• Open Access

Betting on the evidence: Reported gambling problems among the Indigenous population of the Northern Territory


Correspondence to:
Matthew Stevens, Charles Darwin University, Institute of Advanced Studies, School for Social and Policy Research, Darwin, NT, 0909. Fax: (08) 8946 7175; e-mail: matthew.stevens@cdu.edu.au


Objectives: To address a shortfall in evidence with which to justify gambling-specific interventions for the Indigenous population, we analysed two surveys (2002 National Aboriginal and Torres Strait Islander Social Survey and General Social Survey) that contain information on reported gambling problems for the NT.

Methods: Estimates of reported gambling problems are presented for each state and territory by remoteness for the Indigenous and total population for 2002. Factor analysis was used to identify the relationship between gambling problems and other negative life events for the NT Indigenous and total population.

Results: High levels of reported gambling problems were apparent for the Indigenous population particularly in the remote parts of the NT and Queensland. Gambling problems were associated with other stressors relating to social transgressions. Among the NT Indigenous population, gambling problems were correlated with levels of crowding, community involvement, personal and community violence and self-assessed health status.

Conclusions: The high levels of reported gambling problems suggest that gambling is causing significant problems for Indigenous people. The multivariable adjusted associations indicate that gambling-related problems are intimately connected to a range of community contexts.

Implications: Policies of intervention need to address broader social and environmental contexts that are intrinsically associated with gambling (and associated problems), in addition to public education in harm associated with gambling and provision of counselling services to assist problem gamblers.

In 2006 the Northern Territory Government (NTG) commissioned a report to investigate the incidence of child sex abuse in Aboriginal communities in the Northern Territory (NT). Rex Wilde, a respected QC, and Pat Anderson, a long serving administrator and board member of several Aboriginal non-government organisations, were commissioned with this task. In April 2007, they presented to the NTG the Little Children are Sacred (Ampe Akelyernemane Meke Mekarle) report.1 When the report was released publically some 60 days later, its conclusion that child sex abuse was evident in all 45 communities they had visited provoked a chorus of outrage around Australia. This reaction provided the rationale for the Federal Government to intervene in the management of Aboriginal communities in the NT in the form of the Northern Territory Emergency Response (NTER), with legislation passing in August 2007, to allow a number of interventions, including income management and banning of alcohol in all prescribed communities (Northern Territory National Emergency Response Act 2007).

More specifically, the report was seminal in that it positioned gambling as a key public health concern, one that endangered the wellbeing of Aboriginal children. The report mentioned the word ‘gambling’ 36 times, and in every instance except one, used the term in conjunction with alcohol, drug abuse and pornography.1 Thus, gambling was positioned within a discourse of vice and pathology. Twelve months later, the NTG released a policy response to the recommendations contained in the Wild and Anderson report titled Closing the Gap on Indigenous Disadvantage: A Generational Plan of Action.2 This document outlined a number of gambling-specific responses that correlated with specific recommendations by Wild and Anderson (see Table 1). The policy response was to specifically target gambling as a discrete, harmful activity, and to provide education and counselling to ameliorate its negative impacts.

Table 1.  Recommendations from the Wild and Anderson (2007) and the Northern Territory Government (2008) response in the Closing the Gap report.
No.RecommendationNTG Response
88Undertake an education campaign targeting the impacts and risk of gambling and especially the risk posed to children who are unsupervised while parents are gambling.The NTG will undertake a wide-spread education campaign around the negative impact of gambling, including the risks to children and families.
  • Additional Cost: $1.25m over five years [shared with pornography education campaign]
89That options for delivering gambling counselling to Aboriginal communities be explored and implemented including consideration of visiting counsellors for smaller communities and resident counsellors for larger communities.The NTG will investigate the extension of current gambling counselling programs to cover more Indigenous communities (Recommendation 90).
90That further research be carried out on the effects of gambling on child safety and wellbeing, and that consideration be given to the enactment of local laws to regulate gambling as part of the community safety plans to be developed pursuant to recommendation 79 [see below].The NTG will undertake further research on the impact and management of gambling on Indigenous communities by June 2008. This research will include the:
 • Undertake further research on the effects of gambling on child safety and wellbeing.• effect of gambling on child safety and wellbeing,
 • Enact local laws to regulate gambling as part of the community safety plans.• adequacy and enforcement of current laws regulating gambling, and
  • potential for effective gambling counselling programs.

We wish to make two points with regard to this policy response. First, there is an almost complete lack of recent empirical evidence about the effects of gambling, either regulated gambling (i.e. casino games, keno, poker-machine play and TAB betting) in the urban centres or unregulated gambling (i.e. card-games) in Indigenous communities. The research that does exist, described here, presents a somewhat mixed picture of gambling, one that is more complex, and positive, than its positioning purely within a discourse of pathology would suggest. Thus the response by the NTG represents an a priori judgement that gambling is intrinsically harmful for Indigenous people, one that is at this stage not based on reliable evidence. Second, the policy response focuses specifically on gambling itself as an independent and discrete phenomenon. In reality, gambling is a complex phenomenon that is intrinsically intertwined with a range of social forces that reproduce it as a social practice.3,4 The idea that its (negative) effects may be managed through gambling-specific education and counselling alone, is more likely a reproduction of existing policy orthodoxies rather than a considered response to the issue (for example see references 3, 5–7). The need to understand the broader contexts in which gambling occurs is an area of growing research and highlights the role of the environment in which gambling opportunities exist.3,4,7

In this context, this paper makes an empirical contribution to the stock of knowledge about gambling by the Indigenous population. It analyses the 2002 National Aboriginal and Torres Strait Islander Social Survey (NATSISS) and the 2002 General Social Survey (GSS) to examine the level of gambling-related problems among the Indigenous population of Australia for each jurisdiction. It then focuses on the NT to identify the social and demographic variables that are associated with reported gambling problems. The analysis will enable us to answer three key questions. First, is gambling a problem in the Indigenous population, second, how are gambling-related problems positioned in the context of other negative life events, and third, which variables are associated with these problems? Identification of the factors associated with reported gambling problems will provide the evidence with which to develop effective policy responses.

‘Aboriginal’ gambling – the unregulated card game

The reason that gambling is often separated into ‘Indigenous’ and ‘non-Indigenous’ or ‘regulated’ or ‘unregulated’ relates to the type of gambling activity and where it occurs.8,9 More specifically, Aboriginal gambling in Australia has historically consisted of card games (i.e. unregulated gambling), which are an integral part of the social and economic activity of community life. While the types of games, numbers of players, and betting systems vary, card games are invariably located outside of formal western gambling space. Therefore, they fall outside the government-controlled fiscal structures that define regulated gambling, and are not easily amenable to formal mechanisms of governmental monitoring and control.

In contrast to the ‘social vice’ policy discourse, early anthropological research into card games revealed a relatively positive picture.10–12 These studies found that card games enabled the equitable distribution of money at a time when cash resources were scarce, likening gambling with a cash form of hunting and gathering. In addition, these studies noted that the social aspects of gambling played an important function with regards to the principles of sharing and reciprocity that exist in Indigenous social systems.11,12 In other words, card-playing was a good fit into community life, one that could operate within existing socio-economic ideologies and structures of resource distribution. This research has indicated that much of the card game gambling on communities is non-problematic because: (i) the money usually stays in the community, (ii) it allows for the raising of larger amounts of money for ‘big ticket items’ to be purchased (e.g. fridges, boats etc.), (iii) it passes time and relieves people from the monotony of community life where there are limited opportunities for employment, and (iv) card-games have a primarily social function.

However, other research from the same time, but conducted in north-west Australia, tended to emphasise the negative over the positive impacts of community card games. For example, Hunter and Hunter and Spargo emphasised the negative effects of card games on child nutrition, household finances and family relations.13,14 This research also found that gamblers in the Western Australian study communities experienced significantly higher anxiety levels compared with non-gamblers in the same communities. In a similar vein, research by McKnight on Mornington Island, Queensland (QLD), emphasised the broader community contexts in which gambling exists.15 Based on periodic visits to the community over 40 years, McKnight used the terms ‘squanderers’ (i.e. drinkers and gamblers) and ‘non-squanderers’ (i.e. non-drinkers and non-gamblers) to describe the positive-negative dichotomy of behaviours that had developed in the community. Importantly, the social problems were not attributable to any single factor (though excessive alcohol consumption was the most significant problem), but was a result of the many interconnected factors including a lack of housing, low income, few employment opportunities, and regular gambling (card games) by a majority of residences for money or alcohol.

The most recent research has come from key informant interviews conducted by Wild and Anderson and these suggest that card games are causing harm in some instances, with problems relating to children being neglected physically and emotionally when their mothers were gambling (i.e. playing card games) for long periods of time, in addition to the economic cost of losing money that could be spent on other essentials.1 An earlier 1999 report on gambling in the NT noted that some key informants interviewed identified card games as positive, emphasising the social and distributive aspects of these games.16 These studies suggest that card-games have both positive and negative consequences that manifest to different degrees depending on a complex range of social, political and economic factors. Any discussion of gambling must acknowledge that as an activity, its function or worth will be dependent on a range of contexts that exist in the broader community; a situation that has also been noted in public health literature related to the use of alcohol.17,18

Recent evidence on regulated gambling in the Northern Territory

Regulated gambling falls under the fiscal control of governments and as an activity is played in licensed venues. Activities include poker machines, lotto and instant scratch tickets, racetrack betting (e.g. TAB or oncourse), sports betting (e.g. TAB or online) and casino games (e.g. table games and keno). There has been little research into regulated gambling among the Northern Territory Indigenous population since the expansion of poker machines into clubs and pubs in the mid-1990s. Foote conducted an observational study of Aboriginal patrons in the Darwin casino over a six-week period and found that there was a trend of increased patronage by Aboriginal people, particularly women, with a clear preference for poker machines (i.e. 66% of identified Aboriginal patrons were women and 76% of Aboriginal people observed were playing poker machines).19 Foote noted that the new forms of gambling available to Aboriginal people and increasing attendance at the casino pointed towards a transition in gambling from unregulated (i.e. card games) to regulated gambling (i.e. casino games, poker machines, keno and TAB). McMillen and Togni noted some negative aspects of gambling were identified in interviews with key informants from communities, such as inadequate care for children and players and their families going without food, though most negative comments related to poker machines being identified as the cause of problems because money was leaving the communities instead of circulating within them.16

Most recently, the Northern Territory Gambling Prevalence Survey 2005 provided some information on gambling participation and the levels of problem gambling by both the Indigenous and non-Indigenous population.8,9,20 Estimates from this survey indicated that in the general population of the NT, the prevalence of problem gamblers was not significantly different to other jurisdictions in Australia.8 However, according to the two screens used, problem gamblers were two to three times more common in the Indigenous compared with the non-Indigenous (unweighted) sample (i.e. 4.0% v 1.9% and 7.9% v 2.5%).9 However, the estimates produced by this telephone survey of problem gambling in the Indigenous population are heavily limited as only 27% of Indigenous households in the NT have a working telephone and these households are socio-economically better off than households with no phone and were more likely to be located in the larger urban centres of Darwin and Alice Springs.9,21 Consequently, the Indigenous sample was heavily biased towards the urban Indigenous middle-class and generalising this estimate to the total NT Indigenous population is problematic and inevitably inaccurate. The extent of problem gambling within the Indigenous population of the NT is therefore not revealed by this survey. However, if a comparison is drawn with the general population, where problem gambling is higher among the socio-economically worse off, then problematic gambling among the Indigenous population may be well above the levels reported above. Thus, the gaps in our knowledge about the effects of both regulated and unregulated gambling are comprehensive. While individual studies in particular communities have yielded valuable information, we are lacking any assessment of the level of gambling, its consequences, or its associations with other social conditions.22 As a first step in filling this gap, the analysis presented below uses data from the 2002 NATSISS to:

  • 1Describe the level of reported gambling problems among the Indigenous population and the total population for each state/territory by remoteness.
  • 2Examine how gambling-related problems positioned in the context of other negative life events in the NT Indigenous and total population.
  • 3Identify variables showing an independent correlation with reported gambling problems.


Australian Bureau of Statistics data sets

The 2002 NATSISS and the 2002 GSS were analysed to assess independent correlates of reported gambling problems for the Indigenous and general population of the NT. Both surveys select private dwellings using a multi-stage area sample. The NATSISS samples remote and non-remote areas, while the GSS samples non-remote areas only. The surveys are designed to provide robust estimates at the state, territory and national levels by remoteness where the survey included a remote sample. For more detail on the survey methodology please refer to the respective technical papers.23,24

The Negative Life Events Scale

The information on gambling problems is collected as part of the Negative Life Events Scale (NLES), which is a module used in both surveys. The NLES is a measure of emotional and social wellbeing that is intended as a reliable measure for both the Indigenous and non-Indigenous populations, and has also been used in the 2002 GSS.24 The scale was developed by the Australian Bureau of Statistics (ABS) for use in the 2002 NATSISS. The survey development and the items for inclusion were guided by a special advisory group comprising of experts in Indigenous information, research and cultural issues, whose members were nominated “from Indigenous organisations, peak Indigenous information bodies, Commonwealth and state/territory government agencies with Indigenous program responsibilities, and relevant academic research institutions”.23 The scale's development was founded on earlier work in social psychology where negative life events were conceptualised as ‘stressors’, though earlier screens included both positive and negative life events.25,26 In addition to the NLES being used in ABS surveys, a slightly modified version of the NLES showed good psychometric properties when used in a research project in 10 Indigenous communities in the NT.27,28

The NLES, as operationalised in ABS surveys, poses the question: “Have any of the following [list of stressors shown] been a problem for you, your family or a close friend in the past 12 months?” Respondents are then presented a list of up to 16 ‘stressors’ or negative life events (see below). While there are some differences in the listed NLES stressors between remote and non-remote locations and between the GSS and the NATSISS, 11 of the stressors are collected in a similar way in both surveys for the remote and non-remote samples to be compared. They are: gambling problems, alcohol or drug related problems, witness to violence, abuse or violent crime, trouble with the police, divorce or separation, not able to get a job, lost job, made redundant or sacked, death of family member or close friend, serious illness or disability and serious accident by family member. A unique feature of the NLES is that it does not measure problems at the individual level, but rather it measures the extent that individual stressors affect individuals and their social and family networks. This broader conception of gambling problems is consistent with the national definition of problem gambling, which states problem gambling “…is characterised by difficulties limiting money and/or time spent on gambling, which leads to adverse consequences for the gambler, others or the community.”29

Data access and variables

The NATSISS and GSS collect person and household level information across a range of dimensions including demographic, family and community, health, socio-economic status and financial stress, housing, culture and leisure, transport, communication and crime. There were some differences in the data items collected between the two surveys. Fortunately, a large number of variables were available that were equivalent or very similar in both surveys, which are used in the current analysis. The main outcome variable of interest, reported gambling problems is a binary variable indicating whether the respondent or their family or close friends had experienced gambling problems in the 12 months prior to the survey. Data were accessed via the ABS web-based data portal known as the Remote Access Data Laboratory or RADL.30 Ethics clearance was not required as the analyses constitute secondary use of ABS data. However strict confidentiality restrictions were respected as mandated by the Census and Statistics Act, 1905.

Statistical analysis

First, estimates of gambling problems are reported by jurisdiction and remoteness for the NATSISS and GSS. NT estimates for the 11 NLES items are presented with relative risks (95% confidence intervals) to identify significant differences between remote and non-remote estimates of gambling problems for the NATSISS as well as between the non-remote estimates between the NATSISS and the GSS. Relative risks show the ratio between estimates of reported gambling problems for two groups (significant at p<0.05 when confidence interval does not include one). To explore inter-relationships between NLES items, a factor analysis (principle component factor method) was carried out on the 11 NLES items common to the GSS and NATSISS. Factors with Eigen values greater than one were retained and an orthogonal rotation applied to facilitate interpretation of the factor structure. Variable loadings greater than or equal to 0.4 for NLES items were considered sufficient to place each item on a particular factor. The results from the factor analyses for the GSS and NATSISS were compared and contrasted for consistencies and differences in factor structure.

Logistic regression was used to assess multivariable associations between reported gambling problems and explanatory variables (i.e. demographic, socio-economic, socio-cultural, health and community wellbeing, culture and leisure and crime). The same analytic strategy was applied to the GSS and NATSISS. This first involved identifying significant unadjusted (or bivariate) associations between explanatory variables and reported gambling problems. Subsequently, all variables showing a moderate statistical association (p≤0.10) were entered into a logistic regression model and backward selection carried out until all variables remaining in the model were significant at p≤0.05. Other NLES items were not included as explanatory variables in the multivariable modelling as these variables are likely to have significant correlations with reported gambling problems and would reduce the ability of the modelling to identify other important correlates. The logistic regression analyses produced two models, one for the Indigenous adult population of the NT (NATSISS), and another for the total adult population (Indigenous and non-Indigenous) living in non-remote parts of the NT (GSS).

All data were analysed using Stata v8.2© (accessed via the ABS RADL) and confidence intervals were calculated using the jack-knife method (with 250 and 30 replicate weights for the NATSISS and GSS respectively), while adjusting for survey design and weighting using the suite of SVR commands available for use in Stata v8.2.31


Reported gambling problems by jurisdiction

Table 2 presents reported gambling problems by remoteness and jurisdiction for the 2002 NATSISS and 2002 GSS. A comparison of the NATSISS with the GSS for non-remote areas reveal gambling-related problem estimates two to five times higher among the Indigenous population compared with the general population across all states and territories. In non-remote areas of the NT, the estimate of gambling problems was 3.2% in the general population, compared with 11.4% among the Indigenous population, a level more than three times higher. In remote areas, the level of gambling related problems among the Indigenous population is significantly higher than in non-remote areas throughout Australia and this is particularly so for the NT and QLD. While there are some limitations of the current data (discussed below), the findings do indicate that Indigenous people are experiencing substantial problems from gambling in both urban and remote parts of the NT and across Australia more broadly. These problems are most prevalent in remote areas.

Table 2.  2002 NATSISS and GSS reported gambling problems by remoteness.
 2002 NATSISS2002 GSS
 RemoteNon-remoteTotalRR (95% CI)aNon-remoteRR (95% CI)b
  1. Notes:

  2. a) Relative risk for NATSISS remote:non-remote and 95% confidence intervals.

  3. b) Relative risk for non-remote NATSISS:GSS and 95% confidence intervals.

  4. c) Combined ACT and Tasmania estimate due to smaller sample and confidentiality restrictions on the CURF.

  5. Sources: ABS 2002 NATSISS (all estimates from ABS data cubes except ACT/Tasmania estimate which was obtained from the 2002 NATSISS Confidentialised Unit Record File (CURF) and 2002 GSS CURF accessed via the RADL (Australian Bureau of Statistics, 2006).

WA13.2 (2.9)3.6 (1.2)8.1 (1.5)3.67 (1.87-5.47)1.4 (0.2)3.03 (1.47-4.59)
NSW8.7 (2.3)10.3 (1.3)10.2 (1.2)0.84 (0.46-1.23)4.1 (0.4)2.73 (2.27-3.19)
Vic-13.3 (1.6)13.3 (1.6)-3.7 (0.4)3.86 (3.53-4.18)
Qld37.1 (10.7)10.7 (1.6)17.4 (2.9)3.47 (1.79-5.14)3.1 (0.4)3.52 (3.03-4.02)
SA19.3 (5.4)16.5 (2.3)17.2 (2.2)1.17 (0.61-1.73)3.5 (0.5)5.06 (4.55-5.56)
NT31.9 (4.1)11.4 (2.9)28.4 (1.4)2.87 (1.59-4.16)3.2 (0.6)3.32 (2.23-8.41)
ACT/Tasc-7.9 (1.1)7.9 (1.1)-3.1 (0.3)2.55 (2.09-3.01)
Australia26.4 (3.2)10.2 (0.7)14.6 (1.0)2.59 (2.08-3.10)3.5 (0.2)3.19 (2.91-3.47)

NLES estimates and factor analysis

Estimates and relative risks for the 11 NLES items by remoteness for the NT are presented in Table 3. Relative risk 95% confidence intervals with ranges either side one indicate a significant difference (p≤0.05) between the NATSISS and GSS estimates for reported gambling problems. The estimate of reported gambling problems for the NT from the NATSISS was 28.9% (31.9% in remote and 11.4% in non-remote), a similar level to alcohol and drug related problems and witness to violence. Overall reported gambling problems were the fifth highest estimate for the NLES items for remote areas and seventh for non-remote areas. For the GSS, the estimate of reported gambling problems was 3.2%, which placed it as the lowest NLES estimate. Only three NLES items, divorce or separation, not able to get a job, and lost a job, had significantly lower estimates in remote regions compared with non-remote regions for the NATSISS. All other NLES items from the NATSISS were reported at significantly higher levels in remote areas compared with non-remote areas. Comparing the non-remote estimates between the NATSISS and GSS, five items, gambling problems, alcohol or drug problems, trouble with police, not able to get a job, and death of a family member, were all significantly higher in the non-remote Indigenous population (i.e. between 1.5 and 3.5 times higher than in the general population).

Table 3.  Estimates of reported gambling problems and relative risks for NLES items for the Indigenous and general population of the NT by remoteness.
 2002 NATSISS2002 GSS
NLES itemRemote % (SE)Non-Remote % (SE)Total % (SE)RR (95% CI)aNon-Remote % (SE)RR (95% CI)b
  1. Notes:

  2. a) Relative risk (Indigenous remote:Indigenous non-remote) and 95% confidence interval. Bold significant at p=0.05

  3. b) Relative risk (Indigenous:total population) and 95% confidence interval for non-remote sample. Bold significant at p=0.05

Gambling problem31.9 (4.1)11.4 (2.9)28.4 (3.4)2.80 (1.59-4.00)3.2 (0.6)3.56 (2.36-4.76)
Alcohol or drug related problems36.9 (4.2)19.5 (3.5)33.9 (3.6)1.89 (1.38-2.41)8.9 (0.9)2.19 (1.55-2.83)
Witness to violence33.9 (3.8)8.7 (2.7)29.6 (3.2)3.90 (1.69-6.11)5.7 (0.7)1.53 (0.67-2.38)
Abuse or violent crime12.9 (2.5)9.0 (2.8)12.3 (2.1)1.43 (0.75-2.12)5.4 (0.7)1.67 (0.74-2.59)
Trouble with the police15.5 (3.1)12.7 (3.0)15.0 (2.6)1.22 (0.92-1.52)5.3 (0.8)2.40 (1.54-3.25)
Divorce or separation9.2 (1.9)15.0 (3.1)10.2 (1.7)0.61 (0.60-0.62)13.0 (1.0)1.15 (0.72-1.59)
Not able to get a job24.1 (4.0)34.4 (6.2)25.8 (3.5)0.70 (0.60-0.80)12.8 (1.1)2.69 (1.85-3.52)
Lost job, made redundant, sacked3.8 (1.1)8.0 (1.9)4.5 (0.9)0.48 (0.32-0.63)6.8 (0.8)1.18 (0.70-1.65)
Death-family member/close friend55.5 (3.7)43.7 (5.3)53.5 (3.2)1.27 (1.02-1.52)22.5 (1.5)1.94 (1.56-2.33)
Serious illness or disability36.8 (2.2)24.0 (4.3)34.7 (2.0)1.53 (1.03-2.04)20.9 (1.0)1.15 (0.76-1.54)
Serious accident19.6 (2.5)4.3 (1.9)17.0 (2.1)4.56 (0.78-8.34)5.4 (0.5)0.80 (0.12-1.47)

To determine the extent that these items are related, Table 4 presents the rotated three-factor solution for NLES items that were common to the NATSISS and GSS. The factor structure for both surveys was very similar, with only small differences in absolute loadings between the two surveys. The three-factor solution for the NATSISS and GSS explained 48% and 41% of the variance respectively. Factor 1 for both surveys represents items associated with social transgression, including trouble with police, witness to violence, experience of abuse or violent crime, alcohol and drug problems and gambling problems. Factor 2 generally represents problems relating to economic and social loss including not able to get a job, experience of divorce or separation and losing a job. Factor 3 represents health-related problems and their consequences including death of a family member, having a serious illness or disability and being, or knowing someone who has been, in a serious accident.

Table 4.  Rotated factor analysis of common NLES items for the NT.
NLES itemFactor 1Factor 2Factor 3Factor 1Factor 2Factor 3
  1. Note: Factor loadings ≥0.4 in bold font.

Gambling problems0.540.270.290.420.000.28
Alcohol or drug related problems0.570.340.270.680.160.11
Witness to violence0.720.250.110.730.060.04
Physical abuse or violent crime0.690.18-0.040.660.03-0.05
Trouble with the police0.70-
Divorce or separation0.250.610.060.100.500.16
Not able to get a job0.120.730.080.150.640.12
Lost job, made redundant, sacked0.030.550.030.080.72-0.07
Death of family member or close friend0.21-0.050.640.17-0.150.58
Serious illness or disability-
Serious accident0.33-0.060.460.020.070.68
Rotated Eigen-value2.321.561.381.961.361.23
Cumulative % variance explained21.1%35.3%47.9%17.8%30.2%41.4%

Multivariate models for reported gambling problems

Table 5 presents the multivariate models for reported gambling problems for the NATSISS and the GSS as well as estimates of reported gambling problems across the significant explanatory variables for the two surveys. The multivariate model using NATSISS data indicated that four groups of variables had significant multivariate adjusted associations with reported gambling problems. They were household type (structure and crowding), attending and/or participating in social and cultural activities, reported community problems, and self-reported health. Lone person households reported fewer gambling problems compared with one family households, while two and three family households reported more gambling related problems, particularly compared with lone person households. The social and cultural activities collected as part of the NATSISS were very similar in nature (e.g. attended a sporting event and spectator at a sporting event), so for the purpose of comparison with the GSS these variables were grouped together. Respondents who participated in social and cultural activities (or group activities) were around twice as likely to report gambling related problems compared with those that did not attend such activities. The cultural and social events associated with higher reported gambling problems included attending a sports carnival (odds ratio 2.22 (95% confidence interval, 1.35-3.64)) and having involvement in a special interest community event (1.94 (1.26-3.00)). Respondents who reported community problems including family violence and youth gang problems reported significantly higher gambling problems. This was particularly so for family violence community problems, with more than 50% of those reporting community problems also reporting gambling problems. In addition to community problems, respondents who experienced interpersonal violence or threatened violence also reported higher levels of gambling related problems among their family and social networks. Lastly, respondents with excellent health reported significantly less gambling related problems than those reporting very good health and fair health.

Table 5.  Multivariate models of reported gambling problems among the NT population.
 NATSISS: All NTGSS: Non-remote NT
Explanatory variablesOR (95% CI)Gambling problems % (SE)OR (95% CI)Gambling problems % (SE)
  1. Notes:

  2. a) Similar association between gambling problems and explanatory variable for the NATSISS and GSS

  3. b) Different association between gambling problems and explanatory variable for the NATSISS and GSS

  4. ns: Not significant (dropped out in backward selection)    na: Not applicable (variable not contained in the GSS)

Household type/structurea    
 One family1.0025.1 (3.8)ns2.8 (0.5)
 Two families1.17 (0.68-2.03)34.1 (6.5)ns5.7 (6.2)
 Three or more families1.78 (0.94-3.37)36.7 (7.3)nsna
 Mixed family and group0.77 (0.05-10.7)26.7 (20.8)ns6.3 (2.2)
 Lone person0.22 (0.07-0.76)7.6 (3.0)ns2.8 (0.8)
Household equivalised incomeb    
 Lowest income quintilens23.5 (4.3)1.007.8 (1.9)
 2nd quintilens38.1 (6.9)0.43 (0.10-1.89)3.2 (1.8)
 3rd quintilens31.0 (9.1)0.21 (0.07-0.60)1.7 (0.7)
 4th quintilens17.9 (8.2)0.32 (0.11-0.94)2.4 (0.9)
 Highest income quintilens3.9 (4.2)0.44 (0.20-1.01)3.3 (0.9)
 Highest income unknownns30.3 (5.3)na
Cash flow problemsa    
 No problemsns26.9 (4.2)1.002.4 (0.5)
 One problemns33.0 (5.2)2.97 (1.24-7.11)7.8 (2.3)
 Two or more problemsnana1.26 (0.29-5.37)4.7 (1.9)
Social networks & culture    
Recreational or cultural activitya    
 Did not participate inns24.0 (2.6)1.002.2 (0.5)
 Participated inns33.1 (5.1)3.25 (1.61-6.57)6.4 (1.7)
Community special interest groupa    
 Was not involved1.0021.1 (2.4)ns2.6 (0.7)
 Involved1.94 (1.26-3.00)47.5 (6.4)ns6.1 (0.9)
Spectator at sports carnivalb    
 Was not a spectator1.0017.7 (3.2)ns2.1 (0.7)
 Was a spectator2.22 (1.35-3.64)37.8 (4.3)ns4.1 (0.9)
Community youth gang problems    
 No problems1.0016.2 (2.2)na
 Problems2.22 (1.57-3.14)45.4 (4.8)na
Community family violence problems    
 No problems1.0017.1 (2.8)na
 Problems3.32 (2.25-4.89)52.4 (4.6)na
Victim of threatened/physical violencea    
 Not a victim1.0026.5 (3.7)1.002.2 (0.5)
 Was a victim1.77 (1.09-2.89)41.1 (4.5)3.24 (1.48-7.07)8.9 (1.8)
Self-assessed healthb    
 Excellent1.0017.2 (3.2)ns3.0 (1.0)
 Very good2.99 (1.56-5.73)42.1 (5.3)ns2.9 (0.8)
 Good1.49 (0.88-2.52)25.1 (4.2)ns2.6 (1.1)
 Fair3.52 (1.30-9.51)34.9 (8.2)ns4.6 (1.9)
 Poor2.26 (0.82-6.28)22.6 (8.9)ns11.3 (6.0)

The multivariate model using GSS data showed that three groups of variables had significant multivariate adjusted associations with reported gambling problems. They were variables relating to socio-economic status (including financial stress), participation in social activities, and individual experience of interpersonal violence or threatened violence. Respondents living in households whose household income was in the lowest quintile were more likely to report gambling problems compared with all other household income quintiles except those in the second quintile (or second lowest household income group). Respondents experiencing personal cash flow problems also reported more gambling related problems among their family and social networks. Again being a victim of threatened or physical violence and participation in social/cultural activities was associated with increased reporting of gambling problems.

Two variables displayed differential effects for the same explanatory variable between the two multivariate models. In the GSS analysis, the lowest income quintile was clearly associated with higher reported gambling problems, while for the NATSISS it is the second and third quintiles that reflect the highest reported gambling problems. The self-reported health variable also followed a different pattern, with those with the poorest health having the higher reporting gambling problems than all other groups in the GSS, while in the NATSISS people who responded that their health was very good or fair had highest reported gambling problems.


Caveats to the analysis

Before drawing firm conclusions from the analysis several potential limitations of the data and the analyses need to be outlined. First, the NATSISS and GSS differ in scope. Specifically, the NATSISS data was not able to be stratified by remoteness for the multivariable analysis because of data quality concerns associated with sample size and accuracy of the data, even though it was clear that estimates of reported gambling problems were significantly higher in remote areas.21 Therefore, while providing an estimate of gambling problems in remote areas, the analysis was unable to determine potential differences in the correlates of gambling problems by remoteness. Consequently, remoteness may be a confounding factor in the NATSISS analysis. For example, the finding that gambling is more problematic in multi-family households may reflect the high levels of crowding in remote Indigenous communities.32–34 Second, the GSS sample contained approximately 10% Indigenous respondents who were not identifiable in the data set used, and this group would have biased the GSS results towards the NATSISS results, though this would be negligible. Third, in comparing absolute levels of reported gambling problems between the NATSISS and the GSS, it may be expected that the Indigenous population would report higher levels because of the extended family networks associated with Indigenous kinship systems.35 Therefore, the higher levels of reported gambling problems may simply reflect the higher levels of interconnectedness and the increased chances of knowing someone with gambling problems or who has been affected by someone else's gambling. To assess the magnitude of this bias, comparison may be drawn with the study by Kowall et al. of housing and Indigenous child health in 10 remote Indigenous communities across the NT, a study that utilised a modified version of the NLES.27 These authors changed the NLES question to “have any of the following been a problem for you or anyone in this house in the past 12 months” (emphasis added). That is, the measurement of gambling problems was at the household level, rather than measuring the extent of gambling problem in the broader social and family networks. The estimate for the 10 remote communities was 28%, which is very close to the 31.9% reported in Table 2. This similarity suggests that the gambling problems reported by the Indigenous population of the NT compared with the general population are not just a result of the interconnectedness that is part of the Indigenous kinship system, but that gambling problems do occur in both remote and urban Indigenous populations at higher levels than the general population. Fourth, given there is significant variation in gambling participation between discrete Indigenous communities, factors such as community location or population size may also confound the analyses.9 Finally, Indigenous people are known to be highly mobile within jurisdictions and it is not known how this mobility interacts with gambling or gambling problems in remote communities.36 Specifically, it is unclear if problems are associated with commercially unregulated card games in remote communities or with people travelling to larger service centres to participate in regulated forms of gambling.

Gambling-related problems among the Indigenous population

The results of this study support the findings of recent research that report significantly higher gambling-related problems among the Indigenous population compared with the general population, and that problems are significantly higher in remote settings in the NT.8,16,19 The higher levels of reported gambling problems among the Indigenous population, not only in the NT, but across Australia, points to unregulated and regulated forms of gambling as causes of problems for individuals and their social and family networks. In the context of regulated gambling, it is well documented that poker-machine play is heavily associated with problem gambling both in Australia and overseas.8,37,38 Western Australia (WA) is a case in point. This state has no poker-machines outside of the Burswood casino and, not surprisingly, Indigenous people from WA reported the lowest levels of gambling problems compared with other jurisdictions. However, estimates of gambling problems in remote areas of WA, where there is no access to poker machines, were still significantly higher than in urban areas (13.2%cf. 3.6%). This suggests that card-games are associated with gambling problems in remote areas, though racetrack betting in TAB's is also available outside of urban locations in WA. More research is required to convincingly identify the forms of gambling that are most problematic in particular contexts. For example, if card games are causing problems such as neglect of children, then community responses such as the establishment of child minding facilities may go some way to minimising this potential harm.

Relationship between gambling-related problems and other negative life events

For both the NATSISS and GSS, the factor analysis clearly identified gambling problems as being associated with other indicators of social breakdown and transgression including witness to violence, trouble with police, abuse or violent crime, and alcohol and drug problems. It was less related to the other two dimensions of economic/social isolation and physical health. Therefore, gambling problems, at a community level, for both Indigenous and non-Indigenous populations may be addressed by initiatives that increase social function and wellbeing. However, while gambling problems are inter-related with a range of social breakdown/transgression issues, we were unable to assign causation due to the cross-sectional nature of the data. Further research using longitudinal data or through detailed ethnographic research is required to understand the relationships between stressors and their broader structural determinants. More broadly, the associations reinforce the importance of community contexts in mediating potential harmful effects of gambling. For example, in a large remote community in the Top End of the NT, Paterson notes that card games are socially regulated and there are particular rules and social norms associated with games that may be dependent on who is playing the game and the size of the stake.22 However, it is not known how other community contexts, such as the management of alcohol and drugs, mediate (or weaken) this social form of regulation. Given the increased presence of police in remote Indigenous communities in the NT and the introduction of alcohol bans, it could be anticipated that some of the related stressors affecting Indigenous communities may decline, though the effectiveness of the alcohol bans was questioned in the 12 month review of the NTER.39 What is remarkable is the similarity between the Indigenous and general population in the patterning of the negative life events. However, while the structure (or relationships between stressors) is similar, reported levels of stressors related to social transgressions were two to three times higher among the non-remote Indigenous population compared with the non-remote general population in the NT, and three to 10 times higher for the remote Indigenous population (see Table 3).

Correlates of reported gambling problems

The multivariate adjusted analyses revealed gambling problems to be significantly associated with social, family and community variables for both the Indigenous and general population of the NT. Specifically, the multivariate model for the NATSISS contained variables relating to demographic factors (household structure and crowding), social connectedness (participating in community social events and attending sporting events), community safety concerns (family violence, youth gang problems), being a victim of threatened or physical violence, and self-reported health. The multivariate model for the GSS showed some overlap of variables with the NATSISS analysis, with participation in recreational activities (social connectedness) positively associated with reported gambling problems and so to being a victim of physical or threatened violence. Other variables in the GSS analysis to show a significant independent association with gambling problems were the socio-economic variables of household income (lowest equivalised income quintile positively associated with reported gambling problems) and whether the respondent had cash flow problems in the past year.

The significant independent association between household structure and gambling problems illustrates the prevalence of multi-family households and high levels of crowding Indigenous people are exposed to across Australia.34 There has been a stream of reports on Indigenous housing in Australia since the 1990s and all these reports have identified the massive backlog of housing required, particularly in remote areas of Australia.32,33,40 Crowded living conditions have implications for both physical and mental health problems.28,41,42 It is encouraging to see a substantial amount of money (approximately $680 million) allocated to housing in remote communities in the NT as part of the NTER. However, if the number of new houses is to make a difference in the lives of Indigenous people then there needs to be a substantial reduction in crowding (persons per bedroom) before any concurrent benefits will begin to flow.28

The reason why people who appear socially connected (measured by attendance/participation in recreational, sport and social events) reported higher levels of gambling problems may be that these individuals have greater opportunity for gambling and therefore to report individual gambling problems. Alternatively, people attending these events may have larger social and family networks and thus are more likely to know people who have experienced problems because of gambling (either themselves or someone in their social and family networks). Unfortunately, there is no way of further elucidating this result, though the finding does suggest that places where people meet (e.g. sporting and community events etc.) may be good places to promote gambling awareness and educational campaigns. Furthermore, there is little evidence on what problems are associated specifically with gambling (and the type of activity) actually are (e.g. child neglect, not meeting social obligations, running out of money, etc). For example, if the problems are related to children not being adequately cared for because a carer's gambling, then improved support services for families (e.g. child minding facilities, safe houses) may go some way to alleviating problems to do with child neglect associated with gambling.

The associations between community violence (i.e. community family violence problems, community youth gang problems, and victim of threatened/physical violence) and gambling problems supports the factor analysis of the NLES in linking gambling problems with social breakdown and transgression. Once again we are unable to offer causal explanations until further research is conducted on the relationships between violence and gambling. However, the relationship between violence and gambling problems is likely to be mediated by a range of community-level factors. The present analysis was unable to determine if both remote and non-remote Indigenous communities experience similar community problems, or if the association between gambling problems and violence is more characteristic of remote or non-remote settings. Indeed, there exists significant variation between Indigenous communities across a range of social and environmental health dimensions.9,43 In the context of gambling, a comparison may be drawn between the study by Altman in a small community (outstation or homeland) in Arnhem Land, NT, and that by Hunter and Spargo in remote WA communities.12,14 The former found that card games were a social activity that assisted individuals to fulfil cultural obligations to kin, acted to distribute money at a time where cash was scarce, and allowed for larger sums of money to be raised when access to credit was limited, while the latter found gambling to cause significant social tensions and in some instances emotional and physical neglect of children. These studies were differentiated by the size of included communities and the presence of other community issues; predominantly the use of alcohol. Thus, the relationship between gambling problems and other social problems is likely to exhibit significant geographical variation that is heavily context-dependent.

The finding for the Indigenous population that people with both ‘fair’ and ‘very good’ health report more gambling problems than those reporting excellent health was surprising, particularly given the consistent trend observed in the GSS analysis (only those reporting the poorest health experienced higher levels of gambling problems). However, the finding from the NATSISS analysis may well be confounded by measurement error associated with the self-reported health question when used with Indigenous people, particularly those who speak English as a second or third language.44 Taken at face value, we would only be able to speculate about the relationships between reported health and gambling-related problems, which is an area that requires further exploration.


The estimates of reported gambling problems from the 2002 NATSISS and GSS were significantly higher among the Indigenous compared with the general population in all jurisdictions across Australia, and were higher still in remote locations. The statistical analyses conducted on reported gambling problems clearly indicate that gambling is embedded within a range of social, environmental, and geographic contexts. Given this web of associations, any effective policy response would need to move beyond a reactive concern with treatment and counselling proposed by the Closing the Gap report and address the living conditions of Indigenous people, particularly crowding in remote communities, and attempt to provide meaningful opportunities for social and economic inclusion and development. However, this should be done in addition to improved access to counselling and better public education being made available. What is needed is a multidimensional approach to minimising gambling-related harm.3,4,7 More specific recommendations will be dependent on further research that explores what constitutes safe gambling (particularly unregulated gambling in remote communities), who is at greater risk of being affected by the negative consequences of gambling (e.g. mothers of young children, pensioners etc.) and what the causal relationships between types of gambling, specific community contexts and gambling problems actually are.


This research was supported by a grant from the Community Benefit Fund, Northern Territory Government.