Overrepresentation of Indigenous students in school suspension, exclusion, and enrolment cancellation in Queensland: is there a case for systemic inclusive school reform?

Well-established evidence of the ill-effects of exclusionary school discipline, its disproportionate use on students of colour, and association with the “school-to-prison pipeline” has, in the last decade, led to systemic reforms in the United States, which are successfully reducing exclusion and improving outcomes. Few studies, however, have similarly investigated overrepresentation in Australia, with little attention to systemic reform as a result. In this study, we analysed suspension, exclusion, and enrolment cancellation rates in Queensland (QLD) government schools between 2013 and 2019 and found Indigenous students were consistently overrepresented. Suspension incidents proportionate to enrolments increased for all students, but this increase was faster for Indigenous than non-Indigenous students and driven primarily by steep rises in short suspensions during primary school (Preparatory-6). Exclusions increased—again disproportionately—for Indigenous students, chiefly in secondary school (7–12). During 2019, Physical Misconduct had the highest incident rate for both groups; however, Indigenous students were most overrepresented in suspensions for Disruptive/Disengaged behaviours. Further, while Indigenous students were overrepresented in all QLD regions, one region’s Indigenous suspension rate was higher than all others despite no difference in the distribution of Indigenous/non-Indigenous enrolments across regions. The scale and nature of Indigenous overrepresentation in exclusionary discipline incidents in QLD indicate clear need for further research to secure political commitment to systemic inclusive school reform, as well as to produce high-quality evidence capable of guiding that reform.


Introduction
Since the 1960s, education researchers have repeatedly raised concerns that students from marginalised and minority groups are subject to exclusionary school discipline at rates disproportionate to their representation in the total school population (Dunn, 1968;Waitoller et al., 2010). This is concerning given that there is a host of protective factors associated with being in school and negative outcomes linked to being excluded from the school environment (Quin, 2019). For example, mutually respectful teacher-student relationships and inclusive school cultures are associated with higher student attendance and engagement, and lower problem behaviours (Pyne, 2019;Quin, 2019). Further, effective academic instruction and behaviour support structures protect against the risk of suspension, school failure and early school leaving (Christle et al., 2005). Exclusionary school discipline, however, negatively impacts relationships, school belonging, and academic achievement (Jacobsen, 2020;Lacoe & Steinberg, 2019;Noltemeyer et al., 2015). These effects hold when controlling for factors such as prior achievement, indicating that exclusionary school discipline precipitates rather than mitigates problem behaviours, introducing significant long-term risks, which include poorer educational outcomes than might have already been expected (Rosenbaum, 2020), escalating antisocial behaviours (Hemphill et al., 2006), and early school leaving (Homel et al., 2012;Noltemeyer et al., 2015).
Beyond these adverse educational sequelae, disproportionate representation in school suspension and exclusion has been linked to overrepresentation in the criminal justice system (Hemez et al., 2020). Research evidence is particularly rich in relation to the criminalisation and incarceration of low-income African American, Latinx, and American Indian/Alaska Native children and young people (Cruz et al., 2021;Hirschfield, 2008), with exclusionary school discipline now considered a key contributor to a phenomenon described as the "school-toprison pipeline" (Novak, 2019). However, research investigating disproportionality in the use of exclusionary school discipline in settler colonies with a white non-indigenous majority outside of the United States-such as Australia, Canada, and New Zealand-is astonishingly rare, despite long documented evidence that indigenous (or First Nations) peoples are significantly overrepresented in these nations' criminal justice systems (Perdacher et al., 2019). In Australia, for example, Aboriginal and Torres Strait Islander peoples (Indigenous Australians) are estimated to account for just 3.3% of the total population but comprise 29.6% of the adult prison population (Australian Institute of Health & Welfare, 2021). Indigenous overrepresentation in the juvenile justice system is even greater and it varies by state. In the state of Queensland (QLD), for example, Indigenous children and young people accounted for just 7% of the 10-17-year-old population between 2015 and 2019, but 55% of those under youth justice supervision (Queensland Family and Child Commission, 2021).
The possibility of a link between exclusionary school discipline and entry to the justice system in Australia is particularly important at the 30-year anniversary of the 1991 Royal Commission into Aboriginal Deaths in Custody (Johnston, Overrepresentation of Indigenous students in school… 1991), which found "Aboriginal people are more likely to die in custody because they are arrested and jailed at disproportionate rates" (Evershed et al., 2020, np). This is not because Indigenous Australians have a "higher offending rate" as claimed by former Prime Minister Tony Abbott (Karp, 2020), but because they are more likely than non-Indigenous Australians to be stopped, questioned, fined, arrested, sentenced, and incarcerated, rather than receive bail or community service orders, and this can occur for as little as swearing in public or failing to pay fines (Australian Law Reform Commission, 2017). These outcomes strongly indicate a precipitative role for racial bias in community policing and the judicial process; one that remains unaddressed after 30 years (Langton, 2021).
Research from the United States has identified similar racial bias in the use of exclusionary school discipline, with African American and Latinx middle school students more likely to be subjected to exclusionary practices than White students for all types of violations, except disruption in the case of Latinx students (Skiba et al., 2011). This same research found that African American students are four times as likely and Latinx students twice as likely to be suspended or expelled for minor offences in primary school (Skiba et al., 2011). While the overrepresentation of Indigenous students in the use of exclusionary school discipline has also been observed in the Australian context, only a very small number of studies have investigated this to date (Beauchamp, 2012;de Plevitz, 2006;Gardiner et al., 1995;Graham et al., 2020a;O'Brien & Trudgett, 2020). None has been national, most have used only descriptive statistics or reviewed existing literature, and only one (Graham et al., 2020a) has conducted longitudinal or disaggregated analyses. These evidence gaps are principally due to the lack of publicly available data in Australia (Sweller et al., 2012), as well as differences in disciplinary processes and recording across a wide range of education providers (Graham et al., 2020a). These issues have to date prevented Australian researchers from investigating the local contours of this problem (Sweller et al., 2012); knowledge that is critical to secure the political commitment necessary for systemic reform.
Not only does this lack of data transparency prevent public scrutiny and problem identification but it also leads to gaps in public policy. An example is the National Agreement on Closing the Gap (Australian Government, 2020), which involves strengths-based strategies to reduce inequalities in life expectancy, children's mortality, education, and employment. The most recent analysis of progress against 17 agreed targets found that there has been no improvement in the school attendance rate of Indigenous students in the last 10 years (DPMC, 2020). Interestingly, the report does not mention the use of exclusionary school discipline and nor does the Agreement include targets to reduce its use, despite its known negative effects on school attendance, early school leaving, and increased contact with the criminal justice system. This appears to be a missed opportunity, given that 2 of 17 Closing the Gap targets are to reduce Indigenous overrepresentation in the criminal justice system (Australian Government, 2020).
If Indigenous children and young people are suspended and excluded from school at disproportionate rates and for disparate reasons, and if exclusionary school discipline is linked in any way to criminal justice trajectories, then we must, as a matter of urgency: identify overrepresentation in school suspension and exclusion, and any patterns related to it; challenge implicit bias, racism, and discrimination wherever they may exist; and strengthen culturally appropriate evidence-based prevention and intervention frameworks, as well as implement them with fidelity on a systemwide basis. Achieving equity in school discipline outcomes in Australia will require legislative reform and the introduction of safeguards to protect Indigenous students from inappropriate use of suspension and exclusion with mechanisms to monitor system compliance, as well as systemic reform to test viable alternatives suitable for the Australian context and to reduce educators' reliance on exclusionary discipline. Such reforms, however, require a fundamental reset in knowledge, attitudes, and capability both within and outside a largely non-Indigenous teaching profession (Morrison et al., 2019;Riley & Pidgeon, 2019). For a model on how this might be achieved, Australia needs to look no further than the United States, where mounting evidence of the ill-effects of exclusionary school discipline, together with the overrepresentation of African American children and young people, has prompted farreaching education reforms which are gradually making their way through US public-school systems (Cruz et al., 2021;Nese et al., 2021;Steinberg & Lacoe, 2017).

Turning its back on zero tolerance
In January 2014, the Obama Administration, through the Office of Civil Rights, released a set of Guiding Principles on school discipline to (a) warn school administrators against engaging in direct or indirect discrimination based on race, and to (b) encourage public-school systems to engage in evidence-based reforms aimed at improving school climates, teaching quality, and student support (US Department of Education, 2014). While the Guiding Principles referred specifically to race discrimination, it was noted that "much of the analytical framework … also applies to discrimination on other prohibited grounds [e.g., disability]" (US Department of Education, 2014). Later that same year, the US Department of Health and Human Services, and the US Department of Education (2014), jointly issued a statement strongly discouraging the use of exclusionary discipline with young children, due to known ill-effects on children's academic, social-emotional, and behavioural development.
Many US public-school systems have since adopted substantial reforms: California banned the use of suspensions and exclusions for children in the early years of school (K-3; Freedberg, 2019), Chicago Public Schools (CPS) and Philadelphia reduced permissible suspension length, and all three jurisdictions either banned or limited suspensions for minor reasons. Recognising the tendency for exclusionary discipline to reinforce task and school avoidance (Graham, 2018), Arkansas banned out-of-school suspension for truancy and Miami-Dade County Public Schools eliminated suspensions entirely (Anderson, 2020;Hinze-Pifer & Sartain, 2018). Within 1 year of the Guiding Principles being issued, …22 states and the District of Columbia had revised their laws in order to require or encourage schools to: limit the use of exclusionary discipline practices; implement supportive (that is, nonpunitive) discipline strategies that rely on behavioral interventions; and provide support services such as counseling,

3
Overrepresentation of Indigenous students in school… dropout prevention, and guidance services for at risk students. (Steinberg & Lacoe, 2017, p. 44) And, within 2 years of the Guiding Principles being issued, "23 of the 100 largest school districts nationwide had implemented policy reforms requiring nonpunitive discipline strategies and/or limits to the use of suspensions" (p. 44).
While each of these reforms met with opposition from industrial associations and conservative commentators, with claims that they would lead to an increase in school violence (Eden, 2019), empirical evidence over time showed the opposite. Abolishing suspension for minor incidents in Californian public schools, for example, did not result in predicted "chaos" or increased school violence (Losen & Martin, 2018). Rather, longitudinal analyses reveal a significant decline in suspensions, alongside improvements in school climate and academic outcomes. The same research found a large and significant decline in the number of instructional days lost and a narrowing of the racial gap in exclusion rates, attributable to reduced use of suspension for minor incidents (Losen & Martin, 2018). These and other positive outcomes from California's reforms paved the way for the banning of suspension for elementary and middle school children (Kindergarten through Grade 8) for classroom disruption and "wilful defiance" (Freedberg, 2019). Importantly, these reforms-like those in other US public-school systems that have successfully reduced the use of exclusionary discipline, such as Chicago Public Schools-did not just involve banning or limiting the use of exclusionary discipline. In addition to severely limiting and monitoring the use of suspension and exclusion, these systems also required schools to use evidence-based educative responses, delivered through a Multi-Tiered Systems of Support (MTSS) framework designed to improve students' academic achievement, social-emotional understanding, and behavioural interactions. By implementing MTSS, school systems in the United States went beyond discipline reform and instead engaged in systemic inclusive school reform.

What is systemic inclusive school reform?
Upon ratifying the Convention on the Rights of Persons with Disabilities (CRPD; United Nations, 2008), Australia is committed under international human rights law to engage in systemic inclusive school reform (Graham et al., 2020b). General Comment No. 4 (GC4) on Article 24 of the CRPD, which articulates the human right to inclusive education, defines inclusion as follows: …a process of systemic reform embodying changes and modifications in content, teaching methods, approaches, structures and strategies in education to overcome barriers with a vision serving to provide all students of the relevant age range with an equitable and participatory learning experience and environment (United Nations, 2016, para 11).
GC4 also outlines the nine core features of an inclusive education system: whole system, whole school and whole person approaches; committed leadership, supported teachers, respect for and value of diversity, learning-friendly environments, effective transitions, and recognition of partnerships. Critical to the success of 1 3 systemic inclusive school reform are frameworks that promote consistency in policy logics and practice across regions, schools, and classrooms.
MTSS is a "fully braided" (McCart et al., 2014, p. 252) prevention and intervention framework that was developed from research in the fields of implementation science and inclusive education, and which is increasingly being used internationally to support systemic inclusive school reform (Sailor et al., 2018). MTSS guides school-wide consistency in the delivery of processes and practices that enable regular schools to enrol all students in their local catchment. This is achieved through the universal provision of high-quality accessible pedagogies and evidence-based programs at Tier 1 together with reasonable adjustments, followed by timely evidencebased supports targeted to small groups or individuals at increasing levels of intensity and frequency (Tiers 2 and 3) identified through data-based decision making (Sailor, 2015). It is through these tiered approaches that some public-school systems in the United States, such as Chicago Public Schools, have been able to reduce suspension, while improving safety, school connectedness, attendance, and academic outcomes (Hinze-Pifer & Sartain, 2018). Reducing disproportionality has been more difficult, as attitudes that drive implicit racial bias can remain intact, despite structural reforms (Trinidad, 2021).
Systemic inclusive school reform as articulated in GC4 goes beyond the decontextualised or ad hoc provision of prevention strategies and support programs by grounding them in a strong foundation of inclusive school cultures developed through culturally appropriate (relational) pedagogies, positive teacher-student relationships, and mutually respectful parent-school partnerships (Graham et al., 2020a). These need to be contextualised in and for unique Australian contexts and guided centrally with adequate support for educators to determine-quickly and easily-which programs and practices have the best evidence and are most likely to be effective (Laurens et al., 2021b). Research is needed to evaluate how these programs and practices can be adapted for local needs through co-design, particularly with and for Indigenous young people (Shay & Miller, 2021), yet still implemented with fidelity (Laurens et al., 2021b). This is critically important considering the diversity of some Australian schools and the heterogeneity of cultures within them (Keddie et al., 2013). Recent results of a best-evidence synthesis investigating the effectiveness of school-based interventions in reducing disproportionality suggest, however that proscribing or banning the use of exclusionary discipline may be a critical first step to support meaningful change in both attitude and practice, which is a necessary foundation for successful reform implementation (Cruz et al., 2021).

Reforming discipline in Australia
Australian school systems have not yet implemented the systemic reforms necessary to reduce educator reliance on exclusionary discipline, although it is clear from recent reports that viable alternatives are needed urgently. For example, in 2014-the same year that the Obama Administration intervened-the QLD state government instead expanded school principals' disciplinary powers leading to large and sustained increases in the use of exclusionary discipline in QLD state schools (Graham, 2018). Concern over rising suspension rates in other education jurisdictions in Australia prompted the South Australian (SA) state government to commission an independent Inquiry into Suspension, Exclusion and Expulsion processes in South Australian government schools (Graham et al., 2020a) and the New South Wales (NSW) state government to commission an evidence review (Pearce et al., 2019). Both final reports, released in 2020, recommended reforms aimed at improving the provision of appropriate, timely and evidence-based supports to students who experience learning and behavioural difficulties at school: systemic inclusive school reform supported by legislative reform and the implementation of MTSS in SA (Graham et al., 2020a), and a system of social-emotional skill building and complex behaviour support in NSW (Pearce et al., 2019).
The SA government has since committed to many of the Inquiry's 76 recommendations, along with $15 million to "develop a systemic strategy to minimise suspensions for all students-but especially suspensions for Receptions to Year 2 s, Indigenous students and children with disabilities or in state care" (Williams, 2021, para 2). The strategy will implement new data dashboards to enable system monitoring and set reduction targets to stop the use of suspensions for minor reasons, such as not following instructions, minor physical acts and talking in class. Both Flexible Learning Options (FLOs) and exclusions will be abolished, communication to parents and students about their rights will be improved, and complaints and appeals processes strengthened.
Following the release of their evidence review, the NSW government proposed a new Behaviour Strategy (Department of Education NSW, 2021), noting that "suspension rates for vulnerable students in NSW are too high, and disproportionately so for students with disability, Aboriginal students, students in rural and remote areas, students in out-of-home care and students experiencing socioeconomic disadvantage" (p. 7). The proposed NSW Behaviour Strategy made the case for educative alternatives to the use of exclusionary discipline, stating that there is "no evidence that suspension is an effective mechanism for improving or managing student behaviour", and that it "does not provide students with the support they need to achieve behavioural change" (p. 7). Like some US public-school systems, the NSW government proposed a reduction in the maximum suspension length from 20 to 10 days for students in Grades 3-12, in addition to limiting suspensions to a maximum length of 5 days for students in the early years of school (K-2) and only then for serious circumstances. The proposed reductions are mild in comparison to those enacted in many US systems where suspension is banned for any reason in K-2, and across K-8 for wilful defiance (e.g. California), or is limited to 3 days duration even for serious misdemeanours (e.g. Chicago Public Schools; Graham et al., 2020a). Nonetheless, the proposed reductions have been "vehemently" opposed by "the Federation of Parents and Citizens Association of NSW, the NSW Secondary Principals' Council, the NSW Primary Principals' Association, and the NSW Teachers Federation" (para 2), who collectively argue the proposed reforms will reduce the authority necessary for schools to achieve "safe, calm learning environments" (New South Wales Teachers Federation, 2021, para 2).

The Queensland context
The QLD state government school system is Australia's third largest behind NSW and Victoria. In 2021, over half a million students were enrolled in 1254 schools across an immense area two and a half times the size of Texas (Department of Education Queensland, 2021a, 2021b). With more than 220,000 Indigenous Australians, accounting for 4.6% of the state's total population, QLD has Australia's second largest population of Indigenous peoples behind NSW (Indigenous Education and Research Centre, 2021). However, due to higher fertility and higher mortality rates, the age distribution of Indigenous Australians is skewed towards younger age groups (Australian Institute of Health & Welfare, 2021). For this reason, Indigenous students account for 10.9% of QLD state school enrolments (Department of Education Queensland, 2021b).
Various education reforms have been enacted over the past decade in QLD, some of which have impacted the use of exclusionary discipline (Graham, 2018). In 2014, legislation was passed via the Education (Strengthening Discipline in State Schools) Amendment Bill 2013 (Queensland), which resulted in greater autonomy for government (or public) school principals to issue suspensions, reduced obligation to consult with students' families, stripped students of the right to appeal short suspensions, introduced the ability to suspend for criminal offences or non-criminal behaviours occurring outside the school environment, and changed the length of short and long suspension: increasing short suspensions from 1-5 to 1-10 days, and decreasing long suspensions from 6-20 to 11-20 days (Carden, 2018). These changes were intended to relieve the burden on teaching staff obviated by onerous paperwork processes, as well as to reduce suspensions, exclusions, and enrolment cancellations. The latter two consequences were particularly in focus, as their rate had more than doubled in the years (DETE, 2014. Recent research has demonstrated that these legislative reforms have not been effective-and may, in fact, be harmful-especially to young children just commencing school and those transitioning from primary to secondary school (Graham, 2018). While enrolment cancellations did decline after the 2014 change in legislation, exclusions remained steady and suspension rates rose steeply, surpassing enrolment increases. Graham's (2018) analyses disaggregated by calendar year, school phase and grade level, but did not investigate differential impact for Indigenous versus non-Indigenous students. This was noted as a critical gap in the extant research, "given prior evidence that Indigenous students, students in out-of-home care, and students with disability are overrepresented in suspensions and exclusions" (Graham, 2018, p. 16). The use of exclusionary discipline and its disproportionate impact on students in these three priority equity groups has since been raised in Hearings of the current Royal Commission into Violence, Abuse, Neglect and Exploitation against People with Disability, which has focussed a great deal of attention on the QLD government school system (Iaria, 2020). Noted at Hearing 7 was QLD's very high suspension rate, which at 150.37 suspension incidents per 1000 students far outstrips those of comparable Australian jurisdictions: almost double that of NSW (87.62 suspensions per 1000 students), and almost triple that of the Australian Capital Territory (58.05 suspensions per 1000 students; Graham et al., 2020a).

Overrepresentation of Indigenous students in school…
Concerns have also been raised as to the impact of QLD's rising number of school exclusions on Indigenous students and its relationship to the school-to-prison pipeline by those working with youth in the community legal sector, who noted that 38% of the 55,752 Indigenous students enrolled in QLD state schools were either suspended, excluded, or had their enrolments cancelled in 2018 (Iliffe & Stevenson, 2020). Despite these alarming statistics, the current Queensland government has not conducted any reviews of their approach to school discipline or proposed any reforms since the legislative changes enacted in 2014. And, although the United States has backed away from zero tolerance, Queensland's Minister for Education has in recent years defended large increases in suspension and exclusion rates, even those affecting Queensland's youngest Preparatory (Prep) year children, attributing them to her government's "zero tolerance approach" (Caldwell, 2019). Due to a lack of data and research, it has not been possible to examine who exactly is being impacted by this approach or what its long-term effects on vulnerable young Queenslanders might be.

The present study
The present study examined whether Indigenous students are overrepresented in exclusionary discipline incidences in Queensland government schools. Our investigation was underpinned by three research questions: 1. For Queensland government schools throughout 2013-2019, what has been the trend over time for suspensions, enrolment cancellations, and exclusions at the population level? 2. Do these trends vary by a. Short versus long suspensions b. Primary versus secondary school c. Reason d. Education region (geographic location) 3. Are Indigenous students overrepresented and, if so, does their position relative to non-Indigenous students remain stable over time?

Methods
The present study used government school enrolment and exclusionary discipline data publicly available from the QLD Department of Education (DoE) website (Department of Education Queensland, 2021b, 2021c) from the years 2013 to 2019. In 2019, there were 1236 QLD state schools across seven administrative regions enrolling a total of 559,099 students in Prep to Grade 12 (Department of Education Queensland, 2021b, 2021c), with just over one in 10 of these students (10.3%) identifying as Indigenous (Table 1). Records of suspensions, enrolment cancellations, and exclusions are reported by the DoE in terms of the number of incidents, rather than the number of students issued with a suspension, enrolment cancellation or exclusion. This gap in public data availability prevents assessment of the number of Indigenous and non-Indigenous students receiving multiple suspensions, which is of critical relevance in fully understanding overrepresentation and its impact. Nor is there publicly available data to investigate the intersectionalities of poverty, race, gender, disability, and out-of-home care status, all being important factors to consider prior to devising or implementing reforms aimed at addressing disproportionality. This level of analysis has been conducted only once in the context of the SA Inquiry (Graham et al., 2020a) where the researchers were provided access to non-publicly available data. While the present study was constrained to less in-depth analysis by the limited data currently made public, conducting analyses to determine whether overrepresentation exists in QLD would add to what we know of the local contours of this problem. This may, in turn, convince the QLD Parliament of the need to either hold a similar Inquiry to South Australia or even demonstrate the need for national intervention by the Australian Government and Australian Human Rights Commission, like that of the Obama Administration and the US Office of Civil Rights. Reported incident frequency can still be disaggregated by calendar year (2013-2019), semester, region, Indigenous status, grade level, incident type, reason, and education region. As such, in all analyses, incident served as the main unit of analysis. To be consistent with the Department's reporting standards, all analyses used the rate of exclusionary discipline incidents per 1000 students enrolled.
There are four types of formal disciplinary consequence in the QLD government school system: suspensions, exclusions, enrolment cancellations, and charge suspensions.
"Charge" suspensions are issued in the circumstance where a student has been charged by the criminal justice system with a "serious offence". A school principal may also issue these in cases where a non-serious offence has been committed, and they believe it would be in the best interests of the other students and staff for the charged student to be absent. Charge suspensions may last for the duration of the charge or until the principal decides the student can return. Only 50 charge suspensions were reported during the years 2013-2017, and were not disclosed in the reporting period of 2018-2019. Due to inconsistent reporting of this type of suspension in the years under investigation and the small numbers involved, charge suspension incidents were excluded from analyses. We note, however that Indigenous students accounted for 11 of those 50 charge suspension incidents (22.0%).
Suspensions involve the removal of a student from school for a fixed time period. Since the 2014 legislative change, "short" suspensions in QLD have a duration of 1-10 days, and "long" suspensions 11-20 days (Graham, 2018).
Exclusion involves the removal of a student from their school "for a period of not more than 1 year, or permanently, if the principal is reasonably satisfied a ground exists for the exclusion" (Department of Education Queensland, 2020a).
Enrolment cancellations can be issued to students who are above compulsory school age (> 16 years) at the discretion of the principal, usually on the basis of "refusal to participate in the educational program provided by the school" (Department of Education Queensland, 2020b).
For the purposes of the present study, incident rates were examined over time for the type of incident (suspension, enrolment cancellation, or exclusion), the length of suspension incident (short or long suspension), and by the school phase (primary or secondary suspensions and exclusions). To examine patterns in these rates over time, and whether they varied between Indigenous and non-Indigenous students, curve estimation analyses were conducted with time (calendar year) as the independent variable and incident rate per 1000 students as the dependent variable. These analyses were used to test for the presence of linear or quadratic trends, where a linear trend indicated a stable rate of increase or decrease over time, whereas a quadratic trend indicated a curvilinear relationship, or a change in the rate of increase or decrease over time (Sweller et al., 2012). When interpreting these slopes, it is important to note that the use of incident frequency, rather than the number of students who are suspended or excluded, meant that the same student might contribute to the incident rate more than once (i.e. where a student was issued with multiple suspensions in a year). Hence, each slope represents an incident rate over time, rather than necessarily an increase or decrease in the number of students being suspended over time.
Where linear trends were observed for both Indigenous and non-Indigenous incident rates, the slopes of two independent groups were compared using the t statistic, to determine whether there were significant group differences in the rate of increase/ decrease over time. These comparisons were also conducted for the linear trends in instances where both a significant linear and a quadratic trend were detected for one or both of the slopes. The comparisons used the following formula (Howell, 2012), where b represents the slope and s b its standard error: Durbin-Watson values (d) were calculated to test for the presence of any first order autocorrelation in the models, which, where significant, indicate violation of the assumption of independence of errors in linear regression. For a model including one predictor, an intercept, and seven cases, the following upper and lower bounds apply for positive autocorrelation at alpha = .05: d L = 0.700, d U = 1.356 (Savin & White, 1977), where values above d U indicate no significant autocorrelation. The d value obtained in each regression analysis was compared against these values. In instances where d exceeded 2, signifying negative autocorrelation, 4 − d was computed and compared against these values.
For visualisation purposes only, the slopes of incidents over time for all students are presented, in addition to the slopes for the Indigenous and non-Indigenous groups. These overall trends in incidents among all students were not used in any statistical comparison of slopes, as only the slopes for Indigenous and non-Indigenous groups were independent of each other.
Detailed analysis of suspensions and exclusions by School Grade Level and Consequence Reason was conducted using data from the calendar year 2019 only. Suspensions and exclusions by Grade Level (for Indigenous and non-Indigenous groups) were differentiated using enrolment numbers at each level in the calculation of the incident rates. Suspension and exclusion rates by Indigenous status and education region were analysed over the years 2016-2019, due to data availability.

Results
Trends in suspensions, cancellations, and exclusions over time Table 2 presents the separate linear regression models for suspension, cancellation, and exclusion incident rates, each using time (calendar year) as the predictor. Incident rates per 1000 students averaged over all years are included for reference. For some of these slopes, significant quadratic trends were observed in addition to linear, as indicated within the table. Durbin-Watson values all exceeded d U , confirming a lack of significant positive autocorrelation in all models. In several models, values of d exceeded 2, and the subsequent recalculated values (4 − d) fell between the confidence intervals, reflecting inconclusive results with regards to negative autocorrelation. However, no definite negative autocorrelation was observed in any models, and so these analyses were considered acceptable. Figure 1A-D displays the incident rate per 1000 students enrolled, and the predicted values derived from the linear regression models, for (a) suspensions, cancellations, and exclusions for all students (Indigenous and non-Indigenous combined), (b) for suspensions only (Indigenous vs. non-Indigenous), (c) for enrolment cancellations only (Indigenous vs. non-Indigenous), and (d) for exclusions only (Indigenous vs. non-Indigenous).
When considering the total suspension rates (Fig. 1A), a significant increase over time was apparent for all students (see Table 2 for regression results). However, as displayed in Fig. 1B, the rate of increase differed substantially when these slopes were calculated separately for suspensions issued to Indigenous as compared to non-Indigenous students, with a significantly steeper increase over time apparent in the suspension incident rate of Indigenous students, t (10) = 4.52, p = .001. Unlike suspension incidents, enrolment cancellation incidents for all students decreased in the 7-year period 2013-2019, and these decreasing slopes were significant for both Indigenous and non-Indigenous students ( Fig. 1C; Table 2). The comparison of Indigenous to non-Indigenous trends indicated that there was no significant difference between the rate of decrease in enrolment cancellations, t (10) = − 1.50, p = .164.
Exclusion incidents also increased significantly over time for Indigenous students, yet not for non-Indigenous students ( Fig. 1D; Table 2). There was a significant difference between the two groups in terms of the rate of exclusions over time, t (10) = 3.37, p = .007 (Table 3). Figure 2A displays, for all students, the short and long suspension rates per 1000 students enrolled. Figure 2B and C display these disaggregated by Indigenous status. For short suspensions (Fig. 2B), the rates appeared to increase over time for all students, and also when examining the incident rates of Indigenous and non-Indigenous students separately. There were no clear linear trends over time for long suspensions (Fig. 2C), combined or disaggregated. In some instances, significant quadratic trends were observed for one group or the other.

Magnitude of overrepresentation for length of suspension
Change in the definition of short and long suspension days in 2014 may explain the substantial drop in the rates of long suspensions after 2013, as many exclusionary periods previously classified under a "long" suspension may have been absorbed into the now longer "short" suspension category. When examining long suspension rates using all 7 years of data, the direction of the time trends was negative for both Indigenous and non-Indigenous students (although statistically non-significant), likely due to this definitional change.
Accordingly, an alternative analysis which excluded the year 2013 was conducted, and found that the trends were still non-significant, but the direction had changed and instead indicated a marginal increase in both Indigenous and non-Indigenous rates of long suspension in the years 2014-2019. This analysis of the 6 calendar years is presented in Table 2 and the trend lines obtained through these

Magnitude of overrepresentation in suspensions/exclusions by school phase
To determine whether the degree of disproportionate Indigenous representation in exclusionary discipline incidents depended on school phase, Indigenous and non-Indigenous suspension incidents were compared first in primary school and then in secondary school, repeating these analyses for exclusion incidents. Only suspension and exclusion incidents were analysed (enrolment cancellations are issued only for students > 16 years of age). Table 4 displays the linear regression model fits for incident rates per 1,000 students predicted by time, and Fig. 3A-D displays both the raw data and the predicted values from each model. Quadratic trends were also observed in several instances, as indicated in Table 4. In the primary grades, suspension incidents increased over time significantly for both non-Indigenous and Indigenous groups, yet the slope was significantly steeper for Indigenous incidents, indicating that primary school suspensions increased at a faster rate for this group, t (10) = 3.48, p = .006. Similarly, secondary school suspensions increased significantly over time for both groups; however, the difference in the slopes was non-significant, such that incidents increased at a similar rate for both groups, t (10) = 1.73, p = .115.
For exclusions, incidents did not significantly increase in primary school for either group of students. In secondary school, exclusion incidents increased significantly for Indigenous students, but not for non-Indigenous students, and the slopes of these two groups significantly differed, t (10) = 2.97, p = .014.

Magnitude of overrepresentation in suspensions/exclusions by grade level
Figures 4 and 5 display suspension and exclusion incidents per 1000 students enrolled by grade level and Indigenous status, during the year 2019. Enrolment cancellations were not analysed by grade level, as they are reported only for Grades 9-12 (for students who are 16 years or older). A pronounced increase in suspensions and exclusions is evident for both groups in Grade 7, which marks the commencement of secondary school in Queensland. The increase in suspensions per 1000 students for non-Indigenous (152.6%) exceeded that for Indigenous students (69.8%) but not enough to address Indigenous overrepresentation. Suspension incidents peaked in Grade 8 for Indigenous students and Grade 9 for non-Indigenous students, whereas exclusion incidents peak in Grade 9 for both groups. The lowest frequency of suspension and exclusion incidents proportionate to enrolments occurred in Prep, for both groups of students, although the Indigenous suspension rate in Prep was 2.4 times the non-Indigenous suspension rate.

Reasons for exclusionary discipline
Figures 6 and 7 display suspension and exclusion incident rates, respectively, for 2019, disaggregated by reason and Indigenous status. To reduce categories Overrepresentation of Indigenous students in school… of analysis, the original 16 reason categories were collapsed into seven aggregate categories (see Table 5 for original and re-categorised reasons). To compare the relative rates, the ratio of incident rates proportionate to enrolments between the two groups were computed. Cohen's h, which is derived from the z-test of two proportions was used as a measure of effect size. According to convention, a small effect is 0.2, medium is 0.5, and large is 0.8 (Cohen, 1988). For all students, physical misconduct represents the highest number of incidents per 1000 students, for both suspension and exclusion. A greater number of incidents per 1000 students was apparent for Indigenous than non-Indigenous students in every category (Table 4); nonetheless, degree of overrepresentation varied according to reason.
For suspensions, the number of incidents proportionate to enrolments for disruptive/disengaged behaviour, verbal/non-verbal misconduct, and physical misconduct were each between three and four times as high for Indigenous compared to non-Indigenous students (range 3.3-3.8 times; h range 0.24-0.35 [small-to-medium effects]), whereas the smallest effects were apparent for property misconduct and substance misconduct, which were 2.4 and 2.0 times as high, respectively (h = 0.09 and 0.06). Within individual categories, Indigenous overrepresentation was greatest in "Physical Misconduct involving Students not involving an object" (h = 0.30), "Verbal or Non-Verbal Misconduct involving Adults" (h = 0.24), "Persistently disruptive behaviour adversely affecting others" (h = 0.17), and "Refusal to participate in the program of instruction" (h = 0.16).
For exclusions, effect sizes contrasting Indigenous and non-Indigenous rates for each specific reason were negligible in magnitude (h range 0.00-0.05) but followed a similar pattern to suspensions in that the effect was greatest for physical misconduct (h = 0.05).
No analysis of the reasons for enrolment cancellation was conducted, due to the fact that every incident was attributed to a "Refusal to participate in the program of instruction".

Magnitude of overrepresentation in suspensions/exclusions by education region
A final analysis sought to determine whether disproportionality in suspensions or exclusions was related to education region. There are seven education regions in the Queensland government school system: Metropolitan, South East, Darling Downs South West, North Coast, Central Queensland, North Queensland, and Far North Queensland. The Metropolitan and South East regions are the most populous and urbanised, with the remaining regions characterised by smaller urban centres, as well as rural, remote and very remote communities. Table 1 shows the distribution of enrolments in each region, indicating both the distribution of Indigenous and non-Indigenous students across the state of Queensland (by region), as well as the distribution of Indigenous and non-Indigenous students within each region.
The region with the highest proportion of Indigenous students relative to total enrolments statewide (20.6%), as well as the highest proportion of Indigenous student enrolments relative to total regional enrolments (33.4%), is Far North Queensland. The region with the lowest proportion of Indigenous students statewide is Darling Downs South West (10.9%), while the lowest proportion of Indigenous students relative to non-Indigenous students at the regional level is in Metropolitan (4.8%). Figure 8A-D displays the rate of incidents per 1000 students separately for Indigenous and non-Indigenous suspensions and exclusions, with each of the seven regions denoted by separate lines. Indigenous students were overrepresented in suspensions in all seven regions across Queensland, while non-Indigenous suspension rates were similar across regions. Indigenous rates were prominently elevated for the Darling Downs South West region. During 2019, the Indigenous suspension rates per 1000 students in this region were 1.5 times as high as Central Queensland, Far North Queensland, North Coast, and North Queensland, respectively, with small-tomedium effect sizes . When compared with the Metropolitan and South East regions, Darling Downs South West rates were 1.4 times as high (h = 0.28 and h = 0.26, respectively).
Exclusion rates were similarly low for the non-Indigenous group in each region, while there was variability over the 4 years for the Indigenous group, depending on region. For instance, exclusion rates for the Indigenous group tended to reduce over time for South East, Darling Downs, and Central Queensland regions, while North Queensland increased slightly.

Discussion
The present study examined trends over time in suspension, exclusion, and enrolment cancellation incidents both at the population level and disaggregated by Indigenous and non-Indigenous status. Suspension incidents proportionate to enrolments in Queensland government schools increased significantly during the 7 years from 2013 to 2019, and at a faster pace for Indigenous than for non-Indigenous students, particularly during the primary school years. Further analyses revealed that this was principally due to the rise in use of short suspensions, which we note were extended from 1-5 days duration (prior to 2014) to 1-10 days duration (from 2014) as part of the legislative changes expanding QLD state school principals' powers to suspend (Graham, 2018). Recall also that this legislation stripped students and their families of the right to appeal short suspensions. While there were trend increases in short suspensions of non-Indigenous students, and in long suspensions for both groups, these increases were not statistically significant. Indigenous overrepresentation in suspensions has, therefore, increased over time: an outcome that may be contributing to the lack of progress in closing attendance and achievement gaps between Indigenous and non-Indigenous students in the last decade (Australian Government, 2020).

Differences in school phases and across grades
Our analyses at school phase and grade level found a greater rate of increase in suspensions of Indigenous students in the primary school years compared to suspensions of non-Indigenous students. It is important to recall here that the US Department of Health and Human Sciences and Department of Education followed the Obama Administration's Guiding Principles by jointly issuing a statement in late 2014 strongly discouraging the use of exclusionary school discipline with young children due to known negative impacts on children's development. The SA Inquiry final report similarly provided strong evidence of the cumulative impact of exclusionary school discipline over time with take homes and short suspensions leading to longer suspensions and exclusions (Graham et al., 2020a). This graduation in severity of behaviour and consequence occurs because exclusionary discipline does not address the issues underlying children's behaviour, and can instead reinforce it (Graham, 2018).
The present study found that secondary school suspensions increased over time at a similar rate for both groups, with the magnitude of Indigenous overrepresentation decreasing slightly on transition to secondary school in Grade 7. Note that this was not due to a decrease in suspensions of Indigenous students, but because of an increase in suspensions of non-Indigenous students, which was coming off a much lower base. Further, while Indigenous students are overrepresented across all grade levels from Prep to Grade 12, suspension incidents peaked in Grade 9 for non-Indigenous students, but in Grade 8 for Indigenous students. This observation raises questions over the attrition of Indigenous students and whether there is a different peak because early school leaving is potentially higher and earlier among Indigenous students and whether school attrition is primarily affecting those students receiving repeat (or multiple) suspensions. Similarly, there is an important question as to when these now secondary school students' suspension experiences began and whether these students were the early victims of increases in the use of exclusionary discipline in primary schools stemming from the change in legislation in 2014.
As the QLD Department of Education publishes incident data only and not the number of students suspended, it is currently impossible to determine the number of repeat suspensions and whether there is a different pattern of single versus repeat suspensions for Indigenous versus non-Indigenous students. This remains a critical gap in the knowledge necessary to implement successful inclusive school reform. It is also unknown whether the students receiving multiple suspensions also have a disability and/or are living in out-of-home care. Both factors have been identified as contributing differentially to Indigenous overrepresentation, including by Indigenous scholars (Avery, 2018), with each factor requiring careful consideration in the development of response frameworks, such as MTSS (Graham et al., 2020a).

Differences in reasons
Our investigation of reasons for suspension and exclusion indicated that Physical Misconduct accounted for the highest rate of incidents for both of these consequences and for both student groups, although Indigenous students were overrepresented each time. It is important to note that our aggregate Physical Misconduct category includes four subcategories with "Physical Misconduct involving students not involving an object" contributing 6.5 times the incident rate of the next nearest subcategory. Peer-to-peer altercations are the most common among the incidents described as Physical Misconduct, not those involving adults or involving weapons, which is similar to findings from the SA Inquiry (Graham et al., 2020a). Importantly, Indigenous young people interviewed in the conduct of that research spoke of being subject to exclusionary discipline for fighting which they said was often in defence against unaddressed bullying and racism in schools. Similar research is urgently needed in QLD to understand and address factors underpinning these discrepant rates, and that research needs to begin by gathering the perspectives and experiences of suspended and excluded Indigenous students themselves.

And differences in degrees
While Physical Misconduct may have the highest incident rate, the greatest degree of Indigenous overrepresentation was in suspensions in the Disengaged/Disruptive behaviour category, which includes three subcategories: truanting, refusing to participate in the program of instruction, and persistently disruptive behaviour adversely affecting others. This discrepancy is a critically important finding: first, because improving attendance is a key target in the National Agreement on Closing the Gap, on which the most recent report shows there has been no improvement in 10 years; second, because suspending and excluding students for truancy rewards and, therefore, reinforces non-attendance (Chu & Ready, 2018); third, because task and school avoidance-including that expressed through disruptive behaviour-is common among students with disabilities, particularly those affecting hearing, language and executive function (e.g. attention, working memory, self-regulation; Graham & Tancredi, 2019); and finally, because behaviour is a form of communication and refusal to participate in the program of instruction may indicate that the program is not culturally relevant or is not being taught in accessible ways (Graham et al., 2020a;Llewellyn et al., 2018;Morrison et al., 2019).
Importantly, research from the United States highlights the importance of contextualising programs and frameworks to ensure that they are culturally appropriate. For example, there is evidence of the effectiveness of Positive Behaviour Intervention Supports (PBIS) for reducing the incidence of suspension overall but not for reducing disproportionality (Lustick, 2017;Trinidad, 2021). Empirical evidence also suggests that addressing implicit racial bias and improving teacher empathy is necessary to address disproportionality (Ispa-Landa, 2018), given that the evidence base for most programs and frameworks has not been derived from diverse or globally representative samples. It is, therefore, critical to investigate the overrepresentation of Indigenous students to understand what these students are communicating about their education and where reforms are needed.

In different places
We examined differences across the seven education regions of QLD, which vary with respect to population density and diversity. Our analyses found that, from 2016 to 2019, suspensions were higher for Indigenous students in the Darling Downs South West region compared to other regions, with rates during 2019 being between 1.4 and 1.5 times as high as in each of the other six regions (each of which were disproportionately high in comparison to suspensions of non-Indigenous students). Non-Indigenous suspensions tended not to vary by region. This regional analysis is important because it shows that Indigenous overrepresentation in suspensions and exclusions is not an artefact of higher or lower proportions of Indigenous students in some regions. Indeed, Far North QLD is home to the highest proportion of QLD's Indigenous student enrolments, with roughly a fifth of all Indigenous students in QLD living in this region (see Table 1). Yet the highest proportion of suspensions given to Indigenous students was in Darling Downs South West region, which has the lowest proportion of QLD's Indigenous student enrolments of all seven regions (enrolling about one tenth of the state's Indigenous students).
At the region level, however, DDSW has the third highest percentage of Indigenous students compared to non-Indigenous students (Table 1), after Far North QLD and North QLD. High suspension rates in DDSW cannot, therefore, be explained by educator unfamiliarity with Indigenous students due to very low enrolments (as might be claimed in Metropolitan region, for example) or alternatively as the artefact of comparatively high Indigenous enrolment numbers compared to other regions in the state. These findings confound any attempt to provide easy explanations for higher Indigenous suspension rates in this region and demand further investigation. On the ratio of Indigenous students to incidents in the DDSW region, it appears that Indigenous children and young people in this corner of the state have up to a one in two chance of being suspended, although it is likely that a substantial proportion of these students are receiving multiple (repeat) suspensions. Without publicly available data it is impossible to tell how many Indigenous students are receiving repeat suspensions, yet this is critically important information because these are the students for whom suspension does not work but also the students against whom it is most used (Graham et al., 2020a).

Enrolment cancellations and exclusions
Our analyses found a decline in enrolment cancellations over time for all students; however, as this occurred at a similar pace for both non-Indigenous and Indigenous students, the decline did not address the overrepresentation of Indigenous students. A different pattern-more like that of suspensions than enrolment cancellationswas observed for exclusion incidents, which increased significantly for Indigenous students but not for non-Indigenous. This increase was driven by incidents occurring in the secondary schooling years.
When considering incidents in the year 2019, exclusions were highest for Grade 9 students, irrespective of Indigenous status. Exclusion incident rates followed a similar pattern to suspensions, with these being highest for the category of Physical Misconduct, within which the highest rate of exclusion was evident for Indigenous students. Due to the small number of incidents, however, we were unable to determine the degree of overrepresentation. Regional analyses indicated that non-Indigenous exclusions were similar across regions, while there was some minor variability for Indigenous exclusions based on geographical location.
It is worthwhile at this point to recall that the aim of the 2014 legislative changes in Queensland was to reduce the use of exclusionary discipline-especially enrolment cancellations and exclusions-which had been trending up since 2008 (Graham, 2018). Our analyses indicate the legislative changes have only averted increased enrolment cancellations and exclusions for non-Indigenous students, but that Indigenous students are increasingly both differentially and negatively impacted. Given the aims of the National Agreement on Closing the Gap (Australian Government, 2020), it would appear that empirical research to guide systemic inclusive school reform is urgently needed in QLD, such as that proposed in the SA Inquiry (Graham et al., 2020a). The proposed reform includes legislative change to severely limit suspension length and permissible grounds, introduction of safeguards to protect priority equity groups including the right to appeal short suspensions, and reform of discipline policy and procedures, as well as implementation of comprehensive MTSS to provide high-quality accessible pedagogies and evidence-based supports across all three developmental domains: academic, social-emotional, and behavioural.
The known ill-effects of exclusionary discipline, including its association with early school leaving and juvenile justice involvement (Chu & Ready, 2018;Hemez et al., 2020;Mowen & Brent, 2016;Noltemeyer et al., 2015), makes the case for urgent research to determine why and which Indigenous students are overrepresented, not just in Queensland but in all Australian states and territories. It is critical to determine whether they-like African American students-are sanctioned for behaviours that are more commonly overlooked in non-Indigenous students, and/or whether Indigenous students are subjected to harsher consequences than non-Indigenous students for the same infractions (Skiba et al., 2011). Such findings would suggest the need for empathy, cultural responsiveness, and implicit bias training; however, analyses conducted during the SA Inquiry (Graham et al., 2020a) suggest that this is not all that is needed. The Inquiry final report shows that Indigenous students accounted for just 6.6% of enrolments in SA government schools in 2019, but 15.5% of suspensions and 20.3% of exclusions. However, the availability of disaggregated data enabled finer-grained analyses using five reconstituted groups: The Inquiry team's analysis of 2019 data using these five reconstituted groups found that Group 2 accounted for only 4% of suspensions and 5% of exclusions in 2019. A much larger percentage of suspensions (17%) and exclusions (19%) went to Group 4 (those with 2 or more of these characteristics). Within Group 4, the largest share of suspensions (67%, p. 312) and exclusions (63%, p. 314) were for Aboriginal students with a disability. Analyses such as these are important because they suggest that while Indigenous students are overrepresented in exclusionary discipline, the reasons for this are likely multiple and complex (Avery, 2018). Importantly, the high involvement of disability indicates the need for reforms that go beyond empathy, cultural responsiveness, or implicit bias training, although the international evidence suggests these are highly likely to be necessary. Disability requires teachers to engage in high-quality inclusive practice to enable access to curriculum, instruction, and assessment (Graham, 2020;Graham et al., 2020a), and achieving this takes more than policy change in one area-it takes brave and principled whole-of-system inclusive school reform.

The case for systemic inclusive school reform
As the QLD Department of Education publishes only limited data disaggregated by Indigenous and non-Indigenous status, there were many important questions that we could not investigate. We could, however, trace some important contours of this problem as it presents in Australia's third largest education system, enrolling this nation's second highest number of Indigenous students. We found that Indigenous students are significantly overrepresented in QLD state school suspensions and exclusions, and that their overrepresentation is increasing. This is predominantly driven by increases in the use of short suspensions, particularly by primary schools. First, however, short suspensions in QLD are not "short". They can, in fact, be 2 weeks (10 school days) in duration and, since the 2014 changes to legislation, both cannot be appealed and are considerably longer than the maximum suspension permissible in Chicago Public Schools, which is 3 school days. Second, it is currently impossible to tell, due to the lack of publicly available data, whether there are differences in the lengths of short suspensions for Indigenous students versus non-Indigenous students. For example, it could be that Indigenous students are disproportionately being issued with suspensions closer to the 10-day upper limit than non-Indigenous students. Third, it could also be that Indigenous students are receiving multiple "short" suspensions and, thus, spending many, many weeks outside school with all the negative impacts that absence from school entails.
While overrepresented in every suspension reason category, Indigenous students in QLD are most overrepresented in suspensions for disruptive and disengaged behaviours, which include "absences" (or truancy). These are behaviours for which suspension is banned in many large public-school systems in the United States, such as Los Angeles Unified School District and Chicago Public Schools, because it is well known that exclusionary school discipline both rewards and reinforces school avoidance. This is also where the link between exclusionary school discipline and juvenile justice involvement is clearest (Hemphill et al., 2017). School is a protective factor: it assists in keeping vulnerable young people out of trouble simply by virtue of adult supervision, prosocial peer association, and purposeful activities. Without it, young people are more likely to associate with deviant peers, roam the streets unsupervised, get up to mischief they might otherwise be prevented from doing, and thereby come into contact with the justice system.
Reform in QLD must begin with identifying precisely what Indigenous students are being suspended and excluded for, how long and how often they are being suspended and excluded, and whether these students have a disability or are living in out-of-home care. The present study could not answer these questions due to a lack of publicly available data, but it has succeeded in providing evidence of the need. While not as extensive or as deep an investigation as is necessary, both current and future research may help indicate structural aspects of the QLD government school education system that could be ameliorated through evidence-based systemic inclusive school reform, such as that occurring over the last decade in the United States where the effects of exclusionary discipline and its disproportional impact on historically dispossessed people of colour, who have been "denied opportunities by institutionally sanctioned segregationist policies and practices" (Artiles et al., 2010, p. 281), could no longer be ignored.

Limitations
The present study was subject to certain limitations. In particular, rates of suspension, exclusions, and enrolment cancellations were calculated using publicly available data aggregated at the level of exclusionary discipline incident rather than student. This format restricted the types of analysis that could be conducted and the ability to draw inferences at the student level, as individual children may be included more than once in the incident counts. Accordingly, the present analyses characterise where there have been increases or decreases in the number of incidents occurring over time, rather than indicating whether there have been changes in the number of students being suspended over time. Previous research has delineated students' unique and repeated suspension incidents (e.g., Graham et al., 2020a;Krezmien et al., 2006), which enables insight into demographic and student-level characteristics that place a child at risk of exclusionary discipline. Analysis of both single and repeated incidents also indicates the extent to which the same students are being suspended multiple times, for whom suspension is clearly an ineffective solution.
A related limitation pertains to the relative absence of background demographic information in the data obtained. While it was possible to include certain demographic information in the presented analyses (e.g., Indigenous status, school phase, region), there are many other variables which should be considered when investigating exclusionary discipline, such as socioeconomic status, school size, gender, disability, health and educational outcomes, and whether children are living in outof-home care (Hemphill et al., 2014;Laurens et al., 2021aLaurens et al., , 2021bNoltemeyer et al., 2015;Raffaele Mendez & Knoff, 2003;Sullivan et al., 2013;Welsh & Little, 2018). These factors have substantive and unique associations with school discipline outcomes, and they also interact and influence one another over time in complex ways. The present study indicates a critical need for further research in a range of Australian contexts to examine the longitudinal outcomes of school suspension and exclusion, and its bidirectional associations with a range of student, school, and systemic factors. Given the important historical, cultural, and political differences between Australia and the United States, where most research on disproportionality is produced, effective solutions cannot be derived until the precise contours of the problem-as it exists both in and across diverse Australian regions-are known.
A further limitation of the present work is the lack of Aboriginal or Torres Strait Islander researchers among the investigator team. The National Agreement on Closing the Gap includes priority reforms to support the integration of Aboriginal and Torres Strait Islander peoples and perspectives in researching the drivers of inequality between Australia's Indigenous and non-Indigenous populations, and in developing policy recommendations to support systemic reforms aimed at overcoming these inequalities. This should include ongoing support for training and capacity building initiatives that prioritise Aboriginal or Torres Strait Islander peoples' involvement in the co-design of research questions, data collection and provision, and interpretation and dissemination of research on Indigenous overrepresentation in exclusionary school discipline practices. This will help to ensure that culturally appropriate evidence-based prevention and intervention frameworks can be developed and implemented with fidelity on a system-wide basis.

Conclusion
This study provides important preliminary evidence that the incidence of suspensions, enrolment cancellations, and exclusions all increased disproportionally for Indigenous students in Queensland state government schools during the period 2013-2019. Increases were particularly prominent during the primary school years, and in the use of "short" suspensions of up to 10 days' duration. Further critical research is needed at the individual student level to determine whether Indigenous students are sanctioned for behaviours for which their non-Indigenous peers are not similarly disciplined, and/or whether Indigenous students receive harsher penalties than non-Indigenous students for similar disciplinary infractions, and/or whether Indigenous students are receiving longer suspensions on average, and/or whether they are disproportionately receiving multiple suspensions. Making such data accessible, so that they may be interrogated in partnership with Aboriginal and Torres Strait Islander communities to determine the scale and nature of Indigenous overrepresentation in exclusionary school discipline, is fundamental to deriving effective, evidence-based solutions to overcoming structural disadvantages to accessing education in Australia. Such systemic reform will be critical to the success of the National Agreement on Closing the Gap (Australian Government, 2020) agenda to reduce the school attendance and achievement gaps that have long plagued Australian education systems.

Acknowledgement of Country
The authors acknowledge the Traditional Owners of the lands and waters throughout Australia. In doing so, we pay respect to Elders and families past, present and emerging, and pay reverence to Aboriginal children and young people as they emerge as future leaders. We particularly acknowledge the Jaggera, Turrbal, and Dharug people: the Traditional Owners of the various lands on which this paper was written. We recognise that these lands have always been places of teaching and learning, and that they were never ceded. We pay respect to their Elders-past, present and emergingand acknowledge the important role Aboriginal and Torres Strait Islander people continue to play within our community.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.