equity groups

For example, in 2019, Indigenous students had 2.65 times the risk of being issued a short suspension, Indigenous students with a disability had 5.02 times the risk and Indigenous students with a disability living in out-of-home care had 7.79 times the risk. Confidence intervals did not include the reference group (RR = 1), and p values were below the adjusted cutoff of α = 0.007 (all p’s < 0.0001), indicating a significantly increased risk for students in priority equity groups, within each calendar year.

Similarly, in that same calendar year, Indigenous students had 3.13 times the risk of being issued a long suspension, Indigenous students with a disability had 6.13 times the risk and Indigenous students with a disability living in out-of-home care had 12.34 times the risk (Fig. 5). With very few exceptions (NCCD only, 2019; OOHC and Indigenous, 2020), the risk ratios for long suspension were higher than those of short suspension, particularly for the groups in which multiple equity factors are combined. Again, confidence intervals did not include the reference group for any group in any year, nor did p values exceed the adjusted cutoff, indicating a significantly increased risk for students in each priority equity group (all p’s < 0.0001). The more extreme point estimates and wider confidence intervals for long suspensions, particularly for Indigenous students in out-of-home care, are reflective of the fact that the number of long suspensions issued within these groups was relatively small (Girvan et al., 2019).

From visual inspection, there appear to be slight increasing trends over time in the risk of short suspension for the NCCD and Indigenous group, the OOHC and NCCD group, and the final group representing all three priority equity areas. For long suspensions, it is notable that the risk increases in 2020 for the latter two groups, considering that 2020 marked the first year of the COVID-19 pandemic, in which 31 school days were lost due to lockdowns or learning from home. However, with only 5 years of data, and given the relative instability of risk-based metrics over time (Girvan et al., 2019), longitudinal trends should be interpreted with caution. In contrast to risk ratios, metrics such as Raw Differential Representation (RDR) and discipline rate provide a more stable measure of the scale and impacts of disproportionality over time (Girvan et al., 2019). RDR represents the number of additional students issued with a suspension within each priority equity group who would not have been suspended if the suspension rate was consistent with that of students not in a priority group, therefore, providing an indication of the impact of disproportionate suspension use. The discipline rate represents suspension incidents per 100 students enrolled per school day, providing a measure of the prevalence of suspension use in each group. The results of analyses using these metrics are presented in the following section to more fully investigate the available data.

Raw differential representation: number of students

Table 3 displays the RDR of students in each priority equity group, calculated using short suspension data; in other words, the number of additional students in each group who were issued a short suspension as a function of the higher suspension rate for that group, as compared with students not in any priority group. For example, in 2016, the percentage of Indigenous only students who were suspended was 11.6%, while the percentage of students not in any priority equity group was 4.3% in that year. The resulting difference in rates of 7.3% equates to 2527 Indigenous students, as determined by multiplying the rate by enrolled students. Put simply, if Indigenous students without a disability and not living in OOHC were suspended at the same rate as students not in a priority equity group, 2527 fewer Indigenous students would have been suspended during that calendar year. These numbers are particularly alarming for students with a disability, with numbers in excess of 7000 in each year.

Table 3 raw differential representation of students in each priority equity group

Discipline rate

Discipline rate takes into account the number of school days in each year, thereby providing a way to account for the reduced number of school days in 2020. Hence, in this analysis, the discipline rate represents short and long suspension incidents per 100 students per day (see Fig. 6). Different student subgroups are indicated by colour of the dots, and, as an indicator of the impacts at the student level, RDR is represented again by the size of the dots. The highest discipline rates in every calendar year are observed for Indigenous students with a disability living in OOHC; yet the highest RDRs are observed for students with a disability only.

Fig. 6

Discipline rate per 100 students per day and Raw Differential Representation of students in each priority equity group

These data provide insight into the scale and impact of suspensions over time. Another important factor to consider is the extent to which these incidents reflect one-time events, or repeated suspensions, which will be unpacked for each group in the next section.

Single and repeat suspensions per 1000 students

Short and long suspensions are represented separately in Figs. 7 and 8 below. Each graph depicts student-level and repeat incidents, separately for different priority equity groups. For clarity of interpretation, this graph includes five (instead of eight) groups: students not in any priority equity group, students in one priority equity group only (Indigenous, in care, or disability), and students who fall into all three groups. For short suspensions, the highest rates per 1000 students, across all years, were the single and repeat incidents for the group in which all three priority equity factors were present—Indigenous students with a disability living in out-of-home care. Moreover, while the rate of students being suspended does not appear to increase over time, the rate of repeat incidents increases steadily each year, from 480.93 repeat suspensions per 1000 students in 2016 to 592 per 1000 students in 2020. All other groups have markedly higher single and repeat incident rates in comparison to the group that does not fall into any of the priority equity areas. For this reference group, the repeat incident rate is lower than the single incident rate.

Fig. 7

Distribution by priority equity group of Single vs Repeat Short Suspensions from 2016 to 2020

Fig. 8

Distribution by priority equity group of Single vs Repeat Long suspensions from 2016 to 2020

Long suspension rates for these groups were much lower in comparison to short suspension rates (see Fig. 8). Repeat rates are lower than single rates for each group. Again, the highest rate of students suspended is for students who are Indigenous, have a disability and live in out-of-home care.

To determine whether rates of single and repeat suspensions varied according to adjustment category, for students with a disability (NCCD), further investigation was conducted for students in each category, during the year 2019. Rates per 1000 students were calculated separately for single and repeat short suspensions in each category. As depicted in Fig. 9, repeat rates tended to be highest in the social-emotional adjustments category and repeat rates exceeded single rates for every category except Physical, which had the lowest rates in general.

Fig. 9

Single and repeat rates of short suspensions for students with a disability (NCCD) based on their adjustment category, in 2019

Discussion

Overrepresentation of students in priority equity groups in school suspension and exclusion is a critical and ongoing problem within Queensland state schools; however, there is limited publicly available data to determine the extent of the problem and which students it impacts most, which is an essential first step towards meeting unmet needs. Using data obtained through a Right to Information request, this paper makes an important but long overdue contribution by creating eight distinct groups‒comprised by seven equity groups and one referent group‒and then examining suspension rates over the 5 years from 2016 to 2020. Findings indicate that suspensions are issued to students in those seven priority equity groups at a rate grossly disproportionate to the number of students enrolled, a pattern which is not observed for the referent group: students who are not in any priority equity groups. Risk ratios indicated an increased risk of suspension, both short and long, for students in priority equity groups; moreover, these risks increased with increasing intersectionality, such that students who fell into all three priority equity groups (Indigenous + disability + in-care) were at the greatest risk of suspension.

Findings also showed that, in each of the years 2016–2020, the highest discipline rates proportionate to school days and enrolments were observed for Indigenous students with a disability living in OOHC, while the highest number of additional students suspended due to disproportionate rates (RDR) were observed for students with a disability only. These metrics provide different yet complementary information about the nature and impacts of disproportionate suspensions (Girvan et al., 2019). Higher discipline rates capture disproportionality in the overall number of suspension incidents issued, scaled to enrolment and days at school, therefore, providing insight into the prevalence of suspensions for different groups. RDR characterises the broader impacts of disproportionate suspension rates, by indicating the number of students in priority equity groups issued a suspension who would not have been suspended but for the identified disparities in suspension rates between groups. While providing valuable information as to the scale of disproportionate suspension, this metric does not allow interpretation of the potential underlying causes, as it is influenced by multiple factors within the data, such as the number of students enrolled in the groups being compared, disproportionate discipline rates among these groups, and overall discipline rates (Girvan et al., 2019). Hence, the present findings indicate that while prevalence and risk of suspension is highest for Indigenous students with a disability who are living in care, the greatest number of students impacted by the use of exclusionary discipline are those with disability.

When disaggregating by single and repeat suspension incidents, the highest rates per 1000 students, across all years, were observed for students falling into all three priority equity groups: Indigenous + disability + in-care. For this group, repeat suspensions appeared to increase over time, even in 2020 when schools were in lockdown and the total number of incidents decreased as a result. For students with a disability, rates per 1000 students were highest for those represented within the Social-Emotional category of adjustments, and again, the rates of repeated suspensions exceeded those of one-time events. These findings are in direct contrast to the referent group (students not in any priority equity groups), for whom repeat suspensions were fewer than single suspensions. While these findings reflect aggregate rates for each group, they nonetheless show that high suspension rates for students in priority equity groups are being driven by repeat events, rather than single events, providing yet more empirical evidence to refute the claim that suspension acts as a deterrent to future behaviour.

As reported in our previous publications on increases in the use of exclusionary discipline in Australian schools, the international research evidence consistently shows that when the underlying reasons prompting the suspension are not effectively addressed, a vicious cycle of repeat suspensions can occur, with huge costs to students’ exposure to academic instruction (Wilkerson et al., 2022) and the quality of student–teacher relationships (Okonofua et al., 2016), with the tendency to reinforce the behaviours for which it was issued (Wiley et al., 2020). Comprehensive reform that encompasses legislation, policy/strategy, procedures, and—most importantly—classroom practice and the provision of support services is necessary to reduce the incidence of problem behaviour and to mitigate the use of exclusionary discipline.

Implications for reform

Our research adds to an emerging body of work in Australia investigating the overrepresentation of students in priority equity groups in the use of exclusionary discipline and referral to segregated settings (de Plevitz, 2006; Graham, 2012; Graham et al., 2020; Grahm et al., 2023; Rudolph & Thomas, 2023; Sweller et al., 2012). Identifying the existence and nature of overrepresentation is an important first step but the most urgent task now is to identify and address factors driving disproportionality. This is not something that should be dragged out and nor should it happen behind closed doors; the stakes are too high for the children and young people being affected now, as well as all those who may follow in their wake if solutions are derived from inadequate analysis. Therefore, while the QLD government has recently responded to increased pressure from the media, advocacy campaigns and research evidence pointing to unparalleled and inequitable growth in the use of exclusionary discipline, as well as repeated warnings about to its long-term negative impacts on children, families and communities, this work is taking place internally and gains claimed are not being substantiated with data. Further, while there are departmental reports of a decrease in suspensions there has been no acknowledgement of the role of COVID on attendance with its consequential impacts on suspension (Kovacevic, 2022). The government’s focus has also been fixed on Preparatory year children; however, our research demonstrates that the overuse of exclusionary discipline in QLD state schools affects many more children than just those in prep. Note also that those students now being suspended in Grades 1 to 6 are children who entered school after the 2014 legislative change that expanded principals’ power to suspend, lengthened the maximum short suspension to 10 days, and stripped parents’ right to appeal (Carden, 2018). Given the complexity of their learning profiles, older students among those we have identified as most at risk of suspension were also quite possibly among the increasing numbers of prep and early years children suspended from 2014 (Graham 2020). An Inquiry of the depth and rigor of that conducted in South Australia would be able to determine this. Such children should be receiving intensive individualised supports and substantial/extensive adjustments (Swancutt, 2023), not punishment and exile, and we believe an Inquiry of that nature is necessary for effective systemic inclusive school reform.

We say this because the current approach of soft accountability through real-time data access, departmental line of sight and conversations with principals through regional directors is vulnerable to switches in Ministerial attention and any gains made via this approach will wither as soon as that switch occurs. As politically unpalatable as it may be, given the support for suspension from industrial associations, the only lasting fix will come from legislative change to reintroduce the safeguards that were stripped away by the previous government (Carden, 2018). However, this legislative change should also include thresholds like those proposed in the SA Inquiry to trigger escalated review of supports and adjustments once those thresholds are reached (Graham et al., 2020). Stricter suspension criteria that reduce the number of permissible reasons for suspension including bans for minor incidents, as well as a significant reduction in permissible length (e.g. 3–5 days maximum) in line with international best practice is also critical. Specialist advocacy support and the right to appeal with that support is another necessary safeguard to protect the rights of marginalised students, especially those in priority equity groups.

To be effective, legislative reform needs to be supported by tight accountability in the enactment of policy and procedure to make sure relevant safeguards are heeded throughout the system, including on the ground in schools. And schools need to be supported with funding and guidance to implement evidence-based preventions and supports, including in-school suspension to build social-emotional skills. For example, our research shows that the students with a disability receiving the most suspensions are those receiving adjustments in the Social-Emotional category. As found in the SA Inquiry, these students are also receiving more multiple suspensions than they are single, suggesting that the adjustments they have been recorded as receiving are inadequate, irrelevant or, quite possibly, not being enacted at all. These possibilities underscore the importance of proactive universal social-emotional learning that covers all five CASEL competencies, especially Responsible Decision Making which is not covered in the Australian Curriculum Personal and Social Capability (Laurens et al., 2022). Retrospective adjustments are unlikely to be successful when a student is already at the point of overwhelm and this is what we suspect might be the story behind these single/repeat suspension statistics.

Inclusive education reform must also be supported with high-quality professional learning (PL) relevant to the groups that are most at risk of suspension: (i) inclusive practice, (ii) culturally appropriate practice and (iii) trauma-informed practice. PL should focus on Tier 1 (universal) provision for classroom teachers, extending to Tiers 2 and 3 (targeted and intensive) for specialist support staff. Enhancements in inclusive, culturally appropriate and trauma-informed practice need to be combined with the provision of evidence-based social-emotional learning, and all four delivered through a Multi-Tiered Systems of Support (MTSS) model which promotes the use of universal screening and progress measures to support data-based decision-making (de Bruin et al., 2023). Most importantly, teachers need to be upskilled to engage in accessible quality first teaching (Graham & Tancredi, 2023), for many of the students who end up in suspension statistics either have a known disability but are receiving inadequate support and adjustments contributing to an experience of overwhelm, as is often the case for neurodiverse students, or they have an unidentified disability and are receiving nothing but disciplinary consequences for a perceived failure to comply with instructions they not may understand, as in the case of those with a language disorder (Graham & Tancredi, 2020).

Lastly, and despite our critique of the outcome measures, the QLD Department of Education is to be commended for publishing these measures and for striving to achieve them. This is better than many other systems and sectors in Australia, especially Victoria which only publishes expulsion data. However, the outcome measures for QLD should be bold and commit to reducing suspension incidents overall, as well as for specific groups; they should be explicit and measurable, annual progress against them should be publicly reported using sophisticated techniques capable of detecting the influence of differences in attendance across years, and they should avoid creating silos where there are none. For example, our analyses show that the largest group of students in more than two priority equity groups who have been suspended are Indigenous students with a disability, yet there is no recognition of disability in the four outcome measures for the Every Aboriginal and Torres Strait Islander Student Succeeding strategy. While it might be presumed that Indigenous students with a disability will be identified and served through Every Student with Disability Succeeding, this is unwise. There is long-standing evidence from across the world that white middle class school staff misinterpret the needs and behaviours of children from other cultures (Chin et al., 2020; La Salle et al., 2020; Raffaele Mendez, 2003; Skiba et al., 2002), particularly if those children happen to be black or brown (Rudolph & Thomas, 2023). There is also a tendency in Australia also to see every issue affecting Indigenous students through the lens of cultural safety and relevance (Keddie et al., 2013). Culturally appropriate practice is critical, of course, but it will not help that child to access and participate in education if they have a hearing impairment, language disorder, ADHD, autism, or Foetal Alcohol Spectrum Disorder. The same applies for students in out-of-home care only being seen through the lens of trauma-informed practice. When disability is in the frame—whether formally identified or not—the focus needs to shift to the quality and accessibility of inclusive practice (Graham & Tancredi, 2023). Outcome measures for all three priority groups should, therefore, be developed to reflect the intersectionalities between them and to ensure that all roads lead back to understanding what each individual student needs to be safe and successful at school.

Limitations and future directions for research

The present study is subject to limitations imposed by the available data and, as a consequence, by the metrics used for analysis. Aggregate-level frequency data is useful for gaining understanding of exclusionary discipline data broadly, but there are limitations in terms of the types of analyses that can be performed and the implications that can be derived. Although risk ratios provide highly useful information about disproportionality within subgroups, they are not always stable over time (Girvan et al., 2019), making it difficult to use them as a reliable indicator of increasing or decreasing disproportionality. Given this limitation, we also included RDR and discipline rates, which are two recommended alternatives for assessing longitudinal trends and providing an indication of the scale and impact of exclusionary discipline over time (Girvan et al., 2019). This approach to analysis aligns with current approaches to the assessment of disproportionality, where triangulation of different metrics is recommended (e.g. Bottiani et al., 2023; Curran, 2020; Girvan et al., 2019).

The ability to disaggregate data and investigate non-overlapping priority equity groups has provided insight into how suspension incidence changes based on these group characteristics. However, there are other variables, not available to us in this current dataset, such as school attendance, SES, gender and region, with known associations with suspension rates (Graham et al., 2023; Laurens et al., 2021), for which we were not able to account. Moreover, in the present analysis, the reference group against which comparisons were conducted excludes students in the three priority equity groups under consideration; yet there may be other factors potentially placing these students at risk, for which we lack data.

Another important consideration is the disaggregation of single and repeat suspension data. The present analyses underscore the insidious nature of suspension, particularly for certain groups of students, for which the rate of repeated incidents exceeds the rate of one-time events. However, although highlighting systemic issues at play, these data do not provide an indication of the distribution of repeat suspensions across students. It may be that these high repeat rates are driven by a small number of students receiving a high number of suspensions throughout the year. Without transparent, publicly available data for frequency of suspension incidents, it becomes impossible to evaluate whether target outcomes of departmental strategies have been met; for instance, these data cannot speak to whether the department’s Every Aboriginal and Torres Strait Islander Student Succeeding Strategy has been successful in its aim to ‘reduce the number of students receiving multiple school disciplinary absences by 2022’, simply because it is unknown how many students receive multiple disciplinary absences and it is unclear which type of SDA is targeted within this aim.

To get a true sense of the scale of this issue, accounting for a larger number of relevant variables, there is a need for publicly available, longitudinal datasets that provide deidentified information of all suspension incidents at the student-level. Access to federal government funding should be contingent on the provision of these data for all school systems and sectors, nationally. Access to such data would enable modelling of the many contributing factors underlying exclusionary discipline in schools, thereby providing more detailed evidence to inform policy and practice, as well as a means of independently assessing the effectiveness of current government strategies. A similar call for transparent and accessible data reporting was made by the Productivity Commission (2023) in their recent review of the National School Reform Agreement (NSRA). An objective of the NSRA was that ‘academic achievement improves for all students, including priority equity cohorts’. However, the Productivity Commission report noted that despite inclusion of students with a disability in the priority equity cohorts considered under the NSRA, NAPLAN outcomes are not reported for students with disability, thereby prohibiting analysis of academic results, disproportionate underachievement, or persistent learning gaps (Productivity Commission, 2023). Clear and consistent reporting of data is required for the effectiveness of initiatives such as the NSRA to be independently evaluated and for effective education reform to take place.

Similarly, and as we have noted in previous research (Graham et al., 2023), Closing the Gap targets include school attendance; however, there is no target to reduce suspension. Indigenous students are still being suspended for truancy in QLD state schools, and disproportionately so but this is missed by the focus on attendance, which is a separate measure. These anomalies need to be urgently addressed in the next NSRA to put an end to exclusionary practices with an expert view to ensure the improvement of practice. Without this, exclusion may simply be driven underground with absenteeism skyrocketing as it did in Miami Dade Public Schools the year after suspensions were banned, leading to questions about the use of informal suspension and restricted enrolments (Geraty, 2017; McCombs et al., 2022). Given the known damages inflicted by exclusionary discipline on both students and their families, its ineffectiveness in addressing problem behaviour, and its association with harsh and unsupportive school climates, as well as systemic failure to provide reasonable adjustments for students with disability, it is critical that data extensive enough to enable independent expert assessment of public education policy and practice is made public. This could and should be actioned in the next NSRA.

Conclusion

In the present study, we examine Queensland school suspensions from 2016 to 2020, finding clear evidence that students in priority equity groups are consistently and markedly overrepresented in these statistics. By taking intersectionality into account, these analyses reveal that disproportionality in suspensions is increased for students in more than one priority equity group, with the greatest degree of risk and overrepresentation occurring for Indigenous students with a disability (NCCD) living in OOHC. However, and of critical importance, an examination of the group combinations in which overrepresentation is most pronounced shows that disability is the most predominant underlying factor. This is a critically important finding because the response when disability is a factor is different due to the impact of disability on students’ ability to access and participate in education. At the root of sustained increases in the use of exclusionary discipline in QLD state schools over the last decade and their disproportionate use on priority equity groups, are deep-seated issues that were not adequately examined nor canvassed in the 2017 Deloitte Disability Review. They were not, therefore, sufficiently addressed in the review recommendations, which were implemented in full by the Department of Education, and as a result suspensions continue to escalate. We have provided numerous recommendations in this paper to help address this problem; however, an independent Inquiry of the depth and rigor of that conducted in South Australia in 2020 is needed in QLD to ensure that the necessary reform occurs. Importantly though, QLD is not the only state or sector where increases in the use of exclusionary discipline is a major problem, but it is one of the only states that publishes relatively comprehensive data, and which will provide data through Right to Information requests. Ultimately leadership from the Australian government is needed to ensure that the right of children to an inclusive education is upheld no matter which state or territory they live in or in which school sector they are enrolled.

Appendix

See Table 4.

Table 4 Recommendations from the Deloitte Review Final Report