Police and procedural justice: perceptions of young people with mental illness

ABSTRACT

Young people with mental illness are significantly more likely to encounter the police than their counterparts who do not identify as having a mental illness. Yet little is known regarding how this cohort perceives the police and whether they believe the police to be a fair, trustworthy, and legitimate service. Research suggests that young people and other vulnerable groups (such as adults with mental illness) value procedural justice policing as a technique for nurturing fair and trustworthy policing, which in turn, increases satisfaction with police interactions and willingness to cooperate with police. This study uses procedural justice as a lens for analysing the perceptions of young people with mental illnesses regarding the police. Drawing upon survey data from a sample of 3147 Australian participants aged between 14 and 25 years old – a third of which identified as having a mental health condition – results demonstrate that young people with a mental illness offered significantly lower perceptions of the police in relation to procedural justice. Young people identifying as not heterosexual or as trans* or gender diverse, and those who report that they are not seen as Australian, also offered significantly lower perceptions of police procedural justice. A theoretical explanation is offered for why these marginalised young groups perceive the police to be procedurally unjust. Tactics for how the police may nurture more trusting and supporting relationships with young people in general are also discussed.

KEYWORDS:

• Policing
• Youth justice
• Mental illness
• Procedural justice

Introduction

Young people with mental illness, trust, and negative perceptions of the police

Public trust and confidence in the police is essential for predicting public cooperation with law enforcement in a democratic society (Miles-Johnson 2013a). Public cooperation and collaboration with the police is important for the effective control of crime and disorder, especially since the police rely on citizens to report crime and suspicious activities in their communities as well as assisting in the identification of offenders (Murphy 2015). Tyler (2006) argues that public expressions of trust in the police are shaped by the belief that the police are sincerely motivated to help and care for citizens, thus prioritising the best interests of the community. Yet minority and diverse members of society, such as young people with mental illness and young persons in general, typically mistrust the police more than adults, often fearing that the police will not take their concerns seriously and that they may incur abuse from police (Piquero et al. 2005, Miles-Johnson 2013a, Murphy 2015). Mistrusting perceptions of the police by diverse groups are often shaped by first-hand and/or second-hand (vicarious) negative experiences with the police (Watson et al. 2008). When minority and diverse groups hold negative and mistrusting perceptions of police which question perceptions of police legitimacy, they may purposively avoid contact with the police and become reluctant to cooperate with police (Miles-Johnson 2013a).

Young people with mental illness and police contact

Public perceptions of the police are also important since such perceptions may influence the way in which the public and the police interact with one another. These perceptions of the police are especially important for young people with emerging mental health problems, given they are significantly more likely to have early police contact than the general adolescent population (Fazel et al. 2008, Dean et al. 2021). Young people with mental illness are more likely than their peers to be a witness to or a victim of crime (Dean et al. 2021), to be arrested for misdemeanours (Wasserman et al. 2002), be detained in correctional facilities (Fazel et al. 2008), and, unique to this cohort, to be detained under mental health legislation for suffering a mental health crisis (Doulas and Lurigio 2010). Yet early police contact is associated with long-term difficulties for young people, negatively impacting their physical and mental health, their school and occupational outcomes, and their risk of future involvement with the criminal justice system (Craig et al. 2017, Rivenbark et al. 2018, Dean et al. 2021). Combining the negative impacts of early police contact with the increased likelihood of police contact for young people with mental illness, not only diminishes trust between this cohort and the police but also has the potential to leave long lasting psychological scars that many vulnerable young people will never recover from (Asquith and Bartkowiak-Theron 2012).

Prevalence of mental illness amongst young people, lack of resources, and police acting as first responders

Encounters of young people affected by mental illness with police are especially problematic considering mental illness amongst Australian youth is more prevalent than in the general population. Preliminary data from The National Survey of Mental Health and Wellbeing conducted in 2020–2021 demonstrated that rates of mental illness were highest amongst young people aged 16–24. Approximately 40% of 16–24 year old’s experienced a mental disorder compared with approximately 20% of Australians of all age groups combined (16–85) in 2020–2021 (Australian Bureau of Statistics 2022). Despite the prevalence of mental illness and psychological distress amongst Australia’s youth, the last national inquiry examining the state of Australian mental health services revealed major service gaps in the provision of mental health community care for all ages, particularly in relation to appropriately tailored services for children and adolescents who have unique and complex emotional, social, and physical needs (Senate Select Committee on Mental Health 2006).

In Australia and across the Western world, there are also major gaps in inpatient facilities for adolescents with severe mental illnesses where parents, guardians, and young people with mental illness struggle to navigate access to a piecemeal system of underfunded state and expensive private psychiatric beds (Patel et al. 2007). In Queensland, Australia, the decision to close the Barrett Adolescent Centre (an inpatient public mental health facility for adolescents) came under intense public scrutiny since its closure was linked with the tragic suicides of three former inpatients and exposed the state’s vastly under-resourced inpatient mental health system for adolescents (McLeish 2015). The lack of appropriate community mental health services for all ages of people suffering mental illness has also been linked to increased police contact for people who have unmet needs for assistance in the community (Gooding 2016). The process of deinstitutionalisation of mental health services, across Australia and globally, has over-burdened the police acting as first-responders to individuals suffering mental health crises in the community and increased the need for police response (Carroll 2005, Kruger 2020).

Criminalising and degrading police treatment of young people with mental illness and a lack of confidence in the police

Police acting as first responders to mental health crises in the community has contributed to the overrepresentation of young people with mental illness in the criminal justice system. Against a backdrop of deficient alternative community resources, police act as gatekeepers to the youth justice system (Richards and Ellem 2019), and often feel compelled to arrest people with mental illness (of all ages) when confronted with the unfamiliar idiosyncrasies of a mental health crisis (Baksheev et al. 2012). Whilst research concerning police treatment of adults with mental illness has been growing in recent years (Clifford 2010, Boscarato et al. 2014, Evangelista et al. 2016, Furness et al. 2016, Bradbury et al. 2017, Morgan 2021, Morgan and Miles-Johnson 2022), research on the perceptions of young people with mental illness regarding the police is comparatively rare. Research consistently states that adults with mental illness routinely feel criminalised and degraded by the treatment they receive from police when officers respond inappropriately to their needs (Boscarato et al. 2014, Brennan et al. 2016, Bradbury et al. 2017).

Limited research from the USA also suggests that young people with mental illness are often criminalised when encountering the police, with police lacking the necessary training and resources to effectively interact with this cohort and divert them away from the criminal justice system (Rich 2009). Young people’s mental health crises are often exacerbated by punitive law enforcement responses, which can lead to potentially volatile situations when engaging with the police (Doulas and Lurigio 2010). As such, evidence from the USA suggests that police contact with young people with mental illness – especially exposure to police violence – is detrimental to the mental health of this cohort, and contributes to a lack of confidence in police amongst adolescents (Choi et al. 2021).

Procedural justice

Given the increased likelihood of police contact with young people with mental illness and the adverse impacts of early police contact, the perceptions of young people with mental illness of the police are important in understanding the quality of such interactions. Such perceptions of the police are important since they may influence both the way young people with mental illness and the police interact with one another, and whether this vulnerable cohort are willing to contact police or will seek to avoid them. Since police act as decision makers at street level when interacting with citizens, many scholars have used procedural justice as a lens through which to assess public perceptions of the police, specifically in relation to public perceptions of police legitimacy and police fairness when police interact with citizens (Murphy and Tyler 2017). In the broadest sense, procedural justice describes people’s perceptions of the treatment they receive (or others receive) during processes involving decision making, specifically regarding whether treatment received during proceedings is fair and just (Thibaut and Walker 1975, Lind and Tyler 1988).

Procedurally just policing is generally measured by the presence (or absence) of police committing to sincerely emphasising four key principles during engagement with citizens – trust, dignity and respect, neutrality, and voice (Goodman-Delahunty 2010). Trust refers to the level of belief the public has of an authority’s concern for the well-being and interests of citizens and whether police are willing and sincerely motivated to help, listen, and consider the views of the public. Dignity and respect refer to the behaviour of the police and whether it is ‘professional’ in terms of protecting the rights and dignity of citizens such as treating citizens with politeness and courtesy and demonstrating to citizens that they are valuable and that their concerns are taken seriously. Neutrality is the absence of bias within police procedures through transparency, consistency, and even-handedness by using principled police conduct that can be easily explained to members of the public and justified by rules and not by personal opinions. Voice refers to the value of being listened to by the police and the level of community participation afforded in police decision-making processes, thus giving citizens a social standing in the community through demonstrating that police value the public and their opinions (Goodman-Delahunty 2010).

Outcomes of procedural justice policing

Procedural justice policing is considered important in police-citizen interactions based on several theorised outcomes for both police and citizens, such as enhancing: (1) public perceptions of police legitimacy (the recognised right to rule) (2) fair treatment of citizens; (3) public trust and confidence in police; (4) supportive and cooperative relationships between the police and public; (5) public satisfaction with police encounters; and (6) in some cases voluntary compliance with the law and police directives (Mazerolle et al. 2014, Murphy and Tyler 2017). Despite limitations with measuring the effects of procedural justice (see Nagin and Telep 2017), the theorised outcomes of procedurally just policing have enjoyed significant empirical support since its inception (Hinds and Murphy 2007, Murphy et al. 2008, Murphy 2009, Mazerolle et al. 2012, 2013a, 2013b, Higginson and Mazerolle 2014).

Procedural justice and vulnerability

Most importantly for the present research, procedural justice policing has been found to be particularly beneficial for vulnerable persons who encounter the police, such as persons with mental illness (PWMI) and young people. Research from the USA consistently demonstrates that PWMI value procedural justice policing, potentially even more than the general public (McCluskey 2003, Watson and Angell 2007, 2013, Watson et al. 2008, 2010, Wood and Watson 2017). These studies contend that due to PWMI having a tenuous status in society (such as identifying with a vulnerable and marginalised status), these vulnerable individuals pay the most attention to fair police treatment due to expectations of being exploited by police. Not only do PWMI value procedurally just policing more highly, but they also have a better emotional response, feel less coerced, offer less resistance, and cooperate more with police when police engage using procedural justice (Watson and Angell 2007). Conversely, research has shown that vulnerable individuals are also more likely (than the general public) to rebel against police treatment that is disrespectful, forceful, and procedurally unjust (McCluskey 2003, Watson et al. 2008).

Self-uncertainty and the value of procedural justice

Central to this argument is the role of ‘self-uncertainty’, such as when marginalised individuals feel uncertain about their societal status (for example, people who feel uncertainty and vulnerable due to identifying as ethnic minorities, immigrants, young people, victims of crime, or having a mental illness). Self-uncertainty has been shown to increase the value of procedural justice policing, since self-uncertain individuals may respond more favourably to procedurally just police behaviour due to identifying as a marginalised and vulnerable individual when interacting with the police (Murphy and Barkworth 2014, Murphy 2015, Wolfe et al. 2016, Murphy et al. 2018, Murphy and Mazerolle 2018). Adolescents have been found to possess greater levels of self-uncertainty than adults (Meilman 1979, Waterman 1982), and there is a growing body of research to suggest that self-uncertain young people may value procedurally just policing even more than adult populations. Murphy (2015) surveyed 513 adolescents and 2611 adults in Australia and found that self-uncertain adolescents valued procedural justice more than adult populations, and as such were more likely to view the police as more legitimate and cooperate with police when police were considered procedurally just. Other studies in Australia (Hinds 2007, 2009), the USA (Fagan and Tyler 2005, Piquero et al. 2005), and Jamaica (Reisig and Lloyd 2009) have also found positive associations between adolescent perceptions of procedurally just police behaviour and perceptions of police legitimacy. However, although these studies in the youth literature demonstrate positive associations between procedural justice and perceptions of police legitimacy, they often fail to show unequivocal associations between positive perceptions of police legitimacy and willingness to cooperate with the police.

The current study

The theorised pathway from procedural justice, to police legitimacy, and to compliance and cooperation with the police requires further attention in both the adult and youth literature (Murphy 2015, Nagin and Telep 2017). What remains more clear is that when police value self-uncertain individuals by using procedural justice in police-citizen interactions, in turn, it has the potential to increase a positive sense of self for the vulnerable individual (Murphy 2015). Yet there remains a dearth of literature that assesses perceptions of young people affected by mental illness with police and whether police value and fairly treat this cohort. Understanding this phenomenon becomes increasingly complex when considering the overlapping marginalised identities (intersectionality) young people often possess, such as adolescence, mental illness, sexuality, ethnicity, and religion. Intersectionality is an analytical approach that refers to how multiple aspects of an individuals’ identity can lead to overlapping forms of discrimination (Vickery 2018). As such, this research draws upon data collected from the first wave of the Australian Youth Safety Survey (Higginson and Morgan 2020) to assess young people with mental illness’s perceptions of police treatment, whilst also taking into consideration other vulnerable identities adolescents possess and how this may further affect perceptions of police. A procedural justice lens has been applied to the data collection and analysis to determine perceptions of fair police treatment by young people with mental illness.

Methods

Data

Data for this paper were drawn from wave 1 of the Australian Youth Safety Survey (AYSS); an online survey of young Australians aged 14–25, focusing on their experiences of identity, diversity, and safety (Higginson and Morgan 2020). Wave 1 of the AYSS was collected between 28 October 2019 and 23 January 2020.¹ Recruitment was via a paid Facebook advertising campaign, as well as social media sharing of the advertisements. As an incentive, participants were offered the chance to enter a prize draw to win one of thirty $50 gift vouchers. Data were collected in a non-identifiable format using the EFS Survey (Questback 2019) and only surveys where the participant selected ‘Yes, SUBMIT all my answers’ were retained for analysis. Detailed information on recruitment can be found in the AYSS Technical Report (Higginson and Morgan 2020). As can be seen from the distribution of missing values shown in Tables 1 and 2, the AYSS data has a very low rate of missing data on individual items, with the highest number of responses missing on items that required the respondent’s postcode.

Table 1. Descriptive statistics, continuous variables. Higher values indicate higher levels of the variable, except for Relative Socio-economic Disadvantage decile, where higher values indicate areas of less relative disadvantage. For further information on variable construction, see Appendix 1.

Variable	Mean	SD	Min	Max	Median	% missing
Perception of procedural justice policing for youth (PJPY)	2.336	0.752	1	4	2.25	0.89%
Age	18.42	3.139	14	25	18	0%
Religiosity	2.198	1.582	1	6	1	0.54%
Respect for authority	2.435	0.842	1	4	2	0.16%
Parental emotional support	3.536	1.238	1	5	4	0.44%
Self-control	2.890	0.520	1	4	2.889	0.13%
Victimisation	1.819	1.493	0	7	2	0.03%
Delinquent peers	1.216	1.038	0	5	1	0.22%
Delinquency	0.890	1.505	0	13	0	0.16%
Relative wealth	3.670	1.530	1	7	4	0.10%
Neighbourhood disorder	1.635	0.682	1	4	1.4	0.13%
Neighbourhood social cohesion & trust	2.422	0.642	1	4	2.333	0.13%
Neighbourhood attitude to diversity	3.285	0.664	1	4	3.333	0.41%
Relative Socio-economic Disadvantage decile	6.502	2.728	1	10	7	3.53%

Table 2. Descriptive statistics, categorical variables.

Variable	Response category	%	% missing
Mental health condition	Yes No	33.96% 66.04%	1.27%
Gender	Cis female Cis male Trans* & gender diverse	40.42% 51.89% 7.69%	0.19%
Heterosexual	Yes No	48.83% 51.17%	0.35%
Perceived as Australian	Yes No	72.70% 27.30%	0.25%
Other disability	Yes No	23.18% 76.82%	1.27%
Drug use	Yes No	34.41% 65.59%	0.03%
Police contact	Yes No	7.66% 92.34%	0.22%
Student	Yes No	75.24% 24.76%	0.10%
Employed	Yes No	50.145 49.86%	0.10%
Looking for work	Yes No	19.44% 80.56%	0.10%
Remoteness	Major Cities of Australia Inner Regional Australia Outer Regional Australia Remote Australia Very Remote Australia	70.74% 20.33% 7.93% 0.76% 0.24%	3.50%
State	Queensland New South Wales Victoria Tasmania South Australia Western Australia ACT Northern Territory	29.01% 24.91% 21.07% 3.78% 7.18% 10.74% 2.67% 0.64%	0%

As is common to survey research, the AYSS participants opted in to complete the survey, and hence there is inevitably the possibility of selection bias which may limit the generalisability of findings. The paid social media advertising via Facebook was designed as an attempt to reach the largest and most diverse pool of potential participants, as research suggests that Facebook recruitment can provide better representation than traditional methods, particularly with young and difficult to access populations (Whittaker et al. 2017). Overall, the advertisements were viewed a total of 3,181,909 times (‘impressions’) and at least one advertisement was seen at least once by 538,112 unique individuals (‘reach’), which equates to 14% of the target population in Australia. The survey link was clicked on 17,265 times, and a total of 3147 young people (18% of those who clicked through) consented to participate and submitted their responses.

A small number of variables can be compared to official data to assess the degree of representativeness of the AYSS sample. Based on intercensal population estimates (ABS 2020a)², the AYSS was completed by 1 in every 1245 young Australians aged 14–25, although this proportion does vary across Australian states and territories from a low of 1 in 1923 young people in the Northern Territory, to a high of 1 in 638 in Tasmania. The geographic distribution of the sample across remoteness categories is broadly representative of the 2016 Census data³, with a slight over-representation in regional areas: 70% of AYSS participants live in a major city compared to 74% of 14–25-year-olds in the general population, 29% live in a regional area (cf. 24%), and 1% live in a remote or very remote area (cf. 2%) (ABS 2020b). Participant distribution across areas in terms of socio-economic status was assessed against the ABS Socio-Economic Indexes for Areas (SEIFA). AYSS participants are evenly distributed across areas using the SEIFA Index of Economic Resources but are somewhat over-represented in more advantaged neighbourhoods using the SEIFA Index of Education and Occupation and the SEIFA Index of Relative Socio-economic Disadvantage (Higginson and Morgan 2020).

Approximately half the participants were assigned female at birth (49%) and half assigned male (50%), with four intersex participants and three participants who selected ‘something else’ (agender, nonbinary). The AYSS sample over-represents younger participants: 55% of the sample was aged 15–18, which is significantly higher than the comparable proportion (30%) from official population data (ABS 2020a), hence the results of this study may be somewhat more generalisable to adolescents than to 19–24 year olds. The AYSS somewhat over-represents Australian-born participants. 88% of participants were born in Australia, compared to 76% of the general population, and 12% are first generation immigrant, compared to 19% (ABS 2023). The large sample size of the AYSS provides a narrow margin of error for estimates: at a 95% confidence level, the margin of error for estimates of 50% is calculated as ±1.75% (Higginson and Morgan 2020).

Analysis

This study reports the results of ordinary least squares regression analyses with robust standard errors, conducted on unweighted complete case data (N = 2912, 93%).

Key measures

The dependent variable is young peoples’ perceptions of procedural justice policing as it relates to youth. The AYSS included a series of four items measuring the participants’ perceptions of how often police treat young people with respect, make fair decisions when dealing with young people, explain their decisions and actions to young people, and listen to young people. The variable perceptions of procedural justice policing for youth (PJPY) was calculated as the mean of these four items, measured on a four-point Likert scale: Almost never (1), Sometimes (2), Often (3), Almost always (4). The index was not calculated for participants with missing data on any of the component items, and the resulting index had excellent internal consistency (Cronbach’s α = 0.90). Appendix 1 provides full details of variable construction, and Tables 1 and 2 provide descriptive statistics for all variables in the model.

The independent variable of interest is mental health condition, measured as a self-reported disability. The AYSS asked participants whether they had a disability, and participants were able to answer Yes or No to seven categories of disability, including a mental health condition. The item did not require that the mental health condition had been diagnosed by a medical practitioner or psychologist. This was largely in recognition of the many socio-economic and structural barriers for young people to receive a medical diagnosis of a mental health condition, and because the study aimed to give primacy to the young person’s perspective on their own lives. This relatively simple self-report item can be contrasted with the National Study of Mental Health and Wellbeing survey (ABS 2022) that used trained interviewers to conduct the World Mental Health Survey Initiative version of the World Health Organization's (WHO) Composite International Diagnostic Interview, version 3.0, with items based on the DSM-IV and ICD-10. This process enabled the diagnosis of participants with a range of 12 mental disorders across the categories of anxiety disorders, affective disorders, and substance use disorders (ABS 2022). Despite the different methodologies, the AYSS and the ABS study came to very similar estimates, where 34% (margin of error = ± 1.7%) of AYSS participants reported that they had a mental health condition, and the more clinically diagnostic ABS study identified 40% (margin of error = ± 4.8%) of 16–24 year old participants as having a mental disorder in the previous 12 months (Australian Bureau of Statistics 2022).

Control variables

To account for other intersectional vulnerabilities that may increase young people’s perceptions of uncertainty, and in turn may promote lower perceptions of procedural justice policing, the model controls for a number of key identity characteristics associated with marginalisation. Gender is coded from sex assigned at birth and an open-ended item asking about current gender identity. Participants are categorised as either cis-gendered male, cis-gendered female, or trans* or gender diverse. Sexuality is coded as either heterosexual or not heterosexual. These two variables are included in the model as research indicates that people with marginalised gender and/or sexuality identities are more likely to be distrusting of police (Miles-Johnson 2013a, 2013b). Research also indicates that ethnically minoritized or racialized people have more strained relationships with police and can have lower perceptions of police procedural justice (Dandurand et al. 2022). As a measure of racialisation, we include a variable that stands as a proxy for the ‘otherness’ of diverse ethnic identity – participants reporting that they are perceived by others as Australian or as some other ethnicity that is not Australian. We recognise that there will be degrees of exclusion captured within the category of ‘not Australian’ – for example, being perceived primarily as North American may be less marginalising than being perceived primarily as Asian or African – and this lack of granularity may be a limitation of the measure. Although the AYSS also measured country of birth, ethnic identity, language spoken at home, and whether the participant considers themselves to be white, this measure was chosen as it most strongly speaks to the participant’s perception of their own inclusion or exclusion along racialized lines. The model also controls for the presence of any disability other than a mental health condition, as police interactions with persons with disabilities other than mental health are frequently characterised by discriminatory attitudes and practices (Dowse et al. 2021) which could reasonably be assumed to impact perceptions of procedural justice.

The model accounts for individual characteristics that are associated with pro-social values and more positive perceptions of police, by controlling for age, religiosity, respect for authority, parental emotional support, and self-control (Ferdik et al. 2013, Akinlabi 2015, Pickett et al. 2018). The model includes experiential measures, including personal and vicarious experiences of crime and policing, that may influence perceptions of procedural justice policing. These control variables include measures of personal victimisation and delinquency, drug use, peer delinquency, and previous police contact (Akinlabi 2015, Bradford and Jackson 2016, Leslie et al. 2017, Pickett et al. 2018).

Measures of socio-economic status were included both at the participant level and at their neighbourhood level. Individual measures of socio-economic status include a self-reported measure of relative wealth, and whether the participant was a student, employed, or looking for work (Gau et al. 2012, Ferdik et al. 2013, Pickett et al. 2018). Neighbourhood socio-economic status was measured as the decile of relative socio-economic disadvantage (Australian Bureau of Statistics (ABS) 2016), matched to the participants’ postcode. Three more neighbourhood-level variables were included in the model to control for the effect of pro-social local communities on perceptions of police (Gau et al. 2012, Bradford and Jackson 2016, Weisburd et al. 2022). These measures comprise indexes of neighbourhood disorder, neighbourhood social cohesion and trust, and neighbourhood attitude to diversity.

Participants’ state or territory of residence was included in the model, as each Australian state or territory has its own police service. Policing of young people across Australia is broadly consistent, in that all jurisdictions aim to align with the National Youth Policing Model (Commonwealth of Australia 2010). However, we note that the states or territories differ at times in their priorities, practices, and structures of youth justice (Cunneen et al. 2015) and so we include state or territory as a measure as a proxy for jurisdictional differences in policy and practice. Finally, geographic remoteness was included in the model to account for any geographic variation in policing practice due to distance from major metropolitan centres. There are three levels of geography included in this model – state or territory, remoteness, and neighbourhood. However, although neighbourhood nests within state or territory, remoteness is measured by the distance to major metropolitan facilities. In this sense, it is not a strictly nested geographic model, and due to the wide dispersion of participants across the country the model did not need to be adjusted for clustering effects.

Results

Table 3 shows the results of the regression analysis and presents both unstandardised and standardised regression coefficients. We set an alpha level of .05 for all statistical tests. Overall, the model significantly predicted young people’s perceptions of procedural justice for youth and explained approximately 23% of the variance in the outcome (F(35, 2876) = 26.88, p < .001, AdjR² = .23).

Table 3. Results from OLS regression model predicting perceptions of procedural justice policing for youth.

	b	β		se	t	p
Mental health condition	−0.066	−0.042	*	0.029	−2.274	0.023
Gender (ref: Cis male)
Cis female	−0.094	−0.061	**	0.029	−3.276	0.001
Trans* & gender diverse	−0.189	−0.067	***	0.050	−3.788	0.000
Not heterosexual	−0.111	−0.074	***	0.029	−3.857	0.000
Not seen as Australian	−0.064	−0.038	*	0.028	−2.261	0.024
Other disability	−0.003	−0.002		0.032	−0.105	0.916
Age	−0.011	−0.045	*	0.005	−2.133	0.033
Religiosity	0.015	0.032		0.008	1.850	0.064
Respect for authority	0.226	0.252	***	0.017	13.168	0.000
Parental emotional support	0.031	0.050	**	0.011	2.819	0.005
Self-control	0.070	0.049	*	0.028	2.531	0.011
Victimisation	−0.023	−0.045	*	0.010	−2.300	0.022
Delinquent peers	−0.040	−0.056	**	0.014	−2.863	0.004
Any drug use	−0.113	−0.071	***	0.031	−3.647	0.000
Delinquency	−0.006	−0.012		0.010	−0.603	0.547
Police contact	−0.011	−0.004		0.054	−0.201	0.841
Relative wealth	0.023	0.046	*	0.009	2.455	0.014
Student	−0.002	−0.001		0.034	−0.059	0.953
Employed	0.042	0.028		0.028	1.520	0.129
Looking for work	0.042	0.022		0.032	1.292	0.197
Neighbourhood disorder	−0.053	−0.048	*	0.021	−2.447	0.014
Neighbourhood social cohesion & trust	0.100	0.086	***	0.022	4.623	0.000
Neighbourhood attitude to diversity	0.099	0.087	***	0.020	4.997	0.000
Relative Socio-economic Disadvantage decile	−0.009	−0.034		0.006	−1.637	0.102
Remoteness (ref: Major Cities of Australia)
Inner Regional Australia	0.024	0.013		0.038	0.630	0.529
Outer Regional Australia	−0.036	−0.013		0.051	−0.715	0.475
Remote Australia	0.260	0.030	*	0.127	2.045	0.041
Very Remote Australia	−0.004	0.000		0.334	−0.013	0.990
State (ref: Queensland)
New South Wales	−0.002	−0.001		0.034	−0.072	0.943
Victoria	0.134	0.073	***	0.036	3.732	0.000
Tasmania	0.146	0.037	*	0.067	2.162	0.031
South Australia	0.109	0.037	*	0.052	2.109	0.035
Western Australia	0.016	0.007		0.043	0.377	0.706
ACT	0.115	0.025		0.081	1.422	0.155
Northern Territory	0.101	0.010		0.200	0.503	0.615
Constant	1.166		***	0.182	6.421	0.000
N	2912
F(35, 2876)	26.877	***
R^2	0.242
Adjusted R^2	0.232

* p < 0.05 ** p < 0.01 *** p < 0.001.

The key independent variable of interest – mental health condition – was associated with significantly lower perceptions of procedural justice policing for youth (PJPY)(β = −0.04, t = −2.27, p = .02). Three out of the four other key identity markers associated with marginalisation were significantly associated with the outcome.⁴ Both cisgendered female (β = −0.06, t = −3.28, p = .001) and trans* and gender diverse participants (β = −0.07, t = −3.79, p < .001) reported significantly lower perceptions of PJPY than cisgendered male participants. Participants who were not heterosexual (β = −0.074, t = −3.86, p < .001) or were not perceived by others as Australian (β = −0.038, t = −2.26, p = .02) also reported significantly lower perceptions of PJPY. There was no significant association between other disabilities and perceptions of PJPY.

Older participants reported significantly lower perceptions of PJPY (β = −0.45, t = −2.13, p = .03), while respect for authority (β = 0.25, t = 13.17, p < .001), availability of parental emotional support (β = 0.05, t = 2.82, p = .005), and self-control (β = 0.05, t = 2.53, p = .01) were each associated with significantly higher perceptions of PJPY. There was no significant relationship between religiosity and perceptions of PJPY. Only three out of the five measures that controlled for experiences of crime or policing were significantly associated with perceptions of PJPY. Higher lifetime victimisation variety (β = −0.05, t = −2.30, p = .02), higher peer delinquency variety (β = −0.06, t = −2.86, p = .004), and lifetime prevalence of illicit drug use (β = −0.07, t = −3.65, p < .001) were each significantly associated with lower perceptions of PJPY; however, there was no significant association between the outcome and lifetime delinquency variety or last year prevalence of police contact for illegal behaviour.

Whilst higher individual relative wealth was associated with significantly higher perceptions of PJPY (β = 0.05, t = 2.46, p = .01), there was no significant association between perceptions of PJPY and any of the three measures of occupational status (student, employed, looking for work). There was no significant effect of neighbourhood-level relative socio-economic disadvantage; however, higher levels of neighbourhood disorder were associated with significantly lower perceptions of PJPY (β = −0.05, t = −2.45, p = .01), whilst higher levels of neighbourhood social cohesion and trust (β = 0.09, t = 4.62, p < .001), and more positive neighbourhood attitude to diversity (β = 0.09, t = 5.00, p < .001) were associated with significantly higher perceptions of PJPY.

Although the regression model shows that participants in remote Australia had significantly higher perceptions of PJPY than those in major cities and outer regional areas, a postestimation test demonstrated that the remoteness coefficients were jointly equal to zero (Wald F(3,2876) = 1.42, p = .22). There was, however, evidence that state was a significant predictor of young people’s perceptions of PJPY (Wald F(7,2876) = 3.44, p = .001). After controlling for all other variables, participants in Queensland and New South Wales rated PJPY significantly lower than participants in each of Victoria, Tasmania, and South Australia, and participants in Western Australia rated PJPY significantly lower than participants in Victoria.

Discussion

The results from this study demonstrate that a large proportion of the sample of young people – slightly over one third (34%) – reported having a mental health condition. Whilst the survey did not state that the mental health condition had to be diagnosed by a medical or psychological practitioner, this finding parallels publicly available data which states that 40% of young Australians aged 16–24 experienced a mental disorder in 2021–2022 (Australian Bureau of Statistics 2022). This finding adds weight to the evidence demonstrating that mental illness for Australian young people is pervasive.

As predicted and consistent with the extant literature, young people with mental illness provided more negative perceptions of the police in relation to procedural justice than their counterparts who did not identify as having a mental illness. Although not the strongest predictor of perceptions of procedural justice policing for youth (the association was stronger for persons reporting low levels of respect for authority), young people with mental illness perceived police employed procedural justice significantly less often than other youth did. Given only 8% of the entire sample disclosed having contact with the police for illegal behaviour, these negative perceptions of the police may be based on other interactions with the police and/or vicarious experiences with the police. Research suggests that young people with mental illness are significantly more likely to encounter the police by being a witness to or a victim of crime (Dean et al. 2021) or via detention under mental health legislation for suffering a mental health crisis (Doulas and Lurigio 2010). It may be inferred that many of these negative perceptions of the police were based on first-hand police interactions where the police have not treated the young person with mental illness in a procedurally just manner.

The results from the AYSS also demonstrated that young people identifying as not heterosexual or as trans* or gender diverse were one of the cohorts most likely to view the police as lacking in procedural justice. This is not surprising given the wealth of literature that demonstrates significant mistrust and negative opinions of the police amongst LGBTIQA + cohorts which is perpetuated by a police culture of misconduct and discriminatory attitudes towards LGBTIQA + people (Miles-Johnson 2013a, 2013b). Yet LGBTIQA + youth and young people with mental illness are not two distinct groups. Research suggests that LGBT⁵ people are at an increased risk of suffering mental ill-health and are more likely to attempt suicide in their youth compared to those identifying as heterosexual (Rivers et al. 2018). In spite of social attitudes progressing to be more inclusive of LGBT people, young LGBT people still struggle navigating their sexual identities amid dominant heteronormative communities (Rivers et al. 2018). Data from the AYSS demonstrate that 48% of LGBTIQA + participants live with a mental health condition, compared to only 19% of their peers who identified as both cis-gendered and heterosexual, and correspondingly, 73% of young people with mental illness have diverse gender or sexuality, compared to 41% of participants who do not have a mental health condition. This overlap of marginalities demonstrates that the intersectionality between LGBTIQA + youth and mental illness may compound young people’s negative perceptions of the police.

The negative perceptions of the police reported by young people with mental illness (many of whom will have overlapping and marginalised identities) may also be theoretically explained by the role of self-uncertainty. Self-uncertainty has been described as an aversive state where individuals feel insecure regarding aspects of their self-knowledge (Sedikides et al. 2010). Contrary to self-certain individuals, self-uncertain individuals often feel insecure and lack self-esteem in relation to their emotions, behaviours, personal attributes, goals, and aspirations (Sedikides et al. 2010). Since young people in general have been found to possess greater levels of self-uncertainty than adults (Meilman 1979, Waterman 1982), young people with mental illness (with other intersecting and marginalised identities such as LGBTIQA + status) may possess greater levels of insecurity and self-uncertainty. Young people who are ethnically minoritized or who have a migration background also possess elevated levels of self-uncertainty regarding how authority figures view them and have historically mistrusted the police and feel ill-served by them (Murphy and Mazerolle 2018). The data from the AYSS confirms these mistrusting perceptions of the police whereby participants who were not perceived by others as Australian also reported significantly lower perceptions of procedural justice policing. One method by which self-uncertain individuals seek to ameliorate their levels of self-certainty is by identifying and categorising themselves with other like-minded groups to garner shared notions of self-knowledge and self-worth (Sedikides et al. 2010). This form of group aggregation and dichotomisation may perpetuate negative perceptions of young people affected by mental illness of police via the sharing of first-hand and second-hand negative police experiences amongst themselves – thus creating an ‘Us versus Them’ mentality with the police.

Self-uncertain individuals may therefore feel especially vulnerable and anxious during interactions with authority figures such as the police, especially given the authoritative and coercive nature of the police role (Watson and Angell 2013). Young people with mental illness (and persons with mental illness in general) experiencing feelings of uncertainty – due to identifying with a vulnerable and marginalised status – arguably pay the most attention to fair police treatment due to expectations and fear of being exploited by an authority figure (Watson and Angell 2013). Van den Bos (2001) argues that under conditions of self-uncertainty, individuals may be more motivated to attend to the way they are treated in decision-making processes. Given self-uncertainty is an unpleasant psychological state, individuals often elevate their self-esteem levels by using social interactions in the form of procedural fairness to acquire a sense of acceptance, respect, social worth, and social standing during interactions with others (De Cremer and Sedikides 2005). As such, self-uncertainty has been proposed to act as a crucial moderator regarding responses to procedural fairness (De Cremer and Sedikides 2005).

Given young people with mental illness and other vulnerable groups value procedurally fair treatment by the police – yet view the police to be more procedurally unjust than others do – it is important that police meaningfully engage with young people with mental illness using the four principles of procedural justice policing – trust, dignity and respect, voice, neutrality. Most importantly, procedural justice policing with vulnerable citizens (such as people suffering mental health crises) has been shown to enhance the emotional response of self-uncertain citizens, thus increasing feelings of safety for marginalised persons when encountering the police (Watson and Angell 2007). A further outcome of procedural justice policing with marginalised persons is increased cooperation and compliance with police directives leading to therapeutic outcomes for both citizens and police (Watson and Angell 2007). Conversely, police treatment with vulnerable individuals that is disrespectful, forceful, and procedurally unjust has been shown to lead to increased detrimental outcomes for marginalised persons (and police) that are often characterised by conflict and harm, even more so than with the general public (McCluskey 2003, Watson et al. 2008). It is recognised, however, that research encompassing young people more specifically is lacking in the extant literature and requires greater attention to understand the true effects of procedural justice policing with this cohort.

Perceptions of a lack of procedural justice during police encounters with young people with mental illness may have ramifications for this vulnerable cohort (and their guardians) where they might be reluctant to call police for assistance when experiencing early symptoms of a mental health crisis and/or require police assistance for law enforcement purposes. This mistrust of the police is especially problematic given the dearth of emergency mental health resources available to young people with mental illness (and adults) whereby the police are often the only available 24 h emergency response acting as de facto mental health practitioners (Cummins and Edmondson 2016). Interagency collaborative schemes – whereby police form formal and informal cross service partnerships with mental health clinicians/professionals – may provide an ideal solution for reducing the role police play in responding to young people’s mental health needs (Marcus and Stergiopoulos 2022). The format of these schemes are nuanced across jurisdictions and can take the form of co-responder models (where police and mental health clinicians provide a bilateral response to crises situations), Crisis Intervention Teams (where mental health clinicians train special units of police to specialise in emergency mental health response), or telecommunication services (where police can contact a mental health practitioner in a control room via radio or phone for advice) (Miles-Johnson and Morgan 2022).

Despite evidence demonstrating the efficacy of these schemes (Seo et al. 2020), they are often underfunded, not available 24 h a day, and are only available in metropolitan areas across much of Australia (Miles-Johnson and Morgan 2022). The significant mistrust of the police by young people with mental illness (and LGBTI and ethically minoritized youth) raises questions regarding the effectiveness of police policy, training, and operational practice for responding to vulnerable and marginalised persons in the community. Given inappropriate and forceful police responses to persons suffering mental illness sometimes result in volatile situations which can to lead to civilian fatalities (Ruiz and Miller 2004, de Tribolet-Hardy et al. 2015), it is imperative that police follow procedural justice principles in practice to improve the safety of young people with mental illness whilst also benefiting the police by enhancing officer safety and encouraging citizen co-operation (Skogan et al. 2015).

Conclusion

Procedural justice policing is important for young people who have stigmatised and uncertain identities such as mental illness and other intersectional marginalities. This paper has demonstrated how young persons with mental illness and other young people with marginalised identities are more likely than their peers to mistrust the police and believe them to be procedurally unjust as an organisation. Public perceptions of the police are important since they may influence the way in which the public and the police interact with one another and whether young people with mental illness will call for police assistance.

Ethics approvals

Human Research Ethics approval for the AYSS was received from the Queensland University of Technology (QUT) University Human Research Ethics Committee (1800000095), and this project was approved by QUT as a Human Ethics Exempt Application (HE-Ex 2021-4352-4713).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Australian Research Council: [Grant Number DE180100853].

Notes

1 Data collection was completed two days before the first covid-19 case was confirmed in Australia.

2 The Australian Bureau of Statistics estimates that in December 2019 there were 3,916,443 Australians aged 14–25.

3 Corresponding 2021 Census data had not been released at the time of writing.

4 We did run a version of the model that included interaction terms between these four key identity markers, but as none of these terms were statistically significant, we report the more succinct model. As such, our interpretation of the intersectionality of these multiple marginalities is that they additively accumulate in a young person in the form of cumulative disadvantage, not that there is a multiplicative effect.

5 The term LGBT people was specifically used in Rivers et al.’s (2018) research.

Appendix

Table A1. Variables used in the analysis.

Variable	Description & coding
Perception of procedural justice policing for youth (PJPY)	Mean index of four items, measured on a 4-point Likert scale: ‘In your opinion, how often do police do the following things?  … The police treat young people with respect  … The police make fair decisions when dealing with young people  … The police explain their decisions and actions to young people  … The police listen to young people’ Response options: Almost never (1); Sometimes (2); Often (3); Almost always (4) Index was only created if respondent answered all four items. Cronbach’s alpha = 0.8969
Mental health condition	‘Do you have a disability? (Select all that apply to you)  … Mental health condition’ Yes = 1, No = 0
Gender	Cross categorisation of responses to the following two items: ‘What sex were you assigned at birth (what was put on your birth certificate)?  … Female  … Male  … Intersex  … Something else’ ‘What is your current gender identity?’ [open-ended item] Responses were coded as cis male or cis female if sex corresponded with gender (or if the respondent only answered sex and did not answer gender). Responses were coded as trans* & gender diverse if sex did not correspond with gender, or if a specific non-cis gender was specified (such as a gender, questioning, non-binary etc.).
Heterosexual	Coded from responses to: ‘Do you consider yourself to be:  … Heterosexual or straight  … Gay or lesbian  … Bisexual  … Prefer to self-identify as … ’ Responses were collapsed into Yes = 1 if respondent selected Heterosexual, No = 0 otherwise.
Perceived as Australian	Coded from responses to: ‘Which ethnicity do you think other people (who don’t know you) mostly see you as?’ Response categories: Australian; Aboriginal or Torres Strait Islander; Pacific Islander; Māori; New Zealander; South-East Asian (e.g. Vietnamese, Singaporean, Indonesian etc.); North-East Asian (e.g. Chinese, Korean, Japanese etc.); South/Central Asian (e.g. Indian, Bangladeshi, Pakistani etc.); Middle Eastern; European; North American; South American; African; They mostly see me as (write in). Responses were collapsed into Yes = 1 if respondent selected Australian, No = 0 otherwise.
Other disability	Coded from responses to: ‘Do you have a disability? (Select all that apply to you)  … Chronic medical condition  … Intellectual disability  … Learning disability  … Neurological condition (incl. ASD, ADHD)  … Physical disability  … Sensory disability’ Responses were collapsed into Yes = 1 if respondent answered Yes to any of the items, No = 0 otherwise. Index was only created if respondent answered at least one item
Age	‘How old are you?’ Respondents could only complete the survey if they entered an integer between 14 & 25 inclusive.
Religiosity	Coded from responses to: ‘How important to you (personally) is religion in your everyday life?’ Response options: Very important (1); Quite important (2); A bit important (3); A bit unimportant (4); Quite unimportant (5); Very unimportant (6) Responses were reverse coded to range from Very unimportant = 1 to Very important = 6
Respect for authority	Coded from responses to: ‘How wrong do you think is it for someone of your age to do the following?  … Lie, disobey or talk back to older people or people in authority’ Response options: Very wrong (1); Wrong (2); A little wrong (3); Not wrong at all (4) Responses were reverse coded to range from Not wrong at all = 1 to Very wrong = 4
Parental emotional support	Coded from responses to: ‘How well do you get along with your parent(s)? (Please indicate how much you agree or disagree with the following statements)  … I can easily get emotional support and care from my parents’ Response options: Strongly agree (1); Agree (2); Neither (3); Disagree (4); Strongly disagree (5) Responses were reverse coded to range from Strongly disagree = 1 to Strongly agree = 4
Self control scale	Mean index of nine items, measured on a 4-point Likert scale: ‘How much do you agree or disagree with the following statements?  … I act on the spur of the moment without stopping to think  … I do whatever brings me pleasure here and now, even at the cost of some future goal  … I’m more concerned with what happens to me in the short run than in the long run  … I like to test myself every now and then by doing something a little risky  … Sometimes I will take a risk just for the fun of it  … Excitement and adventure are more important to me than security  … I try to look out for myself first, even if it means making things difficult for other people  … If things I do upset people, it’s their problem not mine  … I will try to get the things I want even when I know it’s causing problems for other people’ Response options: Completely agree (1); Somewhat agree (2); Somewhat disagree (3); Completely disagree (4) Higher values of the scale indicate greater self control. Index was created if respondent answered at least one item, missing items were disregarded in calculations. Cronbach’s alpha = 0.7738
Victimisation variety lifetime prevalence (LTP)	Summative index of seven items, measured on a Yes = 1/No = 0 scale: ‘Did any of the following things ever happen to you?  … Someone wanted you to give them money or something else (like a watch, shoes, mobile phone) and threatened you if you refused  … Someone hit you violently or hurt you – so much that you needed to see a doctor  … Something was stolen from you (such as a book, money, mobile phone, sport equipment, bicycle …)  … Someone threatened or verbally abused you because of your religion, the language you speak, the colour of your skin, your social or ethnic background, your gender or sexuality, or for similar reasons.  … Someone committed physical violence against you because of your religion, the language you speak, the colour of your skin, your social or ethnic background, your gender or sexuality, or for similar reasons.  … Someone used social media, the internet, or mobile phone to bully, threaten or humiliate you  … Has your parent or step-parent ever hit you with an object, punched or kicked you forcefully, or beat you up?’ Index was created if respondent answered at least one item, missing items were treated as 0. Cronbach’s alpha = 0.5574
Delinquent peers variety	Summative index of five items, measured on a Yes = 1/No = 0 scale: ‘People sometimes engage in illegal activities. How many friends do you have who have done any of the following? I have friends who have...  … Used soft or hard drugs (like weed, hash, ecstasy, speed, heroin or coke).  … Stolen things from a shop or department store  … Entered a building without permission to steal something  … Threatened somebody with a weapon or beaten someone up, just to get their money or other things  … Beaten someone up or hurt someone badly with a weapon’ Index was created if respondent answered at least one item, missing items were treated as 0. Cronbach’s alpha = 0.5668
Any drug use LTP	Coded from the responses to the following three items: ‘Have you ever … ?  … Used cannabis (weed/marijuana/hash)?  … Used ecstasy, LSD, speed, amphetamines or similar drugs?  … Used heroin, cocaine, or crack?’ Coded as Yes = 1 if respondent answered Yes to any of the three items, No = 0 otherwise. Index was created if respondent answered at least one item. Cronbach’s alpha = 0.6604
Delinquency variety LTP	Summative index of thirteen items, measured on a Yes = 1/No = 0 scale: ‘Have you ever in your life …   … Painted on a wall, train, or bus (graffiti)?  … Damaged something on purpose, such as a bus shelter, a window, a car, or a seat in the bus or train?  … Stolen something from a shop or department store?  … Broken into a building to steal something?  … Stolen a bicycle?  … Stolen a motorbike or car?  … Stolen something off or from a car?  … Used a weapon, force, or threat of force to get money or things from someone?  … Stolen something from a person without force or threat?  … Carried a weapon, such as a knife, gun, stick, or chain?  … Taken part in a group ­fight in a public place?  … Beaten someone up or hurt someone with a weapon (including a knife or a heavy object) so badly that they were injured?  … Sold any drugs or helped someone sell drugs?’ Index was created if respondent answered at least one item, missing items were treated as 0. Cronbach’s alpha = 0.7197
Police contact LTP	‘Have you ever had contact with the police because you yourself did something illegal?’ Yes = 1, No = 0
Relative wealth	‘If you compare yourself with other people of your age: Do you have more, the same, or less money (of your own) to spend?’ Response options: Much less (1); Less (2); Somewhat less (3); The same (4); Somewhat more (5); More (6); Much more (7)
Student	Coded from responses to: ‘What is your occupation? (select all that apply to you)’ Response options: Studying (high school) full-time; Studying (high school) part-time; Studying (post-secondary) full-time; Studying (post-secondary) part-time; Working full-time; Working part-time or casual; Looking for work; Caring for a family member; Not able to work because of disability; Something else (write in) Responses were collapsed into Yes = 1 if respondent selected any of the Studying options, No = 0 otherwise. Responses were only coded if at least one option was selected.
Employed	Coded from responses to: ‘What is your occupation? (select all that apply to you)’ (see above for response options) Responses were collapsed into Yes = 1 if respondent selected Working full-time or Working part-time or casual, No = 0 otherwise. Responses were only coded if at least one option was selected.
Looking for work	Coded from responses to: ‘What is your occupation? (select all that apply to you)’ (see above for response options) Responses were collapsed into Yes = 1 if respondent selected Looking for work, No = 0 otherwise. Responses were only coded if at least one option was selected.
Neighbourhood disorder	Mean index of five items, measured on a 4-point Likert scale: ‘How much do you agree or disagree with the following statements about your neighbourhood?  … There is a lot of crime in my neighbourhood  … There is a lot of drug selling in my neighbourhood  … There is a lot of fighting in my neighbourhood  … There are a lot of empty and abandoned buildings in my neighbourhood  … There is a lot of graffiti in my neighbourhood’ Response options: Completely agree (1); Somewhat agree (2); Somewhat disagree (3); Completely disagree (4) Responses were reverse coded to range from Completely disagree = 1 to Completely agree = 4. Higher values on the index indicate higher levels of neighbourhood disorder. Index was created if respondent answered at least one item, missing items were disregarded in calculations. Cronbach’s alpha = 0.8583
Neighbourhood social cohesion & trust	Mean index of six items, measured on a 4-point Likert scale: ‘How much do you agree or disagree with the following statements about your neighbourhood?  … Many of my neighbours know me  … People in my neighbourhood often do things together  … People around here are willing to help their neighbours  … This is a close-knit neighbourhood  … People in this neighbourhood can be trusted  … People in this neighbourhood generally get along well with each other’ Response options: Completely agree (1); Somewhat agree (2); Somewhat disagree (3); Completely disagree (4) Responses were reverse coded to range from Completely disagree = 1 to Completely agree = 4. Higher values on the index indicate higher levels of social cohesion and trust. Index was created if respondent answered at least one item, missing items were disregarded in calculations. Cronbach’s alpha = 0.8353
Neighbourhood attitude to diversity	Mean index of three items, measured on a 4-point Likert scale: ‘How much do you agree or disagree with the following statements about your neighbourhood?  … People in my neighbourhood prefer that residents in the area are mostly white  … People in my neighbourhood do not like having people from other ethnic groups as neighbours  … Some people in my neighbourhood have been left out of social events because of their skin colour, ethnicity, race or religion’ Response options: Completely agree (1); Somewhat agree (2); Somewhat disagree (3); Completely disagree (4). Higher values on the index indicate more positive attitude to diversity. Index was created if respondent answered at least one item, missing items were disregarded in calculations. Cronbach’s alpha = 0.7532
Relative Socio-economic Disadvantage decile	Australian Bureau of Statistics Socio-Economic Indices for Areas (SEIFA) Index of Relative Socio-economic Disadvantage decile, matched to respondent’s residential postcode. Higher values indicate areas of less relative socio-economic disadvantage.
Remoteness	Australian Bureau of Statistics Remoteness Areas, matched to the respondent’s residential postcode, using a Mesh Block population weighted correspondence file. Categories: Major cities of Australia; Inner regional Australia; Outer regional Australia; Remote Australia; Very remote Australia. Where postcodes crossed multiple remoteness area categories, the category that contained the majority of the population of the postcode was used.
State	‘Which state or territory do you live in?’ Response options: Queensland; New South Wales; Victoria; Tasmania; South Australia; Western Australia; ACT; Northern Territory.