Do patron bans reduce crime? An examination of assault offences in Western Australia, before and after the introduction of police-imposed barring notices

ABSTRACT

This paper explores whether the introduction of police-imposed barring notices in Western Australia (WA) is associated with changes to the number, type and location of recorded assaults. Police-imposed barring notices were introduced in WA in January 2011, and are issued in response to alcohol-related disorderly behaviours in/around licensed venues. A barring notice can exclude the recipient from one or more licensed venues for up to one year. WA Police provided unit level records of 67,750 assault incidents occurring during high-alcohol hours – 8pm to 6am on Friday and Saturday nights – between 1 July 2007 and 30 June 2020. This enabled data from before and after the introduction of barring notices to be compared. The findings identified a number of potentially positive effects of barring notices, including significant reductions (p < 0.001) in non-family assault offences marked with an alcohol flag, assault offences recorded occurring on pathways (which includes areas around licensed premises), and common assault offences marked with an alcohol flag. These categories clearly align with the intended use and effect/s of barring notices, and the findings are encouraging. The effects of other factors on offending numbers cannot be excluded, but the association between the introduction of barring notices and, in particular, a reduction in alcohol-flagged violent offending appears to support the use of barring notices in WA.

KEYWORDS:

• Barring notice
• patron banning
• assault
• alcohol flag

Introduction

Governments in many jurisdictions have responded to concerns about alcohol-fuelled disorderly behaviours by implementing a variety of legislative reforms, regulations, and operational procedures to address violence and risky drinking practices. Across Australia, a key feature of responses to issues of alcohol-related disorder in entertainment districts has been the introduction of one or more methods of patron banning. Banning provisions range from the discretionary facility for licensees to exclude individuals from their venues to on-the-spot police-imposed venue or public area bans, and court-imposed exclusion orders. Since 2007, the use of banning provisions has continued to expand both legislatively and operationally (Farmer, Curtis & Miller, 2018, Farmer & Clifford, 2020), although concern has been expressed about a lack of scrutiny of the use of patron banning (Palmer & Warren, 2014, Miller, Curtis & Palmer, 2016, Curtis et al. 2018, 2022, Farmer, 2018, 2019, 2021a, 2021b; Taylor et al. 2018, 2021, Farmer, Clifford & Miller, 2021). Despite limited evidence-based analysis of their effects and effectiveness as a mechanism to address alcohol-related disorderly behaviours, all Australian jurisdictions have now introduced one or more forms of patron banning (Farmer & Clifford, 2020).

The specifics of banning provisions differ across Australian states and territories (Farmer, Curtis & Miller, 2018, Farmer & Clifford, 2020), but their underpinning rationale is generally consistent. The banning of patrons, from licensed premises and/or wider public areas (such as nightlife entertainment precincts), is generally presumed to change the behaviours of recipients and deter non-recipients from disorderly behaviour and, as a result, increase/ensure public safety and prevent crime (Farmer, Curtis & Miller, 2018). The demonstrable effects of banning are limited, and provisions have typically been implemented without ongoing scrutiny of their use or effects (Farmer, Curtis & Miller, 2018, Farmer, Clifford & Miller, 2021). Bans do remove individuals from a specific place, can provide an immediate response to a behavioural issue within a given location, and may prevent a recurrence if the ban is enforced in a proactive and effective manner. However, the specific effects on subsequent patron behaviours, as well as the community more generally, are less clear. Curtis et al. (2018) undertook a community-based survey and found limited knowledge and understanding of patron banning. They concluded that it is difficult for banning to have a wider effect on behaviours if sufficient numbers of people are not aware of the risk of receiving, or the consequences of, a patron ban. In a study of the offending records of banned patrons from a regional city in Victoria, Australia, Curtis et al. (2022) examined the median number of charges before and after receiving a ban, and found that they were not significantly different. However, when analysing short term effects (two years before and after receiving a ban), Curtis et al. found a significant increase in anti-social offending after receiving a ban compared to before. The study suggests that police-imposed bans in Victoria do not appear to deter recipients from offending or change already established patterns of anti-social behaviours.

In Denmark, Sogaard et al. (2017) and Sogaard (2018) examined police-imposed zonal bans that are operational within urban nightlife precincts. Since the early 2000s, Danish police have been empowered to ban patrons from defined geographical areas for two years. Bans can be issued by an officer at the time of arrest, as a result of the recipient’s behaviour in or associated with a visit to a particular venue. Bans apply to all venues and connecting public areas in the affected area. Offences leading to a zonal ban mostly include possession and dealing of illegal drugs, but can also cover weapons possession and a range of acts of violence. Initial breaches can lead to a fine, with repeat violations punishable with up to four months custody. The rationale for banning is to increase public safety by removing troublesome individuals, and to deter those who have been banned from further problematic behaviours. However, there has been little empirical analysis of the actual effect of banning on offending or community safety. Sogaard (2018) interviewed ban recipients in the Danish city of Alborg to explore their perceptions and the effects of their ban. Sogaard found that ban breaches were common-place and largely normalised. Rather than deterring problematic behaviours, the possibility of being banned was found to increase the risk of violence and other undesirable behaviours. Overall, the expected effects of zonal banning were not apparent from the recipients interviewed. In June 2021 the Danish Parliament passed a bill to allow violent offenders to be banned from establishments serving alcohol between midnight and 5am (Nord News, 2021). The operation of these bans and the impact of the exclusion of violent offenders on alcohol-related harms or problematic behaviour in the night-time economy is yet to be observed.

Various types of patron banning operate across international jurisdictions, but direct comparisons are complicated by differing judicial and operational approaches. Banning is typically positioned within the management of licensed premises and/or policing procedures and provisions which regulate the control of public spaces. Socio-spatial criminology embodies analyses of the control of urban space, perceptions and responses to perceived disorder, contractual governance, and spatial governmentality as a means of pre-emptive and reactive behavioural control. The use of exclusion from public areas in response to anti-social behaviours has also been examined in other countries, such as England and Wales (Crawford, 2009, Heap & Dickinson, 2018, Heap, Black & Rodgers, 2021), the USA (Beckett & Herbert, 2010), Germany (Belina, 2007), and Canada (Sylvestre, Bernie & Bellot, 2015). Persak (2017) and Persak and Ronco (2018) documented the spatial regulation of disorder in a number of European jurisdictions. The literature of socio-spatial criminology is extensive and space precludes a more detailed discussion. However, a common feature is a lack of evaluation of the effects and effectiveness of banning provisions. This paper provides new insights into the potential benefits of patron banning by examining changes in recorded assault offences after the introduction of barring notices in one Australian jurisdiction.

The current study: Western Australia

The focus of this paper is the Australian state of Western Australia (WA), where licensee bans, police-imposed barring notices, and prohibition orders are used to primarily target alcohol-related violence and disorderly behaviours. In common with patron banning more broadly, each provision is presumed to prevent crime, change the behaviours of recipients, and increase public safety by: providing a tangible consequence for recipients by excluding them from licensed premises; deterring the recipient from further behaviours which may lead to another ban; deterring the wider community from engaging in behaviours which may lead to a ban (Farmer, Curtis & Miller, 2018). One key indicator that could point to the overall effectiveness of patron banning is a reduction in the level of problematic behaviours in/around licensed premises – measured by the number/type of recorded offences. This paper draws from research which explored the potential effects of the introduction of police-imposed barring notices on the number, type and location of recorded assault offences. It is acknowledged that any changes in recorded offence data cannot necessarily be attributed solely to the exclusion of banned patrons, particularly where other mechanisms may also be operational. However, given the key rationale for patron banning, there is merit in undertaking an analysis of recorded offences to identify any potential beneficial effects.

In WA, police-imposed barring notices are used primarily in response to alcohol-related violence and disorderly behaviours, and have been operational since 2011. Following an amendment to the WA Liquor Control Act 1988 the Commissioner of Police, or a police officer above the rank of Inspector, is empowered to approve the exclusion of an individual from a specified licensed venue, or a class of licensed venues, if there is reasonable belief that the recipient has ‘(a) been violent or disorderly; or (b) engaged in indecent behaviour; or (c) contravened a provision of any written law’ (s.115AA). The original legislation required the alleged behaviour to have occurred within a licensed venue. A 2018 amendment expanded the permissible imposition of barring notices to include problematic behaviours occurring in the vicinity of licensed premises (Liquor Control Act 1988, s.115AA(2)) – to include persons misbehaving in queues, on footpaths, in car parks close to licensed premises. There is no specific distance which denotes the ‘vicinity’ of a licensed venue.

Unlike other Australian jurisdictions, barring notices in WA are not imposed by police officers on-the-spot. A determination is made by the centralized Liquor Enforcement Unit (LEU), following consideration of evidence such as CCTV and witness statements, and the process can take a number of weeks. The barring notice can be imposed for any period up to 12 months, taking effect from the day on which it is served. Although the barring period starts from the date of the problematic incident. The penalty for breaching a barring notice is a fine of up to $10,000 (Liquor Control Act 1988, s.115AA). Details of banned persons are made available to licensees manually (typically via an online portal). This places the emphasis on venue staff to visually identify individuals who have been banned. Some venues do use ID scanners to manage patron entry, but information regarding banned patrons is generally uploaded to the scanner by the venue. There is currently no mandated network of ID scanners, such as those operational in Queensland, to enable venues to check patrons on entry to ascertain whether they have been banned by police or other venues (de Andrade et al. 2021). Police officers rely on help from venues to identify patrons who may be breaching a barring notice and/or checks undertaken during the process of questioning someone following an incident. Pre-emptive and proactive monitoring of banned patrons is difficult.

Research method

The wider study from which this paper is drawn was initiated and funded by WA Police, in collaboration with Deakin University, to examine the effect/s and effectiveness of barring notices and prohibition orders. Ethical approval was obtained from the Deakin University Human Research Ethics Committee in October 2018 (reference: 2018-297). Approval was also received from the WA Police Research Governance Unit as part of the overall project tender and initiation processes.

Procedure

This paper reports the findings of an analysis of the potential impact of barring notices, which came into effect on 17 January 2011, on nightlife crime in WA. The number, type, and location of recorded violent offences have been examined. As barring notices are intended to be issued in response to alcohol-related disorderly behaviours in/around licensed venues, WA Police provided a dataset comprising unit level records of assault incidents occurring during high-alcohol hours [HAH] – Friday and Saturday nights, 8pm to 6am¹ – between 1 July 2007 and 30 June 2020. This date range enabled data from before and after the introduction of barring notices to be compared. Each record included the type of offence; age and sex of perpetrator and victim(s); type of location; geographic location (based on police district); and date and time of day of offence.

The assault incidents were aggregated into two categories based upon severity: serious assaults, and common assaults. The rationale for this division drew from the 2011 Australian and New Zealand Standard Offence Classification (ANZSOC). Subdivision 021 Assault covers ‘The direct (and immediate/confrontational) infliction of force, injury or violence upon a person or persons or the direct (and immediate/confrontational) threat of force, injury or violence where there is an apprehension that the threat could be enacted’ (Australian Bureau of Statistics, 2011).

Under the ANZSOC, the Assault subdivision is further disaggregated into:

• 0211 Serious assault resulting in injury

  0212 Serious assault not resulting in injury

  0213 Common assault

For the purposes of this analysis, the two serious assault categories were combined. Serious assaults included the following offences:

• act intending to cause grievous bodily harm or prevent arrest,

• act or omission with intent to harm and causing bodily harm,

• act or omission with intent to harm and endangering life,

• assault causing bodily harm,

• assault serious,

• assault with intent,

• cause poison to be administered,

• endanger safety of aircraft with intent to injure or endanger,

• grievous bodily harm,

• maliciously administer poison,

• stupefy to commit indictable offence,

• unlawful act causing bodily harm,

• unlawfully cause explosion to cause grievous bodily harm or prevent arrest,

• urge dog to attack or chase,

• wounding,

• any assault on a police officer.

Common assaults comprised the following offences:

• common assault,

• intoxication by deception,

• set, construct or place dangerous thing,

• urge a dog to attack or chase (not involving a police officer).

Analysis

The dataset has been analysed to determine the nature and extent of any changes in recorded assault offence numbers associated with the introduction of barring notices. Analyses have been conducted by location type and offence category. Specific analyses have examined the dataset to differentiate incidents marked with a family flag and also specifically for offences marked with alcohol flags. The analysis has focused primarily on state-wide data but has also examined assault offences recorded in the cities of Perth and Fremantle – both of which are the locations of major entertainment districts. Other regional differences are acknowledged but are not addressed specifically in this paper.

The dataset was examined using interrupted time series (ITS) analysis in Stata 15. ITS allows simultaneous tests of step (immediate) and slope (gradual) changes within data trends. The itsa Stata command was used for all models (Linden, 2015), this allows the user to choose between two ordinary least squares regression-based time series models: newey and prais. Newey models were used by default, when significant autocorrelation was detected using the Breusch–Godfrey test (Baum, 2006) prais models were used to account for this (Linden, 2015). Monthly lag was set to 0 in all newey models, while prais models have a lag of 1 by design (Linden, 2015). To examine whether and how the introduction of barring notices affected the number and type of recorded assault offences, the step effect was examined post January 2011 [BN (step)], and the slope effect was examined both for the whole period, 2007–2020 [Time (slope)], and for the period since January 2011 [Time x BN (slope)]. A total of 67,750 offence records/incidents were included within the analyses. The modelling included a linear time variable to control for any underlying long-term trends, and a monthly seasonal variable to account for seasonality in the data. The results are reported in a series of tables and figures, with significant findings marked as follows: *** p < 0.001; ** p < 0.005; * p < 0.05.

Analytical considerations and limitations

It is recognised that a range of individual, situational, regulatory and operational factors will influence the risk, likelihood and type of violent offending, in addition to the implementation or availability of particular police provisions. However, barring notices were introduced as a specific response to alcohol-related anti-social and disorderly behaviour in/around licensed venues and within nightlife entertainment districts. Any potential association between their introduction and the number/type of violent offending merits analysis.

The introduction of barring notices coincided with a change to licensed venue trading hours – from 6am to 5am on Saturday and Sunday mornings (Liquor Control Amendment Bill 2010). This change differs from the trading hour restrictions introduced in New South Wales and Queensland, which took effect much earlier in the night (i.e. from 5am to 3am or 3:30am) (Kypri et al. 2014, Coomber et al. 2021), and likely influenced a greater proportion of nightlife venues and patrons as a result. Prior evidence has failed to demonstrate that reductions in trading hours of less than two hours reduce excessive alcohol consumption and related harms (Hahn et al. 2010). While the introduction of barring notices cannot be entirely separated from this reduction in trading hours, the availability of the new power to issue barring notices fundamentally changed how police can operate within these spaces. By contrast, the influence of the reduction in trading hours from 6am to 5am is likely to be more limited, with respect to the operation of a small number of nightlife venues.

Offences dataset – family cases flag: Incidents that involve or relate to family violence are flagged by WA Police within the offence dataset. This relies upon the accurate reporting and recording of such offences, and the approach is not necessarily uniform across WA. To mitigate the effect of omissions, different approaches or embedded errors, the analyses differentiate between family and non-family cases, overall and with respect to the location in which the incident occurred. For the latter, this comprises recreation, non-recreation (of which most are domestic settings) and pathways.

Offences dataset – alcohol flag: Incidents that are regarded as alcohol-related during high alcohol hours (HAH) are also flagged within the offence dataset. It is acknowledged that alcohol flags are not used consistently and can be unreliable as markers (Morgan et al. 2018, from p. 20; Nepal et al. 2019). As a result, analytical findings relating to the use of flags may not necessarily accurately reflect individual behaviours or aggregated real-world trends.

The proportion of recorded assault offences each year with police recorded alcohol or family cases flags changed across the period examined. This is noted within the supplementary Table S1. However, it is not possible to discern whether the relative proportions reflect substantive changes in offending types or changes in the use of the flags.

Wounding: At the request of WA Police, the offence of wounding was examined within the aggregated category of serious assaults and also separately, as a number of high-profile wounding incidents – involving ‘glassings’ or assault using a broken glass – were influential in the introduction of barring notices. Overall, the number of ‘wounding’ offences was low, and caution should be exercised regarding the statistical significance of the analyses. Wounding also includes offences involving a knife, scissors, axe, screwdriver or other sharp instrument, which further highlights the need for caution when interpreting the data.

Covid-19: Across the paper, offence numbers have been examined by month/year. In places, notable changes are evident within the first half of 2020. This coincided with the first period of heightened COVID −19 restrictions and any effects should be considered within this context. Given the purpose of this study, these changes are noted but have not been specifically examined in this paper.

Results

Table 1 shows the distribution of police-imposed barring notices across the period examined, since their introduction in 2011. The subsequent analysis is not able to determine how many of the recorded assault offences relate to barring notice recipients.

Table 1. Number of barring notices (BN) served each year: 2011-2020.

	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020^1	Total
BNs Served	207	237	371	535	737	506	465	381	419	165	4,023

¹ to30 June 2020 only

Table 2 summarises the 67,750 recorded offences by assault category. Common assaults comprised 59% of the offence records and serious assaults accounted for 41%. Wounding was included within the serious assault data and, when considered separately, comprised 4% of the overall offences recorded.

Table 2. Number of recorded offences, by offence category.

Offence category	Total number recorded 2007–2020
Common assault	40,211
Serious assault (including wounding)	27,539
Wounding only	2,737
Total	67,750

Total offence numbers by key location types

Table 3 summarises the dataset according to the location in which the offence was recorded to have taken place. Given that barring notices are intended to be used in relation to alcohol-related disorderly or anti-social behaviours in/around licensed venues, the data has been aggregated into three overall location types: recreation, non-recreation and pathways. The specific locations that sit within each of the three designations are set out in Table 3. For offences within the ‘pathways’ category, it is not known whether the streets/footpaths were in residential areas, entertainment districts, or other locations.

Table 3. Location of recorded offences, by overall location type and sub-category: 2007–2020.

Location Type	Total 2007 - 2020	Percentage
Recreation	Total: 9,660	14%
• Casino	811
• Hall / Function Centre	325
• Hotel / Tavern	3,958
• Night Club	3,552
• Restaurant	674
• Others (church, library, museum / art-gallery, sport / recreation centre, theatre / cinema)	340
Non-Recreation	Total: 38,751	57%
• Commercial	766
• Government	258
• House	32,536
• Land	1,407
• Other Dwelling	1,774
• Other Premises	611
• Professional	453
• Shops	729
• Others (construction, industrial, rural, water)	217
Pathways	Total: 19,339	29%
• Street / Footpath	16,038
• Bus Stop / Train Stop	596
• Car Park	2,052
• Public Transport	317
• Railway Line	190
• Others (airport, bridge / overpass, marina / harbour)	146
Grand Total	67,750	100%

Across the whole period, 14% of the recorded offences took place in recreation locations, of which hotels, taverns and nightclubs were the most common specific locations. The most common single location for recorded assaults was the home – accounting for 84% of the non-recreation offences, and 48% of all recorded offences.

Figure 1 depicts the offence numbers for each of the three location types across the period examined. Across the whole period offence numbers recorded at the recreation and pathways location types decreased, but the number increased in non-recreation locations. The notable drops relating to pathways and recreation offence locations in early 2020 coincided with temporary responses to COVID-19.

Figure 1. Recorded assault offence numbers, by location type: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 4 summarises the results of the time series analyses for each aggregated location type. Across the whole period, there were no significant step or slope changes in the number of offences recorded in non-recreation locations. In recreation locations, a slope decrease was evident across the whole period, but is not specifically associated with the introduction of barring notices. A significant step effect decrease was evident for the pathways location type immediately after the introduction of barring notices (p < 0.001).

Table 4. Time series models for total count of offences, by location type: 2007–2020.

Location Type		β co-eff	95% confidence interval	p-value
Non-Recreation	Time (slope)	−0.056	−0.929, 0.816	0.898
	BN (step)	15.628	−8.849, 40.106	0.209
	Time x BN (slope)	0.762	−0.131, 1.656	0.094
Recreation	Time (slope)	−0.392	−0.705, −0.08	0.013
	BN (step)	−1.542	−10.29, 7.21	0.728
	Time x BN (slope)	0.045	−0.274, 0.365	0.778
Pathways	Time (slope)	−0.018	−0.669, 0.632	0.954
	BN (step)	−33.988	−52.24, −15.736	<0.001 ***
	Time x BN (slope)	−0.302	−0.969, 0.364	0.372

Total offences recorded, by ‘Recreation’ location type

A further set of analyses examined the 9,660 offences recorded within the recreation location category, looking more specifically at the three different offence types. As they only accounted for a small number of offences (n = 340), the following recreation location types were excluded from this analysis: sports/recreation centre, library, church, theatre/cinema, museum/art gallery.

Figure 2 depicts the numbers of each type of assault offences recorded in the ‘recreation’ location within each year examined. Across the whole period examined the number of woundings, serious assaults, and common assaults recorded in recreation locations all decreased. The notable drops in early 2020 coincided with temporary responses to COVID-19. Table 5 summarises the results of the time series models for the three offence categories recorded in recreation locations. Serious assaults in recreation locations decreased significantly across the 13 years examined (p = 0.001), but there were no significant step or slope changes associated with the introduction of barring notices for any of the offence categories.

Figure 2. Recorded offence numbers within recreation locations, by assault category: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 5. Time series models for offences recorded within recreation locations: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.04	−0.09, 0.01	0.117
	BN (step)	−0.29	−1.49, 0.92	0.639
	Time x BN (slope)	0.04	−0.02, 0.09	0.170
Serious assault	Time (slope)	−0.33	−0.52, −0.13	0.001 **
	BN (step)	−0.89	−6.77, 4.99	0.765
	Time x BN (slope)	0.19	−0.01, 0.39	0.062
Common assault	Time (slope)	−0.01	−0.25, 0.24	0.949
	BN (step)	−1.43	−8.26, 5.40	0.680
	Time x BN (slope)	−0.18	−0.42, 0.07	0.161

Analysis of all recorded offences

Figure 3 depicts the recorded offence numbers for each of the three offence categories within each year examined. Across the whole period serious assaults in general, and wounding specifically, decreased, but common assaults show an increasing trend. Table 6 summarises the results of the time series models for each offence category. Although common assaults increased across the whole period, this increase was not significant and there was no notable step or slope changes associated with the introduction of barring notices. For all serious assaults, the slope analysis shows a decrease across the overall period examined (p < 0.02), and the step analysis points to a decrease in serious assaults after the introduction of barring notices (p < 0.05). For all wounding offences, the slope analysis shows a decrease over the whole period 2007–2020 (p < 0.02) but no notable effects associated with the introduction of barring notices.

Figure 3. Recorded offence numbers, by assault category: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 6. Time series models for total count of offences: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.15	−0.28, −0.03	0.019 *
	BN (step)	1.47	−2.40, 5.34	0.455
	Time x BN (slope)	0.07	−0.06, 0.20	0.269
Serious assault	Time (slope)	−0.99	−1.76, −0.22	0.012 *
	BN (step)	−23.39	−44.69, −2.09	0.032 *
	Time x BN (slope)	0.71	−0.07, 1.50	0.074
Common assault	Time (slope)	0.45	−0.10, 1.00	0.106
	BN (step)	6.72	−13.41, 26.85	0.510
	Time x BN (slope)	−0.19	−0.78, 0.40	0.520

Total recorded offences: family v non-family incidents

Figure 4 depicts the recorded offence numbers for each year for family and non-family incidents. This analysis includes all assault offences, with or without an alcohol flag. Across the period non-family offence numbers decreased, and family cases increased. The notable drop in non-family assaults in early 2020 coincided with temporary responses to COVID-19.

Figure 4. Recorded assault offence numbers, by family/non-family incident type: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 7 summarises the results of the time series models for total offences for family and non-family incidents. For non-family incidents across all offence types there was a step decrease immediately after the introduction of barring notices (p < 0.005). For family incidents a step increase in the number of recorded offences is noted, although this was not significant (p < 0.05).

Table 7. Time series models for total count of offences, by family/non-family incident type: 2007–2020.

Offence Type		β co-eff	95% confidence interval	p-value
Family	Time (slope)	0.407	−0.353, 1.167	0.292
	BN (step)	21.907	0.569, 43.245	0.044 *
	Time x BN (slope)	0.651	−0.127, 1.431	0.1006
Non-Family	Time (slope)	−0.875	−1.86, 0.11	0.081
	BN (step)	−41.809	−69.451, −14.167	0.003 **
	Time x BN (slope)	−0.145	−1.155, 0.864	0.775

Total recorded offences with an alcohol flag

Figure 5 depicts the number of assault offences within each category that were flagged as alcohol-related during HAH for each year examined. It is acknowledged that alcohol flags are not used consistently, can be unreliable as markers, and can result in misleading trends within the data analysed.² Analytical findings relating to the use of flags may not necessarily accurately reflect individual behaviours or aggregated real-world trends.

Figure 5. Recorded alcohol involved offence numbers, by assault category: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 8 summarises the results of the time series models for each offence category marked with an alcohol flag. For serious assaults there was a step reduction (p < 0.05) associated with the introduction of barring notices. For common assaults, there was a significant (p < 0.001) slope effect reduction associated with the introduction of barring notices, although numbers increased over the whole period (p < 0.005). Given the absence of a significant step effect, caution should be exercised as other factors may have influenced this longer-term change.

Table 8. Time series models for assault offences recorded with an alcohol flag: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.09	−0.20, 0.01	0.084
	BN (step)	1.16	−2.23, 4.55	0.500
	Time x BN (slope)	0.02	−0.09, 0.13	0.740
Serious assault	Time (slope)	−0.48	−1.07, 0.11	0.110
	BN (step)	−17.65	−33.10, −2.19	0.026 *
	Time x BN (slope)	0.02	−0.58, 0.62	0.943
Common assault	Time (slope)	0.59	0.19, 0.99	0.004 **
	BN (step)	−2.70	−16.04, 10.63	0.689
	Time x BN (slope)	−0.78	−1.20, −0.37	<0.001 ***

Total recorded offences, excluding family cases

Within the dataset, incidents were flagged where they relate to family/family-related violence. Table 9 sets out the total number of offences within the dataset for family and non-family incidents. The rationale here is to differentiate between offences that were more and less likely to have occurred in a domestic setting or in relation to a domestic issue. Again, it is acknowledged that the reliability of the family case flag, with respect to the consistency of its use, cannot be assured. Across the whole period, 57% of incidents were not marked with the family case flag.

Table 9. Total number of recorded assault offences by family/non-family Type: 2007–2020

Offence Type	Total 2007 - 2020	Percentage
Assault (Family)	28,980	43%
Assault (Non-Family)	38,770	57%
Total	67,750	100%

To explore the non-family cases in more depth, Figure 6 depicts the number of offences recorded within HAH, excluding incidents between family members, for each of the three assault categories. Across the period all three offence types decreased. The notable drops in early 2020 coincided with temporary responses to COVID-19.

Figure 6. Recorded offence numbers, excluding family cases, by assault category: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 10 summarises the results of the time series models for each offence category, excluding family cases. For both serious and common assaults, there was a step decrease immediately after the introduction of barring notices (p < 0.05). For serious assaults in general (p < 0.05), and wounding in particular (p < 0.005), there was a slope decrease across the whole period, 2007–2020, but no significant slope decrease associated with the introduction of barring notices.

Table 10. Time series models for assault offences recorded excluding family cases: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.13	−0.21, −0.05	0.001 **
	BN (step)	0.67	−2.03, 3.36	0.625
	Time x BN (slope)	0.06	−0.02, 0.14	0.156
Serious assault	Time (slope)	−0.94	−1.66, −0.23	0.010 *
	BN (step)	−19.17	−38.30, −0.05	0.049 *
	Time x BN (slope)	0.48	−0.24, 1.22	0.187
Common assault	Time (slope)	−0.04	−0.70, 0.62	0.904
	BN (step)	−20.33	−40.42, −0.25	0.047 *
	Time x BN (slope)	−0.53	−1.21, 0.14	0.118

Comparing alcohol involved offences, by family/non-family type

To offer a further layer of analysis, the data was examined for family and non-family incidents that were also marked with an alcohol flag. Figure 7 depicts the recorded offence numbers for family and non-family incidents marked with an alcohol flag. Across the period non-family offence numbers decreased, but family cases increased. The notable drop in non-family assaults early 2020 coincided with temporary responses to COVID-19.

Figure 7. Recorded alcohol involved assault offences, by family/non-family type: 2007–2020 (the vertical line indicates the introduction of barring notices).

Table 11 summarises the results of the time series models for recorded assault offences for family and non-family incidents marked with an alcohol flag. There were no findings of significance for family cases marked with an alcohol flag. For non-family incidents marked with an alcohol flag, there was a slope decrease associated with the introduction of barring notices (p < 0.05), and a significant step decrease immediately after their introduction (p < 0.001).

Table 11. Time series models for total count of alcohol involved offences, by family/non-family type: 2007–2020.

Location Type		β co-eff	95% confidence interval	p-value
Family	Time (slope)	0.38	−0.16, 0.91	0.165
	BN (step)	10.82	−4.07, 25.70	0.153
	Time x BN (slope)	−0.04	−0.59, 0.50	0.872
Non-Family	Time (slope)	−0.22	−0.84, 0.40	0.484
	BN (step)	−30.95	−48.26, −13.64	<0.001 ***
	Time x BN (slope)	−0.75	−1.39, −0.12	0.020 *

Perth and Fremantle

To acknowledge the intersection between the intended purpose/use of barring notices and the prevalence of nightlife districts in Perth and Fremantle, this section reports key findings for assaults recorded in these geographic areas.

All assault offences

Table 12 summarises the results of the time series models for each of the three assault categories for offences recorded in the Perth area. There were no significant step or slope effects for common assaults or for wounding. There was a slope increase in the number of serious assaults after the introduction of barring notices (p < 0.05), although serious assaults numbers decreased across the whole period examined (p < 0.05). Given the absence of a significant step effect, caution should be exercised as other factors may have influenced these longer-term changes.

Table 12. Time series models for assault offences recorded in Perth: 2007 – 2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.04	−0.09, 0.02	0.180
	BN (step)	0.53	−0.86, 1.92	0.450
	Time x BN (slope)	0.02	−0.03, 0.08	0.379
Serious assault	Time (slope)	−0.45	−0.77, −0.13	0.006 *
	BN (step)	−4.94	−13.38, 3.49	0.248
	Time x BN (slope)	0.44	0.11, 0.76	0.010 *
Common assault	Time (slope)	0.15	−0.14, 0.43	0.309
	BN (step)	−6.73	−14.99, 1.53	0.109
	Time x BN (slope)	−0.21	−0.49, 0.08	0.161

Table 13 summarises the results of the time series models for each of the three assault categories for offences recorded in the Fremantle area. Serious assaults show no significant changes across the period. For woundings there was a step decrease (p < 0.05) associated with the introduction of barring notices. For common assaults, there was an increase across the whole period (p < 0.005), but step and slope decreases associated with the introduction of barring notices (p < 0.05; p < 0.005).

Table 13. Time series models for assault offences recorded in Fremantle: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	0.01	−0.02, 0.03	0.653
	BN (step)	−0.93	−1.80, −0.05	0.038 *
	Time x BN (slope)	−0.01	−0.04, 0.02	0.576
Serious assault	Time (slope)	−0.07	−0.25, 0.11	0.417
	BN (step)	−4.19	−8.98, 0.60	0.086
	Time x BN (slope)	0.05	−0.13, 0.23	0.567
Common assault	Time (slope)	0.19	0.07, 0.32	0.003 **
	BN (step)	−4.84	−8.75, −0.94	0.015 *
	Time x BN (slope)	−0.21	−0.34, −0.08	0.002 **

Assault offences with an alcohol flag

Table 14 summarises the results of the time series models for each assault category for offences marked with an alcohol flag in Perth. For common assaults, the introduction of barring notices was associated with significant slope decrease (p < 0.001) and a step decrease (p = 0.005). For serious assaults, there was an overall decrease in serious assaults across the whole period but not necessarily associated with barring notices. There were no significant findings relating to the number of woundings marked with an alcohol flag in the Perth region.

Table 14. Time series models for alcohol flagged assault offences recorded in Perth: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.032	−0.069, 0.005	0.091
	BN (step)	0.282	−0.772, 1.337	0.597
	Time x BN (slope)	0.021	−0.017, 0.059	0.285
Serious assault	Time (slope)	−0.221	−0.372, −0.068	0.005 *
	BN (step)	−4.002	−8.26, 0.256	0.065
	Time x BN (slope)	0.137	−0.018, 0.292	0.083
Common assault	Time (slope)	0.197	0.028, 0.366	0.022 *
	BN (step)	−6.865	−11.61, −2.119	0.005*
	Time x BN (slope)	−0.321	−0.494, −0.147	<0.001 ***

Table 15 summarises the results of the time series models for each assault category for offences marked with an alcohol flag in the Fremantle area. For common assaults, there was a significant overall increase in recorded offences (p < 0.001), but the introduction of barring notices was associated with significant step and slope decreases (p < 0.001). Woundings and serious assaults both recorded step decreases (p < 0.05).

Table 15. Time series models for alcohol flagged assault offences recorded in Fremantle: 2007–2020

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	0.013	−0.013, 0.04	0.314
	BN (step)	−0.834	−1.581, −0.087	0.028 *
	Time x BN (slope)	−0.017	−0.045, 0.009	0.202
Serious assault	Time (slope)	0.011	−0.073, 0.095	0.792
	BN (step)	−2.991	−5.359, −0.624	0.013 *
	Time x BN (slope)	−0.046	−0.132, 0.04	0.294
Common assault	Time (slope)	0.176	0.088, 0.264	<0.001 ***
	BN (step)	−4.416	−6.889, −1.944	<0.001 ***
	Time x BN (slope)	−0.223	−0.313, −0.132	<0.001 ***

Assault offences, with family cases removed

Table 16 summarises the results of the time series models for each assault category, for all offences excluding family cases. For woundings in the Perth region, there was an overall decrease across the period (p < 0.05), but not associated with barring notices. For serious assaults, there was a significant slope increase after the introduction of barring notices (p < 0.001), but a significant overall decrease across the period examined (p < 0.001). For common assaults, there was a step decrease immediately after the introduction of barring notices (p < 0.05).

Table 16. Time series models for assault offences with family cases removed, recorded in Perth: 2007–2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	−0.043	−0.084, −0.003	0.034 *
	BN (step)	0.345	−0.794, 1.485	0.55
	Time x BN (slope)	0.032	−0.009, 0.074	0.123
Serious assault	Time (slope)	−0.389	−0.596, −0.183	<0.001 ***
	BN (step)	−5.614	−11.406, 0.178	0.057
	Time x BN (slope)	0.378	0.167, 0.59	<0.001 ***
Common assault	Time (slope)	0.108	−0.118, 0.336	0.347
	BN (step)	−8.111	−14.49, −1.733	0.013 *
	Time x BN (slope)	−0.202	−0.435, 0.031	0.088

Table 17 summarises the results of the time series models for each assault category for offences recorded in the Fremantle area with family cases removed. For woundings, the introduction of barring notices was associated with a step effect reduction (p < 0.005). For common assaults, there was a significant overall increase in recorded offences across the period (p < 0.001), but significant step and slope decreases after the introduction of barring notices (p < 0.001).

Table 17. Time series models for assault offences with family cases removed, recorded in Fremantle: 2007 –2020.

Category		β co-eff	95% confidence interval	p-value
Wounding	Time (slope)	0.012	−0.015, 0.041	0.373
	BN (step)	−1.202	−2.002, −0.402	0.003 **
	Time x BN (slope)	−0.014	−0.043, 0.015	0.339
Serious assault	Time (slope)	−0.083	−0.193, 0.026	0.137
	BN (step)	−2.869	−5.964, 0.225	0.068
	Time x BN (slope)	0.054	−0.058, 0.167	0.341
Common assault	Time (slope)	0.183	0.093, 0.274	<0.001 ***
	BN (step)	−7.821	−10.361, −5.28	<0.001 ***
	Time x BN (slope)	−0.227	−0.319, −0.134	<0.001 ***

Sensitivity analyses

Due to the substantial impact of COVID-19, sensitivity analyses excluding the year 2020 were undertaken. The impact of the removal of time points from the year 2020 was negligible, resulting in no changes to the direction of the effects or to the significance of any results.

As the immediacy of the impact of patron bans was unknown, models without significant autocorrelation were tested using monthly lag values ranging from 0 to 4. These tests did not significantly change the direction or significance of any models tested and, as such, only the base models included in this paper are reported on.

Discussion

A consistent expectation associated with the use of patron banning provisions, such as the barring notices that are operational is WA, is a reduction in crime and a corresponding increase in community safety. The threat or reality of a ban is presumed to act to change the behaviour of both recipients and the community more broadly (Farmer, Curtis and Miller, 2018). With respect to the latter, Farmer et al. (2022) report a general absence of community knowledge and understanding of patron banning in WA, and contend that this limits the potential for more general deterrent effects. As such, while the analysis in this paper is not able to determine how many of the recorded assault offences related to recipients of a police-imposed barring notice, it is likely that any effects on recorded offences that can be attributed to police-imposed barring notices reflect more specific deterrent effects of the provisions – while a ban is active and/or after it expires. However, before discussing the analysis of recorded assault data in WA, it is acknowledged that it is not possible to definitively attribute specific changes or reductions in certain types of offending solely to the introduction of barring notices. The complexity of determining the precise cause/s of behaviour change has been established across a number of research and operational contexts – particularly those that reflect intersections of individual, situational, regulatory and operational factors (Treno et al. 2008, Wells et al. 2008, DeWall, Anderson and Bushman, 2011, Miller et al. 2016a; Eckhardt and Parrott, 2017). Policy responses to problematic behaviours in the night-time economy typically comprise multiple provisions and it can be difficult to discern/isolate specific effects (Miller et al. 2019). As such, care is taken here to discuss the findings in terms of possible associations rather than specific correlations. Nevertheless, this paper represents a notable addition to the evidence base in relation to the potential effects of patron focused interventions.

It is recognised that the data provided by WA Police covers all assault offences. Not all of these incidents will fit with the expected scope of barring notices, the most notable example being violent acts in the home (Miller et al. (2016b) found that around 60% of alcohol related violent offences in the home were not reported to police). To mitigate the risk of skewed findings or results that cannot reasonably be expected to be affected by the barring notice provisions, the data that was provided by WA Police intentionally included assault offences that were recorded during high alcohol hours [HAH] (Morgan et al. 2018, Nepal et al. 2019). The analysis then used specific locational factors alongside the additional family and/or alcohol-related flags to discern the offences of most relevance to the context in which barring notices are imposed. The results were presented for WA as a whole. Additional analysis was also conducted for offences that were recorded in the Perth and Fremantle areas as the geographical locations with the most populous nightlife entertainment districts.

The findings identified a number of indicators of potentially positive effects of barring notices on the number of recorded assault offences. The following categories recorded significant reductions in recorded offences immediately after the 2011 introduction of the barring notice provisions (where p < 0.001 or p < 0.005): non-family assault offences (p < 0.005); non-family assault offences marked with an alcohol flag (p < 0.001); assault offences recorded as taking place on pathways (which includes areas around licensed premises) (p < 0.001); common assault offences marked with an alcohol flag which took place in the Perth area (p < 0.005); common assault offences marked with an alcohol flag which took place in the Fremantle area (p < 0.001); common assault offences excluding family incidents which took place in the Fremantle area (p < 0.001); wounding offences excluding family incidents which took place in the Fremantle area (p < 0.005).

In addition, for the following categories, recorded offence numbers decreased for the rest of the period following the introduction of the barring notice provisions (2011 – June 2020 – where p < 0.001 or p < 0.005): common assault offences marked with an alcohol flag (p < 0.001); common assault offences marked with an alcohol flag recorded in the Perth region (p < 0.001); common assault offences in the Fremantle area (p < 0.005); common assault offences marked with an alcohol flag recorded in the Fremantle area (p < 0.001); and common assault offences, excluding family incidents, recorded in the Fremantle area (p < 0.001). One category of relevance recorded a significant increase in recorded offences across the period after the introduction of barring notices: serious assaults, excluding family cases, in the Perth region (p < 0.001).

The most significant reductions (immediate and ongoing) following the introduction of barring notices related to non-family assault offences marked with an alcohol flag; common assault offences marked with an alcohol flag; and assault offences recorded as taking place on pathways (which includes areas around licensed premises). All reflect behaviours that are directly relevant to the purpose, location and intended application of barring notices. The timing of the substantial decrease in non-family assault offences coincides with a smaller rise in family assault offences, and this could suggest that the drop in non-family assaults is due partially to a change in police reporting practices. However, this would not account for all of the recorded drop, but this cannot be assessed further from the current dataset.

Notwithstanding the accepted limitations regarding the use of alcohol flags within police records (Morgan et al. 2018, Nepal et al. 2019), the clear association between offences marked with an alcohol flag and reductions in recorded assault offence numbers following the introduction of barring notices is a notable finding. Previous patron banning research, while admittedly limited, has been unable to establish such a clear association. Sogaard (2018) found that banning provisions in Denmark increased the risk of violence and other undesirable behaviours. In Australia, Taylor et al. (2021) examined the effectiveness of police-issued ten-day bans applied across Queensland’s three largest night-time entertainment precincts (Brisbane CBD, Fortitude Valley and Surfers Paradise CBD). The research analysed banning notices imposed on particular weekends and compared the quantum of specific offences recorded on subsequent weekends. The study focused on immediate rather than ongoing effects, but the authors found no significant relationship between the utilisation of bans and the number of serious assaults, common assaults or good order offences over the following weekend. There was no evidence to indicate that police bans were effective in reducing alcohol-related harm in the short-term. One perceived issue was the lack of real effect afforded by a ten-day ban – as it would typically only cover one weekend. Miller et al. (2019) recommended extending the permissible length of an initial banning notice to a maximum of one month, rather than ten days. This recommendation was approved, in principle, by the Queensland Government (2019). Part 8 of the Criminal Code (Consent and Mistake of Fact) and Other Legislation Amendment Act 2021 then amended the Police Powers and Responsibilities Act 2000 to permit initial police banning notices to be imposed for up to one month. Any effects of this change of recorded offence numbers has not yet been examined.

Differences in the scope and operation of provisions can make it difficult to directly compare the effects of patron banning across jurisdictions. However, one of these differences may help to explain the more positive findings identified in WA – where barring notices are typically imposed for longer periods than in most other Australian jurisdictions (Farmer, Curtis & Miller, 2018, Farmer & Clifford, 2020). In WA, barring notices can be imposed for up to 12 months, compared with the ten days in Queensland, or 72 h in Victoria. Farmer et al. (2023) reported that mean lengths of barring notices ranged from 4.46 months in 2015–6.82 months in 2019. Longer initial bans (of 6–12 months) were associated with a lower likelihood of a subsequent ban, which provides one measure of the potential specific effect of barring notices on recipient behaviour.

For the current study, the most significant longer-term/ongoing reductions in offence numbers related to common assault offences marked with an alcohol flag (overall, in Perth and in Fremantle), serious assault offences excluding family incidents in Perth and Fremantle, and common assault offences excluding family incidents in Fremantle. While the effects of other factors on offending numbers cannot be excluded, these categories clearly align with the intended use and effect/s of barring notices, and these findings are encouraging. There were no significant findings associated with wounding offences – despite a number of high-profile ‘glassing’ incidents in/around licensed venues being one of the key drivers for the implementation of the barring notice provisions (Legislative Assembly (WA), 2010). It is acknowledged that some offence categories recorded significant increases across the period examined. This included: common assaults marked with an alcohol flag (p < 0.005); common assaults in Fremantle in general, those marked with an alcohol flag and with family cases removed (p < 0.005; p < 0.001; p < 0.001). There is no theoretical reason why the introduction of barring notices would result in an increase in crime. It is reiterated that this study has focused on potential associations between patron banning and recorded levels of offending – other factors can affect crime rates as well.

The use of patron banning across multiple jurisdictions has been subject to criticism regarding its presumed effectiveness, absence of sufficient individual oversight/review options, discretionary and summary imposition mechanisms, the potential for misuse, and the general lack of ongoing scrutiny of the use of banning provisions (Farmer 2018, 2019, Farmer, Clifford & Miller, 2021). The imposition approach adopted in WA addresses most of the concerns about the use of patron banning that have been set out in previous research. Barring notice decisions are not made on-the-spot by a patrolling officer – determinations are made centrally following consideration of relevant information and evidence. Barring notices are then served in person, usually at the home of the recipient, by a police officer sometime after the event. Unlike other Australian jurisdictions, where bans are typically imposed on-the-spot, recipients in WA are less likely to be intoxicated or in an otherwise heightened state when receiving the ban, and may have a better chance of understanding the rationale and potential consequences (it is acknowledged that these are assumptions that have not been subject to specific research). Despite these potentially positive operational differences, from the findings of this study it is possible to identify a number of refinements which may further improve the operation and effectiveness of barring notice provisions in WA.

In common with most of Australia, family violence incidents continue to increase. This trend is evident within the dataset examined for this paper. Since 2007, WA Police have been able to seek the imposition of prohibition orders, another type of patron ban which can exclude recipients from licensed premises, and wide geographical areas, for up to five years. As there is no specific legislative requirement for prohibition orders to be linked to behaviours in/around licensed venues (Liquor Control Act 1988, Part 5A), WA Police now seek their imposition in response to alcohol-related family violence incidents. Given the association between alcohol consumption and family violence (Leonard, 2005, Mayshak et al. 2020), it is suggested that WA consider amending the Liquor Control Act 1988 (s.115AA) to expand the permissible scope for the issuing of a barring notice to include alcohol-related offences that do not necessarily occur in or around a licensed venue. This could include assaults in other locations, including in/near the home, and would have the effect of preventing the in-person purchase of alcohol by the recipient for the period of the barring notice. This change would bring barring notice imposition in line with prohibition orders – but with a potentially more immediate effect as the latter are imposed under the auspices of the Department of Racing, Gaming and Liquor and decisions typically take longer than for barring notices. In addition, while a prohibition order determination is being made, an on-the-spot barring notice could be considered for family violence situations (to cover the period of the prohibition order decision). It is acknowledged that this would mark a change from the current barring notice imposition procedures. However, there are notable potential benefits as an immediate response to family violence incidents. Any concerns regarding the discretionary imposition of a barring notice in specific situations or the potential misuse of barring powers would be subject to immediate and automatic review as part of the process of deciding the merits of a prohibition order.

This paper has presented findings relating to state-wide data. Other than drawing out key data for the major cities of Perth and Fremantle, the paper has not differentiated between regional locations across WA. There are acknowledged differences in offending behaviours, trends and policing priorities, particularly when comparing more remote regional areas and the metropolitan centres of Perth and Fremantle. For example, a voluntary Banned-Drinkers Register [BDR]³ (whereby people are not permitted to purchase or possess alcohol following repeated alcohol offending) using ID scanning technology for take-away alcohol is being trialled in three regional areas of WA (Kimberley, Pilbara and Goldfields).⁴ However, the BDR is not currently operational in Perth or Fremantle where the state’s major nightlife districts are located. These differences sit outside of the scope of this paper, and will be addressed in a separate and more focused examination.

Conclusion

There are a number of implications arising from this study and some notable limitations have been acknowledged. Nevertheless, the association between the introduction of barring notices and a reduction in alcohol-flagged violent offending appears to endorse the use of barring notices in WA. There are no notable issues identified with respect to their operation, although a number of suggested refinements may further enhance the operational use and potential beneficial effects of the barring notice provisions.

Acknowledgements

The authors wish to acknowledge the support and assistance provided by WA Police in undertaking this research. WA Police Research Governance has approved the publication of this paper. The views expressed in this publication are not necessarily those of WA Police and any errors of omission or commission are the responsibility of the author/s.

Data availability statement

The datasets analysed during the current study are not publicly available as the data is held by WA Police.

Disclosure statement

Professor Peter Miller has received funding from the Australian Research Council (ARC) and Australian National Health and Medical Research Council (NHMRC), grants from NSW Government, National Drug Law Enforcement Research Fund, Foundation for Alcohol Research and Education (FARE), Cancer Council Victoria, Central Australian Aboriginal Congress, Northern Territory Government (NTG), Australian Rechabites Foundation, Northern Territory Primary Health Gaming and Racing and the Australasian Drug Strategy Conference. He has acted as a paid expert witness on behalf of a licensed venue and security firm.

Additional information

Funding

This research was funded by WA Police.

Notes

1 As established by Coghlan, Sutherland & Millstead (2016).

2 For context and relevant research, Morgan et al. (2018) and Nepal et al. (2019).

3 In contrast to a mandatory BDR in the Northern Territory (Miller et al., 2022).

4 https://www.mediastatements.wa.gov.au/Pages/McGowan/2021/12/Goldfields-third-WA-region-to-trial-Banned-Drinkers-Register.aspx