Abstract
This Special Issue pays tribute to our dear friend and mentor Anna Stewart, Emeritus Professor at Griffith University, Queensland, Australia, who passed away in April of 2021. Anna was a pioneer in the construction and use of large-scale, linked administrative datasets for criminological research, especially in Australia. She recognised the potential of administrative data to help close critical knowledge gaps and promote evidence-informed government responses. Anna was also deeply committed to mentoring the next generation of scholars and moulding “young” minds – graduating 17 scholars with PhDs and laying the foundation for another three, as well as graduating 28 Honours and Master’s scholars. Many of the guest editors of this Special Issue were these young minds. Anna's enduring commitment to social justice drove her research, bringing together and inspiring others to form cross-disciplinary teams and use their expertise to advance more equitable outcomes across society. We are honoured to continue her work. Our goal in this Special Issue is to highlight Anna's enduring contributions. We begin with a synopsis of her research journey from her PhD through to her final works, highlighting the foundations she laid and visions she had for ongoing work in this area. We then introduce our contributors to this Special Issue, celebrating their collaborations, innovations, and new frontiers that extend beyond Anna's legacy.
Anna Louise Stewart (1955–2021)
Anna was born and raised in Aotearoa New Zealand. She achieved her undergraduate and master's degrees in psychology in her native Aotearoa New Zealand before pursuing a PhD at the University of Queensland. Her thesis, entitled “An investigation of decision making by child protection workers”, marked the start of a career-long interest in reducing harms to vulnerable children, and was grounded in her early, formative work in Aotearoa New Zealand's Department of Social Welfare. After gaining an academic appointment in 1994 at Griffith University's then fledgling School of Justice Administration (later School of Criminology and Criminal Justice), Anna continued to research issues affecting vulnerable women and children, initially in the areas of domestic and family violence (Stewart, 2000; Stewart & Maddren, 1997), and justice responses to sexual assault complaints (Schuller & Stewart, 2000). She also published on criminal justice processes as they were experienced by women and children, whether as victims or offenders (Stewart, 2001; Stewart & Smith, 2004). Perpetrator behaviour was another interest, again in the context of understanding how and why decisions were made to harm vulnerable victims (Knust & Stewart, 2002; McCarthy & Stewart, 1998).
Anna developed an interest in pathways to offending, and particularly the role played by child maltreatment and other harmful experiences common among justice-involved youth (Allard et al., 2010; Dennison & Stewart, 2001; Dennison et al., 2006; Livingston et al., 2008; Stewart et al., 2002, 2008). Her work reflected her interest in the then-evolving developmental and life-course approaches to criminology. She also began exploring another passion, the improvement of system responses to disadvantage, as a way of disrupting pathways to offending. Recognising that the first step was understanding those responses at a system level, she led the development of a micro-simulation model of Queensland's juvenile justice system, intended to demonstrate to policy makers the likely impacts on offending pathways from policy changes, such as the increased use of police cautions as an alternative to arresting children (Livingston et al., 2005, 2006; Stewart, Hayes, et al., 2008).
From this point, Anna's research was focused on justice systems, and how they could be understood, modelled, and ultimately improved to not just reduce offending pathways, but to better the broader life outcomes for vulnerable people. To advance this work, Anna led a series of funded projects aimed at linking and using the everyday administrative records collected by a wide range of government bodies. Her purpose was to use these data to understand how and why people's life trajectories develop, including the intersection of various types of disadvantage, such as child maltreatment, crime, and poor mental health. Bringing together data from diverse agencies raises many sensitivities and complexities, and Anna negotiated these through her tenacity and by building and cultivating strong relationships with people across the public sector, who later became advocates and supporters of her work. Anna's larger than life personality, ability to hold a room, and skill in creating a sense of camaraderie and belonging should not be discounted in her success in building both large datasets and meaningful collaborations. Beyond this, her statistical skills, deep knowledge of the administrative data, and understanding of criminal justice and social services sectors ensured she could translate research to policy and practice. In recognition, she held several advisory appointments, including with the Crime Research Reference Committee in the Queensland Treasury and the National Crime Statistics Advisory Group of the Australian Bureau of Statistics. Her career-long commitment to improving outcomes for women and children also led to her appointment as a member of Queensland's Child Death Case Review Committee from 2010 to 2014.
An enduring outcome of Anna's work was the creation of the Queensland Cross-sector Research Collaboration (QCRC; formerly the Queensland Linkage Project), a population-based linked administrative data repository encompassing all individuals born in Queensland during 1983, 1984, and 1990 who had contact with government systems spanning welfare, health, and criminal justice systems (Stewart et al., 2015, 2021). The guest editors collaborated with Anna in the evolution of this dataset through a series of data linkages over the past 20 years, and national and international collaborations continue to emerge from the use of QCRC data.
These data are held in Griffith University's Social Analytics Laboratory (SAL), a secure research facility custom-built for the purpose of storing, managing, and analysing data of a sensitive nature. Anna was instrumental in establishing the SAL, which also hosts extensive state-wide administrative data holdings from Queensland Police and Queensland Corrective Services systems. Together, these data holdings have produced an impressive volume of research, and attracted eminent researchers from across the world, including several contributors to this Special Issue (e.g., Allard et al., 2014; Allard et al., 2017; Allard et al., 2020; Broidy et al., 2015; Craig et al., 2021; Hurren et al., 2017a; Gilbert et al., 2022a, 2022b; Kuluk et al., 2021; Little et al., 2018; Matthews et al., 2022; McKenzie et al., 2021, 2024; McKenzie et al., 2025; Ogilvie et al., 2021, 2022, 2023, 2024; Thompson et al., 2014).
Administrative data in criminological research: Foundations and early visions
Anna had grand visions for future uses of administrative data in criminological research. She led an article in 2015 advocating for the expanded use of linked administrative data to advance criminological research (Stewart et al., 2015). Here, we highlight the foundations and visions she left us in that article.
Understanding life-course patterns with nuance and balance
Stewart et al. (2015) argued that (linked) administrative data uniquely enable the creation of very large representative longitudinal samples that provide opportunities for (i) exploring life-course offending patterns and associated correlates, potentially for entire population-based cohorts, capturing the full breadth of pathways; (ii) deriving “lifetime” population-based prevalence rates of system contacts, and the accumulation and interaction of system contacts over time, and (iii) examining rare behaviour, offending subtypes, and heterogeneity in timing, patterns, and pathways. Stewart et al. (2015) asserted that this approach could provide a more panoramic view of system contacts and pathways that is not possible in smaller traditional longitudinal designs, as well as a more nuanced understanding of the timing, types, and accumulation of factors associated with variations in these pathways. Such knowledge is essential for theory-building and testing.
De-siloing and reducing inequality
While the limitations of relying on official system contacts are well-known (Payne & Piquero, 2018), the value of understanding individuals’ entry and pathways through multiple systems should not be underestimated. Stewart et al. (2015) illuminated the benefits of understanding longitudinal patterns of contacts across typically siloed government systems. Specifically, examining cross-sectoral contacts provides opportunities to understand when, where, and why systems should collaborate to holistically support vulnerable individuals and families, thereby enhancing service delivery systems and the timing and targeting of interventions. These data are crucial for (potentially quasi-experimental) evaluation of interventions and system responses to ensure efficacious and evidence-based policy and practice to improve outcomes for children, young people, and their families. Moreover, patterns of inequality in system contacts can be highlighted as an impetus and steppingstone for positive change.
Adapting to contemporary and changing contexts
Stewart et al. (2015) argued that, unlike traditional longitudinal designs, multiple cohorts can be derived from administrative data with relative ease to facilitate examination of societal change and legislative/policy impacts. Equivalent studies using traditional longitudinal designs are expensive and take many years to prospectively collect data (Thompson et al., 2023). Since administrative data are routinely collected by agencies, new data sources (as well as cohorts) can be added retrospectively and continuously over time, to explore contemporary patterns, issues, and factors as they emerge (e.g., life-time contacts across criminal justice departments, health, child protection, education, births, deaths, marriages, mental health, and social welfare), as well as the impact of significant policy changes and unexpected events. Furthermore, the ability to create multiple cohorts within and across jurisdictions, and to integrate additional sources of administrative data as they become available and/or relevant, fosters replication and differentiation studies to test and progress theoretical understanding (Stewart et al., 2015).
Expanding (linked) administrative data in criminological research
Stewart et al. (2015, p. 411) noted “these methods are rarely used in criminology and criminal justice research, [but] they offer many benefits” and urged researchers to consider the use of linked administrative data in the pursuit of improved knowledge and practice. In the decade since, there has been a proliferation of criminological research using linked administrative data. Indeed, the value of linked administrative data for complementing traditional longitudinal research is increasingly recognised by field leaders (Laub & Sampson, 2020), as is the role of (particularly linked) administrative data for overcoming key methodological challenges in the field (Malvaso, 2023; Thompson et al., 2023).
Key contributions of the Special Issue papers: From foundations to new frontiers
In this Special Issue, we celebrate new frontiers by showcasing a collection of papers that illustrate the power, challenges, and responsibilities of using administrative data to advance criminological knowledge, including examples that have brought to fruition the visions of Stewart et al. (2015). Many contributors were also Anna's dear friends and colleagues. We know Anna would be delighted with this collection. Below, we briefly review the papers and highlight some of their key contributions to the field, and the unique potential of administrative data. We group the papers into broad themes.
Identifying and understanding inequality in justice system responses
In this Special Issue, Gorton et al. (2025) and Weatherburn et al. (2025) illustrate the value of population-based, linked administrative data for illuminating inequality in justice system responses. Gorton et al. (2025) convened the first linked dataset using Scotland's police and health administrative data. They linked police data to five population health datasets, household size data, and property type data, to explore associations between health vulnerabilities (mental illness and substance use) and police enforcement of Fixed Penalty Notices (FPNs) during the COVID-19 pandemic. This world-first study reveals important justice inequalities, with higher rates of FPNs among people who accessed health services for psychiatric conditions or substance abuse compared to those who did not. Risk appeared to be especially strong for individuals with multiple health conditions. This use of population-based administrative data, and case-control methodology, illustrates justice inequality in the disproportionate impact of the new policing powers on people with health vulnerabilities and has important implications for police responses to persons with mental illness as well as preparedness for future pandemics. The authors illustrate the unique advantages of linked administrative data for uncovering these inequalities and encourage replication of their findings to determine whether police responses to COVID-19 resulted in global justice inequality.
Weatherburn et al. (2025) utilised state-wide linked administrative justice data from New South Wales (NSW), Australia, to explore the potential contribution of racial bias to sentencing outcomes. They investigated differences in imprisonment risk for Aboriginal and non-Aboriginal defendants for assault and theft offences. While most of the gap could be explained by the comprehensive legal factors in their model (Oaxaca-Blinder decomposition), the authors report an inexplicable gap of between 17% (assault) and 23% (theft), which translated into 150 more adult Aboriginal defendants receiving prison sentences for these offences alone over the study period (2019–2023). The results can be explained as the contribution of bias in how courts respond to legal factors for Aboriginal people. Of course, neither Gorton et al. (2025) or Weatherburn et al. (2025) could control for every possible variable that may have influenced their results. Nonetheless, both studies shine a light on inequality in justice system responses to drive change and illustrate the power of linked administrative data for doing so.
Challenges and ethical considerations
While administrative data can be used to highlight inequality in justice system responses, it also has the potential to reinforce systemic bias towards Indigenous peoples. As Gumbaynggirr and Dunghutti scholar Lockwood (2025) argues, such data have been used for the monitoring and control of Indigenous people in Australia and around the world. In particular, Lockwood highlights the centrality of surveillance, objectification, racism, problematising, construction of deficit racial narratives and stereotypes, and settler-colonial institutions, systems, and violence in administrative data. Drawing attention to the historical favouring of positivist, quantitative approaches, at the expense of Indigenous worldviews, experiences, and priorities, Lockwood calls for critical evaluation of methods and methodologies to assess their impact on hyper-incarcerated populations. Building on Indigenous research principles and methodologies, Lockwood proposes a six-part Indigenous data justice framework to guide the use of linked administrative data. While acknowledging the inherent biases and limitations of government data, Lockwood maintains that administrative data and Indigenous data threads can be woven together to support Indigenous justice projects – when Indigenous priorities and community-led research are privileged. Beyond its obvious importance to Australian criminology, the paper raises key ethical and political considerations that are equally important in international settings.
Exploring rare phenomena and distinct subgroups
Several studies in this Special Issue demonstrate the value of linked administrative data for investigating “rare” phenomena and distinct subgroups, and comprehensive depictions of engagement with multiple systems over the life-course, to provide new insights into important theoretical and policy-relevant questions.
White-collar offending
Narvey et al. (2025) demonstrate the capacity of administrative data for providing insight into longitudinal patterns of white-collar offending over the life-course; a nascent area of research. They utilised comprehensive arrest histories compiled from administrative data from Texas to explore the lifetime criminal career patterns of a state-wide cohort of 43,204 individuals released from prison. These authors compared the life-course offending patterns of those who engaged in white-collar offending (N = 7,227) to those who did not, in what is arguably “the most comprehensive longitudinal dataset on white-collar offenders spanning the life-course” (Narvey et al., 2025, p. 593). Narvey et al. demonstrate the important benefit of large, longitudinal administrative data for developmental/life-course analyses of a comparatively rare offence type, uncovering important similarities and differences in criminal careers for those who do and do not engage in white-collar offences. At the same time, they highlight the potential value of future research incorporating both quantitative and qualitative methodologies to further unpack nuances in experiences, and differences in underlying offender characteristics, that are typically not captured in large-scale administrative data holdings. Indeed, most contributors in this Special Issue acknowledge the important role of other data sources to complement administrative data; as did Anna (Stewart et al., 2015).
The links between Child Protection System (CPS) and Criminal Justice System (CJS) contact
Given Anna's passion for better understanding the links between child maltreatment and offending, and improving outcomes for children, it is fitting that four papers in this Special Issue use linked administrative data to explore the CPS–CJS nexus to advance knowledge and inform policy and service provision (Gillespie et al., 2025; Malvaso et al., 2025; Thompson et al., 2025; Tzoumakis et al., 2025).
Gillespie et al. (2025) merged administrative data from Orange County, California, with self-report data to compare recidivism amongst male youths involved only with the juvenile justice system, with those involved with both the juvenile justice and child welfare systems (dual system youths). Their use of combined (administrative and self-report) data sources revealed that, while both groups were equally likely to self-report recidivism, dual system youths were more likely to be detected for this recidivism. This paper demonstrates the value of combined data sources in distinguishing between behaviour and the detection of that behaviour within systems – important factors when considering both offending pathways and social justice. Gillespie et al. advocate for programmes aimed at decreasing re-arrest among dual system youths, particularly considering the negative consequences of criminal justice processing.
Tzoumakis et al. (2025) used the NSW Child Development Study data to describe population-based patterns of CJS contacts in adolescence among individuals placed in Out-Of-Home Care (OOHC) in childhood (before age 10), compared to those who were not. Their findings revealed that OOHC was associated with a higher prevalence of all types of adolescent CJS contacts examined (including both perpetration and victimisation); however, most individuals in OOHC did not have court contacts or periods of custody. The patterns were comparable across sex and different OOHC types. Tzoumakis et al. highlight that much research on the link between CJS involvement and OOHC utilises high-risk samples that do not provide a panoramic view of the OOHC–CJS nexus and associated nuances. They emphasise the advantages of large, linked administrative data for exploring relatively rare phenomenon such as OOHC contact across both a range of CJS outcomes and sex, and for providing population-based patterns to inform policy and practice.
Thompson et al. (2025) used the QCRC data holdings created by Anna and her colleagues to explore the high-risk features of CPS contacts on CJS involvement. Using state-wide linked administrative data on all individuals registered as born in Queensland in 1990, this study provided a rare opportunity to simultaneously explore a range of CJS outcomes through to age 24, with consideration of both lifetime mental health contact and CPS dimensions (including the timing and type of maltreatment and OOHC). Extending on Anna's earlier work in this space (e.g., Stewart, Livingston, et al., 2008), their findings revealed that many CPS-involved individuals (and those who experience OOHC) remain free of CJS involvement (particularly in relation to more serious or frequent contact) (echoing some of the findings from Tzoumakis et al., 2025). However, developmental timing and the nature of CPS contact were important. More serious CJS contact was associated with adolescent-only and persistent maltreatment, neglect, OOHC experienced after age 10, and comorbid CPS and mental health contact.
Together, the findings of Tzoumakis et al. and Thompson et al. support trauma-informed responses for those working in the CPS and CJS sectors and cross-sector holistic care for maltreated individuals. They also illustrate the value of large representative population-based longitudinal linked administrative datasets for delineating variations in patterns of CJS contact across population subgroups. This is often challenging when using traditional methodologies due to the requisite analyses being statistically underpowered.
Informing early intervention: Timing and targets
Malvaso et al. (2025) also examined the OOHC–CJS link; however, they take a particularly novel approach by exploring the predictive capacity of linked administrative data for targeting early developmental crime prevention efforts to those who need it most. They used population-level linked administrative data from the South Australian Better Evidence Better Outcomes Linked Data (BEBOLD) platform to determine risk of youth justice supervision among young people who experienced early OOHC (<10 years of age). Utilising a large number of “predictors” spanning child and maternal sociodemographic factors, perinatal factors, and child protection characteristics, they illustrate the “technical” potential of linked administrative data for developing risk prediction models to identify and guide intervention opportunities for children in OOHC who are at higher risk of transitioning to the justice system. However, they also carefully canvas the ethical and practical considerations associated with such an approach and call for greater research critically investigating the feasibility and potential use of this approach before linked administrative data-based risk prediction models are applied in practice.
Replication and differentiation studies to advance theory and explore social change
Payne et al. (2018) identified what looked to be a major decline in youth offending between two cohorts (1984 and 1994), with the associated explanation challenged by Anna at a conference where the findings were presented. In this Special Issue, Payne (2025) used state-wide linked administrative data from police and courts in New South Wales, Australia, to produce a replication of Payne et al. (2018), exploring differences in longitudinal criminal trajectories between the two birth cohorts. Utilising an improved data-matching process, a longer observation period, and extended data sources, Payne identifies fundamental variations in conclusions from the replication study compared to the original study, with important implications for understanding the crime decline. This study emphasises the importance of replication studies and the value of linked administrative data for conducting such studies, whereby new sources (and greater lengths) of data can be added to validate, refine, and/or expand theory and knowledge. It also reminds us that data linkage methodologies and data sources can meaningfully influence results and, therefore, careful scrutiny of these processes and decisions during interpretation of findings is essential (also see Shaw et al., 2022).
Payne also demonstrates the power of linked administrative data for comparing multiple cohorts across distinct periods of time and examining the influence of macrolevel social change on offending patterns; an approach that is typically unfeasible using traditional longitudinal designs (Hurren et al., 2017b; Stewart et al., 2015). The potential of such cohort comparisons for exploring macrolevel change and ensuring theory transcends the changing social world is obvious (Thompson et al., 2023). Indeed, Laub and Sampson (2020) claim that understanding macrolevel change is one of the “biggest frontiers” for developmental and life-course research. Linked administrative data can play a significant role in this pursuit. As a final minor point, while we think Anna would have found this replication study particularly interesting, we suspect she would have especially enjoyed the fact that Payne admitted in writing that she was right (see Payne, 2025)!
Conclusion
Taken together, the contributors to this Special Issue demonstrate new frontiers in criminological research using administrative data. Their work draws attention to the value of (linked) administrative data for (i) identifying and understanding inequality in system contact and responses to drive more equitable outcomes across society, (ii) exploring rare events, subgroups, and developmental sequences to examine nuances to advance theory, policy, and prevention, (iii) informing cross-sector collaboration and the timing and targets for intervention to better support those who need it most, and (iv) conducting cohort-comparisons and replication studies that facilitate validation, refinement, and expansion of theory and knowledge to ensure theories remain fit for contemporary purposes. Finally, these papers acknowledge, as Anna did, that linked administrative data can complement, but not replace, traditional methodologies. Administrative data are an important tool, but not the only tool, to tackle social problems and drive practice to achieve a more equitable and just society.
The positive contribution of administrative data to criminological research relies upon responsible and ethical use by researchers. As argued by Lockwood (2025, p. 582) in her discussion of administrative data pertaining to First Nations peoples in Australia, “the strength of the linked administrative data thread depends on the Indigenous-led threads and governance structures that hold it together”. Global use of administrative data for criminological research must be guided by the lessons learned from prior and current harms and errors (alongside successes), acceptance of what these data can and cannot contribute, and incorporation of diverse worldviews, methods and methodologies.
As Anna's former friends and colleagues, we benefited from her leadership, vision, and mentorship, and from her abiding belief that criminological research should aim to make a difference to the lives of those who need it most. We have not only been inspired by her work, but also by her care for others. We sincerely thank all contributors to this Special Issue and praise their innovation and dedication. We will continue to use administrative data to inform our criminological research, always striving to increase knowledge, support practice, and improve life outcomes for vulnerable people. It is our hope that this Special Issue will inspire others to do the same.
Footnotes
ORCID iDs
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.