Content uploaded by Bruce Doran
Author content
All content in this area was uploaded by Bruce Doran on Nov 30, 2020
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=cage20
Australian Geographer
ISSN: 0004-9182 (Print) 1465-3311 (Online) Journal homepage: https://www.tandfonline.com/loi/cage20
Placing Bets: gambling venues and the distribution
of harm
Martin Young , Francis Markham & Bruce Doran
To cite this article: Martin Young , Francis Markham & Bruce Doran (2012) Placing Bets:
gambling venues and the distribution of harm, Australian Geographer, 43:4, 425-444, DOI:
10.1080/00049182.2012.731302
To link to this article: https://doi.org/10.1080/00049182.2012.731302
Published online: 10 Dec 2012.
Submit your article to this journal
Article views: 547
View related articles
Citing articles: 20 View citing articles
Placing Bets: gambling venues and the
distribution of harm
MARTIN YOUNG, FRANCIS MARKHAM & BRUCE DORAN,
Southern Cross University, Australia; Menzies School of Health Research,
Australia; Fenner School of Environment and Society, Australian National
University, Australia
ABSTRACT
The liberalisation of gambling in Australia has resulted in the dispersal of
200 000 electronic gaming machines (EGMs) across the country, generating substantial
revenues for State governments and the gambling industry as well as causing significant
gambling-related harm. While the spatial distribution of EGM venues has been shown to
follow a gradient of community disadvantage, little is known about the distinctions between
the venues themselves (i.e. pubs, clubs, and casinos), either in terms of the catchments they
service or the harm they produce. To this end, we constructed a sexpartite typology of EGM
venues in the Northern Territory of Australia derived from venue location and licensing
variables. We also conducted a geocoded mail survey (n7041) of households in three
urban centres to describe the composition of markets and problem-gambling outcomes across
the six venue categories in the typology. Venues in accessible locations and those with a
higher numbers of EGMs, particularly casinos and clubs located near supermarkets, were
most closely associated with gambling-related harm, even when differing player socio-
demographics were accounted for. We argue that gambling risk is a function of the
interaction of geographic accessibility to markets on the one hand and venue effects on the
other. An understanding of the geography of EGM gambling may help improve supply-side
approaches to regulation, as well as shed insights into contemporary urban processes within
Australia’s regional settlements.
KEY WORDS
Gambling; electronic gaming machines; spatial distribution; venues;
Northern Territory; gambling-related harm.
The liberalisation of electronic gaming machines (EGMs) or ‘pokies’ over the past
two decades has produced an extensive network of gambling venues that traverses
Australia. In 200809, there were 5683 gambling venues (i.e. hotels, clubs and
casinos) dispersed across nearly every settlement in the country, with the exception
of Western Australia that restricts EGMs to its sole casino (Productivity Commis-
sion 2010). Hotels, or pubs in the vernacular, are relatively small, ubiquitous
venues that are the equivalent in many respects of bars in other countries. Access is
freely available to the public. Clubs, on the other hand, are larger sporting,
Australian Geographer, Vol. 43, No. 4,
pp. 425444, December 2012
ISSN 0004-9182 print/ISSN 1465-3311 online/12/040425-20 #2012 Geographical Society of New South Wales Inc.
http://dx.doi.org/10.1080/00049182.2012.731302
recreation, and returned services venues that technically require a membership to
enter, although visitors may use the club after signing in at the door. Casinos, of
which there are currently 13 in Australia, are gambling-specific venues located in
major cities and regional centres. In 2009 these venues contained 197 820
operational EGMs, with 12 306 EGMs in casinos, 69 592 in hotels and 115 922
in clubs (Productivity Commission 2010). This network of venues produces
massive revenues for the state and industry. During the 200809 financial year total
EGM expenditure (i.e. amount actually lost) in clubs and hotels alone for Australia
was AUS$10.5 billion, equating to AUS$630 per adult (Queensland Government
2011). Over one quarter (AUS$2.9 billion or 28.1 per cent) of this expenditure was
collected as taxation revenue by individual State governments and accounted for
5.8 per cent of aggregated own-State tax revenue
1
(ABS 2010). Not surprisingly, a
parallel concern with the public health impacts of EGM gambling has emerged
(Marshall 2009). In particular, the issue of ‘problem gambling’has received
widespread national attention since highlighted by a Productivity Commission
report in 1999. More recently, the Commission (2010) has estimated the social cost
of problem gambling to be at least AUS$4.7 billion a year.
A concern with problem gambling, especially its identification, enumeration, and
associated risk factors, has captured the lion’s share of the research attention devoted
to EGM gambling in Australia (Young 2012). There has been relatively less
attention paid to the geography of EGM gambling, despite the inherently geographic
nature of the phenomenon, involving as it does a set of spatial relationships between
individual gambling venues and the characteristics of supporting populations. For
example, we know that EGM venues are disproportionately located in poorer areas
both in Australia (Marshall & Baker 2000, 2001a, b) and overseas (Gilliland & Ross
2005; Wheeler et al. 2006; Pearce et al. 2008; Robitaille & Herjean 2008). In
addition, spatial and temporal accessibility has been linked to both gambling
participation (Productivity Commission 1999; Marshall et al. 2004; Baker &
Marshall 2005; Marshall 2005; South Australian Centre for Economic Studies
2005; Young et al. 2012) and problem gambling (Storer et al. 2009).
Unfortunately, we know relatively little about the venues themselves. While a
‘community-based’category is commonly deployed in Australia to describe
gambling venues other than casinos (Productivity Commission 2010), in reality,
EGM venues display considerable variation in terms of machine numbers, other
gambling facilities, opening hours, location relative to existing infrastructure, the
proximity of competitors, ownership and control, licensing controls, gatekeeping
practices, recreational facilities, and marketing and promotion capabilities (Marshall
et al. 2004; McMillen & Doran 2006). As a consequence, different venues (i.e. pubs,
clubs and casinos) draw on different markets at different geographic scales (Fisher
2000; Young & Tyler 2008; Doran et al. 2007; Doran & Young 2010; Young et al.
2011). However, while a handful of studies have examined the markets of particular
groups of venues such as casinos (e.g. Fisher 2000) or clubs (e.g. Hing & Breen
2001), comparisons across venue types have rarely been drawn. There exists a
pressing need to relate demand for, and supply of, EGMs as mediated by the
network of gambling venues to more fully understand the incidence of gambling
behaviour and associated harm across a range of geographic contexts (Young & Tyler
2008). From a policy perspective, an understanding of the localised relationships
between venues and markets would allow the development of geographically tailored
harm-minimisation measures targeted at vulnerable subgroups. To this end, we seek
426 M. Young et al.
to develop a typology of gambling venues that is sensitive to geographic context, to
examine the differences in markets between these venue types, and to determine
which are the most risky venues defined by problem-gambling outcomes.
The geography of EGM gambling
EGM distribution and socio-economic disadvantage
The most distinctive feature of the geography of EGMs is their concentration in
poorer areas. Studies of the distribution of EGMs in Melbourne and Sydney have
shown socio-economically disadvantaged local government areas (LGAs) to be
more heavily supplied than their better-off counterparts (Marshall & Baker 2000,
2001a, b). This association is produced by a whole raft of urban geographic
processes affecting the operation of markets including government policy (restric-
tions on supply), local political action, ethnic and cultural variations in host areas,
and the historical patterns of development (Marshall & Baker 2001b). For
example, in the case of the recently developed Melbourne market, the spatial
distribution of EGMs was shown to move from a relatively random initial
distribution towards increasing concentration in disadvantaged areas over time
(Marshall & Baker 2002). The link between gambling supply and low-income areas
has been similarly documented in New Zealand (Wheeler et al. 2006) and Canada
(Gilliland & Ross 2005; Wilson et al. 2006; Robitaille & Herjean 2008).
Overprovision in poorer areas is socially problematic because the density of EGMs
has been linked to increased gambling participation and expenditure (Marshall 1999,
2005; Productivity Commission 1999; South Australian Centre for Economic
Studies 2005). People who live closer to EGM venues are more likely to participate
in gambling and spend more time and money within a venue (Marshall et al. 2004).
Moreover, the residential distance to gambling venues has been inversely correlated
with problem gambling in studies from Australia (Young et al. 2012), the USA (Welte
et al. 2004), Canada (Rush et al. 2007), and New Zealand (Pearce et al. 2008).
Gambling outcomes and temporal accessibility
In comparison to spatial accessibility, our knowledge of the effects of temporal
accessibility on gambling outcomes is relatively meagre. The key geographic work
was a study by Baker and Marshall (2005). These authors applied a time-economic
assignment model, originally developed by Baker (1994) in a retail trip context, to
gambling behaviour. Their goal was to simulate how individual gamblers apportion
time to gambling trips given certain parameters (i.e. opening hours, trip distance,
individual time for the week gambling, and socio-economic status). The model was
parameterised using Marshall’s (2005) study of EGM gambling in northern NSW,
drawing in particular on a demand-side typology of ‘average’and ‘involved’
gamblers. The time spent on gambling sessions and the frequency of gambling
acted as proxies for gambling behaviour, with involved gamblers more likely to
gamble in regular blocks than the more random interactions characteristic of
average gamblers. Importantly, the model predicted that an increase in venue
opening hours had the potential to increase the opportunity for involved gambling
and, by inference, the level of gambling-related harm. The implication of this study
is that a change in temporal accessibility can change the patterns of gambling harm
independently of spatial and social variables.
Placing Bets 427
Venue-specific influences on gambling behaviour
In the context of problem gambling and venue type, the limited evidence to date,
derived largely from prevalence surveys, is inconsistent. Venue patronage and
gambling harm vary across jurisdiction and by venue type. Drawing primarily on
the 2009 Victorian prevalence survey, the Commission (2010) argued that if harm
was conceived more broadly than just problem gambling (which did not significantly
vary across venue type in Victoria) then the highest risk venues were pubs and clubs
rather than casinos. In a similar vein, an exploratory study using a convenience
sample in New Zealand found that EGM use was a stronger predictor of problem
gambling in pubs and clubs compared to casinos, after controlling for demographic
factors and gambling activities (Clarke et al. 2010). However, due to the diversity
within the pubs and clubs category it is not entirely clear what these comparisons may
mean. To provide a more meaningful categorisation of EGM venues than the
‘community-venue’vs casino binary, and to link this categorisation to gambling
outcomes, we argue for the empirical development of an EGM venue typology.
The case for an empirical typology of EGM venues
A classic example of a venue typology is represented by the ethnographic work of
Cavan (1966) in the context of bars in the USA. Cavan identified four groups of
bars including convenience bars*accessible outlets in places of mass congregation
(e.g. downtown, sports stadiums, etc.); nightspot bars*entertainment-focused
venues; marketplace bars*pickup bars and other transactional bars (e.g. for drugs,
sex, gambling, and stolen goods); and home territory bars*dominated by one group
of regulars with a common characteristic (e.g. residential location, ethnicity, sexual
orientation, etc.). Research in this vein is valuable because not only does it make
clear that groups of similar venues are associated with distinctive social outcomes
but it also implies that regulation could usefully be adapted to these venue types.
This supply-side approach has yet to be fully developed in the context of gambling
venues, although there are several examples of efforts in this direction. In the US
casino context, Eadington (1998) argues for three distinct types: destination resort
casinos, urban casinos, and spatially dispersed, convenience-gambling venues. At a
smaller spatial scale, Posey (1998) conducted a functional classification of video
lottery terminal (VLT) venues in South Carolina. Ancillary establishments were
primarily non-recreational (e.g. gasoline stations and convenience stores),
comparative establishments were oriented towards the provision of other recrea-
tional services (e.g. movie theatres and bowling alleys), while dedicated establish-
ments, whose principal function was the provision of gaming machines, were also
identified. More recently, Young and Tyler (2008) have proposed a conceptual
typology of venues that position gambling venue types within two-dimensional
space of (a) accessibility, defined as the ease, in the sense of both time and distance,
with which venues may be visited, and (b) level of involvement in, or engagement
with, the host community. This typology produced four categories of venues
including (a) casino resorts, (b) urban casinos, (c) sporting clubs and neighbour-
hood pubs, and (d) shopping mall parlours and gaming arcades. Finally, an
empirically based supply-side typology was developed by Young et al. (2009) in the
context of the Northern Territory. Three general spatial patterns of EGM
expenditure were identified, namely suburban gambling complexes,city-centre
428 M. Young et al.
gambling agglomerations, and opportunistic gambling nodes. While these typologies
provide a useful starting point for teasing out some of the diversity among venues,
the link to gambling harm remains elusive.
The implicit argument of these supply-side typologies is that we need a more
nuanced understanding of the range of gambling venues if we are to develop
strategies for effective regulation. In Australia, the regulation of EGM venues is the
remit of the State government, and this has led to inconsistent regimes across the
country and over time (Productivity Commission 1999, 2010). While there is
regulatory variation across the States, within States there is little heterogeneity in
licensing practices between venues. All jurisdictions differentiate between hotels,
clubs, and casinos. However, no distinction is drawn between venues within these
three broad classes. Given the diversity of venues and the harm that they produce, a
one-size-fits-all policy response is too inflexible to minimise the adverse effects of
EGMs. We need better ways to target gambling problems among certain at-risk
groups in the context of particular venue configurations. Therefore, in this paper we
specifically seek to measure the relationship between venue type, associated
markets, and gambling outcomes. Specifically, we ask:
(1) How may EGM gambling venues be categorised?
(2) In what ways do EGM markets differ by venue type?
(3) How does gambling behaviour vary by venue type?
(4) Are some venues riskier than others, even when differences between their
markets are adjusted for?
Method
Study area
The Northern Territory (NT) of Australia is sparsely settled, with 63 per cent of its
2010 estimated residential population of 229 711 concentrated in the three largest
settlements of Darwin-Palmerston (107 430 persons), Alice Springs (27 987
persons) and Katherine (10 104 persons) (ABS 2011). The NT is notable for its
large Indigenous population: 28 per cent of residents identified as Indigenous in the
2006 census compared with 2 per cent for the rest of the country. However, this
proportion is reduced to 13 per cent for the three largest NT towns (ABS 2007).
The geography of poverty, as measured by the ABS Index of Relative Socio-
economic Disadvantage (IRSD: Australian Bureau of Statistics 2008) follows a
similar distribution. Thirty per cent of areas in the NT are classified in the most
disadvantaged decile of areas nationally. Most of these poorer areas (60 per cent)
are located outside the three largest towns.
Unlike much of the rest of Australia, where EGMs are located in economically
disadvantaged areas (Productivity Commission 1999; Marshall & Baker 2001b),
gambling opportunities in the NT are concentrated in the relatively large, more
affluent population centres. In June 2010, 88 per cent of EGMs (n1798) in the
NT were located in or around Darwin, Katherine or Alice Springs (see Figure 1),
dispersed across 64 licensed gambling venues. These venues consisted of casinos in
both Darwin and Alice Springs (833 EGMs), 26 clubs (612 EGMs) and 36 hotels
(353 EGMs). Clubs, such as sporting or returned servicepersons clubs, are
Placing Bets 429
not-for-profit entities restricted by a cap of 45 EGMs. Hotels or pubs are private
businesses capped at 10 EGMs each.
Data
EGM supply configuration. We selected five EGM venue-level variables that have
previously been implicated in gambling outcomes. These included licence type
(Productivity Commission 2010), number of EGMs (Storer et al. 2009), proximity
F
IGURE
1. Main urban centres in the Northern Territory.
430 M. Young et al.
to a centre of community congregation such as a supermarket (Marshall 2005;
Doran et al. 2007), distance from the central business district (Young et al. 2009),
and venue density (Young et al. 2009). For each of the EGM venues in Darwin,
Alice Springs and Katherine we recorded:
(1) Venue type (i.e. casino, club or hotel).
(2) Number of EGMs: based on the NT Department of Justice licensing database.
(3) Proximity to a supermarket: indicated by the presence of a supermarket within
750 m of the gambling venue. Geocoded supermarket locations were recorded
from the websites of duopoly operators Coles and Woolworths. We calculated
distance from each EGM venue to the nearest supermarket on the road
network.
(4) Distance to CBD: operationalised by distance to the nearest General Post
Office (GPO) as a proxy measure. We obtained GPO locations from the
Australia Post website and calculated the log transformed distance to the
nearest GPO for each venue on the road network
(5) Venue density: measured using a kernel density estimator, adopting smoothing
parameters from previous research in the same geographic area (bandwidth
1000 m, cell size 50 m: Young et al. 2009).
EGM markets and gambling outcomes. We investigated EGM-venue markets and
gambling outcomes using a mail survey. Between April and August 2010 we mailed
questionnaires to all 46 263 addresses in Darwin, Katherine and Alice Springs to
which Australia Post would deliver unsolicited mail and which were zoned as
residential. We extracted addresses from the Geocoded National Address File
(G-NAF), an authoritative, geocoded address database produced by PSMA
(PSMA Australia 2010). We selected an additional 2300 addresses for hand
delivery in Alice Springs and Darwin’s peri-urban fringe where Australia Post does
not deliver, using a spatially stratified cluster sample design.
The survey instrument elicited information about: age, sex, household structure,
education, income, the most frequently visited EGM venue in the last month,
EGM gambling participation and duration during last visit, and the Problem
Gambling Severity Index (PGSI) for the past 12 months. Residential distance to
gambling venue was calculated as the distance along the road network between the
respondent’s geocoded address and the location of their most frequently visited
gambling venue. We used the PGSI as our measure of gambling-related harm as it is
a clinically validated nine-item scale routinely used in Australia and overseas to
estimate problem-gambling risk in the general population (Ferris & Wynne 2001;
Neal et al. 2005). Responses to the PGSI range on an ordinal scale from 0 to 27 that
categorised respondents as being at no risk (PGSI 0), low risk (PGSI 12),
moderate risk (PGSI 37) or high risk (PGSI 8) of being a problem gambler
(Ferris & Wynne 2001).
Analysis
Typology derivation. We performed cluster analysis on the EGM supply configura-
tion variables to create a typology of EGM venues based on their spatial and
regulatory characteristics. We used agglomerative hierarchical clustering and
standardised interval measures, adopting Gower’s method for creating a dissimilarity
Placing Bets 431
matrix for mixed data and Ward’s method for cluster formation. The number of
clusters was determined by selecting the clustering output with the greatest average
silhouette width (Rousseeuw 1987).
Market characterisation. We investigated differences in the socio-demographic
markets and gambling outcomes across venue types through the creation of
contingency tables. We allocated each survey respondent to a single venue cluster
based upon the venue they nominated as their most frequently visited in the last
month. In order to account for non-response bias, we conducted post-stratification
of survey responses, stratifying by age bracket (1530, 3045, 4560 and 65),
gender and survey region (Darwin urban, Darwin peri-urban, Katherine, Alice
Springs urban, and Alice Springs peri-urban).
Multivariate analysis of problem-gambling risk. We conducted a multivariate analysis
to investigate the association between venues type and gambling-related harm while
controlling for the socio-demographic characteristics of their markets*risk factors
in their own right. We employed negative binomial regression to investigate the
independent predictors of the PGSI. We investigated interactions between venue
type and the socio-demographic variables using the likelihood ratio test but none
were found to be significant. Because age and gender are adjusted for in the
multivariate analysis, weighting was unnecessary. We excluded cases with missing
data list-wise.
Results
Venue typology
The cluster analysis produced a typology with six classifications (average silhouette
width 0.71; see Table 1). The first cluster contained only two venues, SKYCITY
Casino in Darwin and Lasseter’s Hotel Casino in Alice Springs. This casino cluster
was differentiated from other venues by licensing arrangements and consequently
hosts many more EGMs per venue. While these casinos also offer table games, TAB
(off-course totalisator betting) and Keno (state-wide instant electronic lottery),
EGMs dominate, with EGM gambling expenditure accounting for 79 per cent of
casino gaming expenditure in NT in 200607.
2
The second cluster contained seven
clubs, six of which had reached their cap of 45 EGMs. We labelled these venues
supermarket-attached clubs due to their location proximate to supermarkets. The
third cluster contained the remaining 19 clubs. While this category included some
service clubs located in isolated areas, the cluster was typified by sports clubs
located on or at facilities such as golf courses. Because these venues were largely
located away from shopping centres and CBDs we labelled them peripheral clubs.
While these venues typically had fewer machines than other clubs (median
16 EGMs per venue), this category masks some diversity in that several larger
suburban clubs located some distance away from supermarkets were included. The
fourth cluster contained 10 hotels in the CBD of Darwin. These agglomerated pubs
are centred on Mitchell Street and oriented towards the night-time economy. All
these venues had reached their cap of 10 EGMs and were geographically
concentrated (median 7.5 venues per km
2
). The fifth cluster of venues consisted
of nine pubs located proximate to centres of community congregation. Labelled
432 M. Young et al.
T
ABLE
1. Venue cluster profiles
Casino
Supermarket-attached
club Peripheral club
Agglomerated
pub
Supermarket-attached
pub Peripheral pub
(n2) (n7) (n19) (n10) (n9) (n17)
Licence type Casino Club Club Hotel Hotel Hotel
Number of EGMs
a
417 (289544) 45 (1345) 16 (345) 10 (1010) 10 (1010) 10 (410)
Venues per square km
a
1.1 (1.01.1) 2.3 (1.07.8) 1.1 (1.02.1) 7.5 (5.68.7) 2.5 (1.03.3) 1.0 (1.04.6)
Distance to supermarket (km)
a
2.3 (2.02.6) 0.3 (0.20.7) 1.5 (0.8433.9) 0.4 (0.10.6) 0.4 (0.00.7) 4.0 (0.8133.8)
Distance to GPO (km)
a
2.4 (2.22.7) 12.8 (0.320.7) 4.2 (1.1434.2) 0.6 (0.10.8) 13.7 (0.037.8) 11.4 (1.0134.0)
Notes:
All clusters contained only a single licence type.
a
Median value with range in parentheses.
Placing Bets 433
supermarket-attached pubs, all these venues had reached their EGM cap of
10 machines. Typically these venues are bars featuring EGMs, TAB and Keno
facilities, integrated into a suburban shopping complex. The final venue cluster
identified 17 peripheral pubs located in diverse locations away from urban shopping
complexes and the CBDs. These venues were generally situated to capture passing
trade, either on high traffic routes as with roadhouses or at destinations such as
airports and beaches.
Variations in markets
We received 7041 completed questionnaires (an overall response rate of 14.5 per
cent). Significant differences in the socio-demographic characteristics of visitors
were identified across the six venue types (see Table 2). Compared to the sample
frame, those whose preferred venue was a casino were more likely to be female, and
less likely to be educated at a technical college. The clientele of supermarket-attached
clubs was older than the sample frame and more working class, with a greater
proportion of visitors having a technical education and a smaller proportion having
a university education. Peripheral club users were also older than the sample frame,
but were also more likely to be male and less likely to live in shared accommodation.
In contrast, pubs attracted a younger market, one that is segregated by income.
Visitors to agglomerated pubs were the youngest of all venue types and of higher
income than the sample frame, less likely to be living with children, and more likely
to be living in shared accommodation. In contrast, while those who preferred to
visit a supermarket-attached pub were also younger than the sample frame, they
earned a lower income. While again younger than the sample frame, visitors to
peripheral pubs earned higher incomes, were more likely to be male and live in
shared accommodation.
Gambling outcomes
Gambling behaviour varied significantly by venue type (see Table 3). The
proportion of visitors who played EGMs was significantly higher for casinos
(40.4 per cent) and supermarket-attached clubs (26.5 per cent) compared to 14 per
cent for all venues combined. Conversely, a lower proportion of supermarket-
attached and agglomerated pub-goers gambled on EGMs compared to the whole
sample. Mean EGM gambling session duration was highest for the casinos (130.3
minutes), followed by clubs (over 1 hour for both types), and lower again for the
pubs. Visitors travelled further on average to visit the casinos (10.2 km) than other
venues. The distances travelled to venues that were near supermarkets were
significantly lower (4.5 km for supermarket-attached pubs and 4.7 km for
supermarket-attached clubs) compared to visits to all venues (7.3 km).
In terms of gambling-related harm, the cross-tabulation identified two venue
types with a higher proportion of problem-gambling patrons than the sample frame.
The casino was most associated with problem gambling, with 6.7 per cent of those
who prefer to visit casinos at high risk of problem gambling*more than double the
estimated rate in the sample frame (2.8 per cent). Supermarket-attached clubs were
also associated with high rates of gambling-related harm, with 10.1 per cent of
434 M. Young et al.
T
ABLE
2. Socio-demographics by venue type
Casino % (c.i.)
Supermarket-
attached club %
(c.i.)
Peripheral
club % (c.i.)
Agglomerated
pub% (c.i.)
Supermarket-
attached
pub % (c.i.)
Peripheral pub %
(c.i.)
Sample frame %
(c.i.)
n1069 n999 n1331 n449 n447 n683 n7041
N16 530 N14 133 N19 874 N11 107 N8799 N14 819 N112 541
Age bracket
1529 34.4 (30.138.9) 27.3 (22.732.3) 25.0 (21.3
29.0) 60.8 (55.8
65.5) 41.5 (35.0
48.1) 43.8 (39.0
48.8) 31.9 (31.931.9)
3044 31.2 (28.134.5) 27.4 (24.3
30.7) 30.8 (28.133.6) 26.2 (22.4
30.3) 35.0 (30.140.2) 31.0 (27.434.8) 30.9 (30.930.9)
4559 23.1 (20.725.7) 26.9 (24.129.8) 28.2 (25.8
30.6) 11.1 (9.1
13.6) 17.6 (14.5
21.1) 19.5 (17.0
22.4) 24.4 (24.424.4)
6011.4 (9.913.0) 18.5 (16.4
20.7) 16.1 (14.5
17.8) 1.9 (1.3
3.0) 5.9 (4.4
7.9) 5.7 (4.5
7.2) 12.8 (12.812.8)
Sex
Female 52.9 (49.0
56.8) 47.0 (43.051.0) 44.9 (41.8
48.2) 52.7 (46.858.6) 51.0 (44.857.1) 41.9 (37.5
46.4) 48.5 (48.548.5)
Male 47.1 (43.2
51.1) 53.0 (49.057.0) 55.1 (51.9
58.2) 47.3 (41.453.2) 49.0 (42.955.2) 58.1 (53.6
62.5) 51.5 (51.551.5)
Education
Primary 1.6 (0.55.2) 1.5 (0.92.7) 1.3 (0.82.1) 0.7 (0.22.1) 1.0 (0.52.2) 0.8 (0.4 1.9) 1.3 (1.01.7)
Secondary 34.2 (30.438.3) 36.7 (32.640.9) 32.8 (29.536.3) 27.3 (22.233.1) 37.6 (31.743.8) 34.0 (29.339.0) 31.7 (30.233.2)
Technical 15.1 (12.5
18.2) 24.7 (21.3
28.4) 21.5 (18.624.6) 18.7 (14.423.9) 21.0 (15.927.1) 19.7 (16.024.0) 19.9 (18.621.2)
University 49.0 (44.9 53.2) 37.1 (33.1
41.3) 44.4 (41.047.9) 53.3 (47.359.3) 40.4 (34.446.7) 45.5 (40.550.6) 47.1 (45.648.7)
Income bracket
B$149 7.8 (5.610.9) 7.0 (5.29.3) 6.9 (5.48.7) 3.3 (1.9
5.4) 10.8 (6.717.0) 3.6 (2.4
5.4) 6.8 (6.07.7)
$150$599 17.1 (14.420.2) 19.8 (16.923.0) 15.4 (13.217.9) 13.2 (9.817.7) 20.9 (16.226.6) 14.0 (10.917.8) 17.8 (16.719.0)
$600$1599 59.8 (55.763.7) 58.0 (53.862.1) 57.7 (54.3 61.2) 66.3 (60.6
71.6) 57.2 (50.863.4) 66.3 (61.5
70.8) 58.9 (57.460.4)
$1600 15.3 (12.818.3) 15.3 (12.318.8) 20.0 (17.3 22.9) 17.2 (13.222.1) 11.0 (8.0
15.0) 16.1 (12.820.0) 16.4 (15.317.6)
Household structure
Couple
without
children
35.5 (31.639.6) 35.2 (31.139.4) 34.9 (31.638.4) 39.2 (33.545.4) 27.9 (22.833.7) 32.9 (28.337.9) 33.4 (31.934.9)
Couple with
children
33.4 (29.637.4) 29.7 (26.233.4) 35.7 (32.539.1) 11.7 (9.1
15.0) 37.2 (31.543.3) 25.9 (22.130.2) 31.3 (30.032.7)
Placing Bets 435
T
ABLE
2(Continued )
Casino % (c.i.)
Supermarket-
attached club %
(c.i.)
Peripheral
club % (c.i.)
Agglomerated
pub% (c.i.)
Supermarket-
attached
pub % (c.i.)
Peripheral pub %
(c.i.)
Sample frame %
(c.i.)
n1069 n999 n1331 n449 n447 n683 n7041
N16 530 N14 133 N19 874 N11 107 N8799 N14 819 N112 541
Group 14.2 (11.118.0) 11.7 (8.715.6) 10.5 (8.1
13.5) 27.5 (22.2
33.6) 16.5 (11.722.9) 21.7 (17.2
26.9) 14.8 (13.516.2)
Single parent 4.0 (2.95.6) 5.2 (4.06.7) 3.7 (2.84.8) 3.0 (1.94.5) 4.8 (3.27.1) 4.5 (3.26.4) 4.6 (4.15.1)
Single
person
12.9 (10.815.3) 18.3 (15.521.5) 15.2 (13.017.8) 18.5 (14.523.4) 13.6 (9.419.2) 14.9 (12.018.3) 15.9 (14.917.0)
Notes:
nunweighted sample size, Nweighted sample size, c.i.95% confidence interval. Category estimates where the c.i. does not overlap the sample
frame’s c.i. are indicated in bold.
436 M. Young et al.
T
ABLE
3. Gambling outcomes by venue type
Casino % (c.i.)
Supermarket-
attached club %
(c.i.)
Peripheral club
% (c.i.)
Agglomerated
pub% (c.i.)
Supermarket-
attached pub %
(c.i.)
Peripheral pub %
(c.i.)
Sample frame %
(c.i.)
n1069 n999 n1331 n449 n447 n683 n7041
N16 530 N14 133 N19 874 N11 107 N8799 N14 819 N112 541
Played
EGMs
40.2 (36.1
44.4) 22.4 (18.7
26.5) 14.7 (12.117.6) 6.2 (3.3
10.5) 7.2 (4.6
10.8) 10.8 (7.614.7) 14.1 (12.915.3)
EGM
session
(mean
minutes)
130.3 (111.3
149.4) 63.9 (53.5
74.3) 62.4 (53.0
71.8) 37.7 (21.7
53.6) 58.7 (34.582.9) 54.3 (37.6
71.0) 88.2 (78.697.9)
Distance
travelled
to venue
(mean km)
10.2 (9.4
11.0) 4.7 (4.2
5.1) 6.8 (5.68.0) 8.6 (7.7 9.6) 4.5 (3.9
5.1) 8.0 (6.99.2) 7.3 (6.97.7)
PGSI
Non-
problem
71.5 (67.4
75.3) 76.4 (72.1
80.2) 84.5 (81.487.2) 83.2 (77.987.4) 79.3 (72.784.6) 82.8 (78.786.3) 83.1 (81.784.4)
Low risk 13.0 (10.3
16.1) 9.9 (7.612.7) 8.7 (6.711.3) 12.7 (9.017.8) 11.0 (7.4 16.1) 8.6 (6.311.7) 8.9 (7.99.9)
Moderate
risk
8.8 (6.8
11.4) 10.1 (7.1
14.0) 4.5 (3.36.2) 2.7 (1.2 5.7) 5.9 (3.410.1) 6.1 (4.09.4) 5.3 (4.56.1)
High risk 6.7 (4.4
10.1) 3.6 (2.25.9) 2.2 (1.14.4) 1.4 (0.73.1) 3.8 (1.410.1) 2.4 (1.24.6) 2.8 (2.1 3.6)
Notes:
nunweighted sample size, Nweighted sample size, c.i.confidence interval. Category estimates where the c.i. does not overlap the sample frame’s
c.i. are indicated in bold. Estimates are percentages, unless indicated otherwise in row labels.
Placing Bets 437
those who preferred to visit a supermarket-attached club at moderate risk of
problem gambling compared to 5.3 per cent in the sample frame.
The multivariate analysis (see Table 4) revealed that, after adjusting for the
effects of socio-demographic characteristics across venue types, casinos were most
strongly associated with gambling harm. The average estimated PGSI of casino-
goers was 3.6 times that of agglomerated pub-goers (the base category). Super-
market-attached club-goers were also at increased risk, with an average estimated
PGSI 2.2 times that of agglomerated pub-goers. There was also a significant
association between PGSI and peripheral clubs, with mean PGSI 1.5 times that of
patrons of agglomerated pubs. Other significant risk factors identified were
consistent with previous studies in that males, younger people, less well-educated
people, lower-income people and those not living with a partner had a higher PGSI
score.
T
ABLE
4. Risk factors for gambling-related harm among those who visited a venue
PGSI ratio (c.i.)
Constant 0.0 (0.00.1)
***
Venue type
Agglomerated pub 1.0 (ref. group)
Casino 3.6 (2.4, 5.3)
***
Supermarket-attached club 2.2 (1.43.3)
***
Peripheral club 1.5 (1.02.2)
*
Supermarket-attached pub 1.2 (0.72.0)
Peripheral pub 1.3 (0.82.0)
Age
651.0 (ref. group)
4564 2.0 (1.42.9)
***
3044 2.5 (1.73.6)
***
B30 2.5 (1.63.9)
***
Gender
Female 1.0 (ref. group)
Male 2.2 (1.82.8)
***
Education
University 1.0 (ref. group)
Technical 1.2 (0.91.6)
Secondary 1.7 (1.42.2)
***
Primary 2.2 (1.15.2)
Income
$16001.0 (ref. group)
$600$1599 2.0 (1.52.7)
***
$150$599 2.1 (1.53.0)
***
B$149 1.6 (1.02.5)
Household structure
Couple without children 1.0 (ref. group)
Couple with children 1.0 (0.81.4)
Group 2.6 (1.93.7)
***
Single parent 1.9 (1.33.0)
**
Single person 1.5 (1.12.0)
**
Notes:
Nagelkerke’sR
2
0.13, n4789. c.i.95% confidence interval.
*
pB0.05,
**
pB0.01,
***
pB0.001. PGSI ratio is obtained by exponentiating the coefficient estimates from a
negative binomial multiple regression.
438 M. Young et al.
Discussion
Constructed using licence type, number of EGMs, venue density, and distance
from a supermarket and CBD, our typology produced six clusters: one for casinos,
two for clubs, and three for pubs. In the context of clubs, the typology drew a
distinction between those near centres of community congregation (such as
supermarkets and CBD) and those located in the urban periphery. The super-
market-attached clubs contained, on average, twice the number of EGMs, and were
located in areas of double the venue density, compared to the peripheral clubs. The
typology produced three categories of pubs. Two mirrored the peripheral vs
supermarket-attached distinction drawn between the clubs, while a third, an
agglomerated pub category, was also created. This category comprised the cluster
of venues in the Darwin CBD.
Our analysis revealed that these venues types were associated with different levels
of gambling harm. Larger venues (i.e. regional casinos and clubs adjacent to
shopping centres) were ‘riskier’or, put another way, their clientele comprised a
higher proportion of problem gamblers compared to other venues. Indeed, 15.5 per
cent of casino patrons and 13.7 per cent of supermarket-attached club-goers were
moderate- or high-risk gamblers (PGSI 3) compared to 8.1 per cent across all
venues. Venue proximity to areas of community congregation also proved to be
significantly associated with gambling harm. The clientele of supermarket-attached
clubs (14.7 per cent PGSI 3) and pubs (9.7 per cent PGSI 3) comprised more
problem gamblers than their peripheral counterparts (clubs 7.7 per cent PGSI 3
and pubs 8.5 per cent PGSI 3 ).
There appear to be two forces at play here: accessibility and venue size. In terms
of accessibility, the location of EGM venues in or near local shopping centres means
that more people are likely to interact with them than other venue types (Doran &
Young 2010). In a time-geographic perspective, EGM venues are accessible when
they fall within a gambler’s potential path area, the region reachable by a gambler
over the course of the day when constrained by their existing travel behaviour
(Golledge & Stimson 1997). Thus, it is likely that gambling venues proximate to
locations visited by many gamblers such as supermarkets and other centres of
community congregation have a greater spatio-temporal accessibility than would
otherwise be the case (e.g. Doran et al. 2007; Marshall 2005). Our finding that the
mean distance travelled to supermarket-attached venues is significantly lower than
for other venue types supports this argument. The accessibility effect is similarly
reflected in the higher levels of problem gambling within supermarket-attached
pubs compared with peripheral ones. Geographically accessibility may be plausibly
linked to gambling harm through a simple exposure model (cf. Young & Tyler
2008).
As a counterpoint, while the highly risky casinos in both Darwin and Alice
Springs are relatively accessible to the majority of town residents, they are not as
geographically accessible as the supermarket-attached clubs. What appears to be
more influential here is venue size. In the case of the casinos, we argue that the
reason for the high number of problem gamblers relates to the specific
characteristics of the venue itself. Casinos are gambling-specific venues and attract
proportionately more EGM gamblers (40.2 per cent of visitors) than pubs (6.2
10.8 per cent) and clubs (14.722.4 per cent) simply by virtue of the gambling
opportunities available. Our data show that casino visitors travel further than
Placing Bets 439
visitors to other venues, with a mean casino trip length of 10.2 km compared to 7.3
km for all venues. The gambling attractiveness of the casinos is further indicated by
the fact that the mean EGM session time is 130 minutes, twice that of any of the
other venues. This more ‘involved’gambling behaviour (Baker & Marshall 2005)
clearly translates into higher levels of problem gambling, with casino-goers 3.6
times as likely to be problem gamblers than visitors to the base venue category (i.e.
agglomerated pubs: see Table 4). The size effect is also evident for the supermarket-
attached clubs (six of the seven had reached their maximum allocation of 45
EGMs). Almost a quarter of visitors to supermarket-attached clubs (22.4 per cent)
played EGMs on their last visit (compared to 14 per cent for the sample), and they
were also over twice as likely be problem gamblers compared to the base category
(see Table 4). These are clearly venue effects because the market characteristics (i.e.
age, sex, education, income, education, household structure) were adjusted for in
the analysis.
This raises the question of what makes larger venues so risky. Not only are large
venues able to provide more EGMs, they can also provide a greater range of
machines and more features such as linked jackpots. Indeed, the marginal profits
gained by increasing the number of EGMs allow activities such as marketing and
promotion campaigns, provision of courtesy buses and more ancillary facilities to
be funded. This is consistent with the limited research into gambling venue
preferences that has identified service, security, range of EGMs, membership costs,
social mix, linked jackpots, and specific ‘bonus features’as attractors (Hing & Haw
2009, 2010). This means that larger venues may be more attractive to dedicated
gamblers. The result is an economy of scale, or ‘unvirtuous cycle’, whereby more
EGMs equal more gamblers which equal larger linked jackpots and more revenue
for marketing and promotion, and so on.
One implication of these findings is that policy makers can affect gambling
outcomes by restricting not only spatial and temporal access but also venue size
expressed as number of EGMs. For example, we would expect greater concentra-
tion of problem gamblers in the larger supermarket-attached clubs in the NT if the
EGM cap were raised. Such an increase has already occurred in other jurisdictions
where clubs may host several hundred EGMs. One of the policy options recently
suggested is the notion of ‘destination-style gambling’, where gambling products
are centralised in fewer larger venues (Victorian Department of Justice 2008). The
general idea is that this approach would reduce accessibility and allow improved
monitoring of problem gambling within venues. Our current results suggest that
problem-gambling rates would be much higher within these larger venues and that
any reduction in problem gambling at the population level would depend on a
significant reduction of supply outside these venues.
Conclusion
We have demonstrated that gambling markets and outcomes vary between and
within the orthodox venue categories of hotels, clubs and casinos. We have argued
that gambling risk is a function of the interaction of geographic accessibility to
markets on the one hand and venue effects on the other. While our analysis suggests
that venue size is a crude indicator of riskiness, more research needs to be
conducted on what makes a venue risky. In addition, our typology has been devised
using a fairly small range of venues when compared to the national scale. We
440 M. Young et al.
emphasise the need for supply typologies in other jurisdictions where the supply
configuration of EGM venues may vary greatly. For example, many clubs in NSW
and Victoria have far more machines than do casinos in the NT. Indeed, the
distinctions between venue categories are blurring as they merge technologies to
produce spaces of gambling consumption (Austrin & Curtis 2004). If this industry
is to be regulated effectively, governments need a convincing supply-side
conceptual apparatus that is empirically verified. Here geographers have much to
add. For example, we need to know more about the relationship between gambling
venues and the action and activity spaces of particular individuals. The role of
distance is clearly important here, both physical (Young et al. 2012) and cognitive
(e.g. Hawthorne & Kwan 2012). In addition, while we know the time available to
an individual affects gambling outcomes (Baker & Marshall 2005), we know little
about the temporal sequencing of gambling behaviour and the way it fits into daily
spatial patterns. There is a strong case for time-geography studies in gambling that
examine the space-time constraints operating on different gamblers relative to
different venues. Finally, there is also a need for a broader political-economic
project that links gambling as a form of consumption to broader trends in the
evolution in capitalism and associated neoliberal governance at a range of spatial
scales (Young 2011).
Acknowledgements
The research was supported by Australian Research Council Linkages Grant
LP0990584, the Community Benefit Fund of the Northern Territory Government
and the Northern Territory Research and Innovation Fund. The authors thank two
anonymous reviewers for their helpful comments.
Correspondence: Martin Young, School of Tourism and Hospital Management,
Hogbin Drive, Coffs Harbour, NSW 2450, Australia.
E-mail: martin.young@scu.edu.au
NOTES
[1] Authors’calculation based on Queensland Government (2011) and ABS (2010) data.
New South Wales, Victoria, Queensland and Tasmania report EGM and Keno tax
revenue in aggregate form only. Here we adjusted EGM taxation revenue based on the
Productivity Commission’s (2010) assumption that Keno revenue totals 5 per cent of
combined revenues.
[2] This estimate excludes TAB wagering. Authors’calculations, based on figures in
Queensland Government (2011) and NT Department of Justice data, adjusted for
inflation into AUS$ 2007.
REFERENCES
A
USTRALIAN
B
UREAU OF
S
TATISTICS
(ABS) (2007) Basic community profile, Cat. No.
2001.0, Commonwealth of Australia, Canberra.
A
USTRALIAN
B
UREAU OF
S
TATISTICS
(ABS) (2008) An introduction to Socio-economic Indexes
for Areas (SEIFA), 2006, Cat. No. 2039.0, Commonwealth of Australia, Canberra.
A
USTRALIAN
B
UREAU OF
S
TATISTICS
(ABS) (2010) Taxation revenue, Australia, 200809,
Cat. No. 5506.0, Commonwealth of Australia, Canberra.
Placing Bets 441
A
USTRALIAN
B
UREAU OF
S
TATISTICS
(ABS) (2011) Regional population growth, Australia,
200910, Cat. No. 3218.0, Commonwealth of Australia, Canberra.
A
USTRIN
,T.&C
URTIS
, B. (2004) ‘The desegregation of gambling media and the emergence
of a single form of gambling’,Social Policy Journal of New Zealand 21, pp. 418.
B
AKER
, R.G.V. (1994) ‘An assessment of the space-time differential model for aggregate
trip behaviour to planned suburban shopping centres’,Geographical Analysis 26(4),
pp. 34163.
B
AKER
, R.G.V. & M
ARSHALL
, D.C. (2005) ‘Modelling gambling time and economic
assignments to weekly trip behaviour to gambling venues’,Journal of Geographical Systems
7(34), pp. 381402.
C
AVAN
, S. (1966) Liquor license: an ethnography of bar behaviour, Aldine, Chicago.
C
LARKE
, D., P
ULFORD
, J., B
ELLRINGER
, M., A
BBOTT
,M.&H
ODGINS
, D. (2010) ‘An
exploratory study of problem gambling on casino versus non-casino electronic gaming
machines’,International Journal of Mental Health and Addiction 10(13), pp. 10721.
D
ORAN
, B., M
C
M
ILLEN
,J.&M
ARSHALL
, D. (2007) ‘A GIS-based investigation of gaming
venue catchments’,Transactions in GIS 11(43), pp. 24565.
D
ORAN
,B.&Y
OUNG
, M. (2010) ‘Predicting the spatial distribution of gambling
vulnerability: an application of gravity modeling using ABS mesh blocks’,Applied
Geography 30(1), pp. 14152.
E
ADINGTON
, W.R. (1998) ‘Contributions of casino-style gambling to local economies’,
Annals of the American Academy of Political and Social Science 556(1), pp. 5365.
F
ERRIS
,J.&W
YNNE
, H. (2001) The Canadian Problem Gambling Index: final report, Canadian
Centre on Substance Abuse, Ottawa.
F
ISHER
, S. (2000) ‘Measuring the prevalence of sector-specific problem gambling: a study of
casino patrons’,Journal of Gambling Studies 16(13), pp. 2551.
G
ILLILAND
, J.A. & R
OSS
, N.A. (2005) ‘Opportunities for video lottery terminal gambling in
Montreal’,Canadian Journal of Public Health 96(13), pp. 559.
G
OLLEDGE
,R.&S
TIMSON
, R. (1997) Spatial behavior: a geographic perspective, Guilford
Press, New York.
H
AWTHORNE
,T.&K
WAN
, M.-P. (2012) ‘Using GIS and perceived distance to understand
the unequal geographies of healthcare in lower-income urban neighbourhoods’,
Geographical Journal 178(1), pp. 1830.
H
ING
,N.&B
REEN
, H. (2001) ‘Profiling lady luck: an empirical study of gambling and
problem gambling amongst female club members’,Journal of Gambling Studies 17(13), pp.
4769.
H
ING
,N.&H
AW
, J. (2009) ‘The development of a multi-dimensional gambling accessibility
scale’,Journal of Gambling Studies 25(4), pp. 56981.
H
ING
,N.&H
AW
, J. (2010) The influence of venue characteristics on a player’s decision to attend a
gambling venue, Centre for Gambling Education and Research, Southern Cross University,
Lismore, NSW.
M
ARSHALL
, D.C. (1999) ‘Adelaide’s pokie geography: distribution of, and expenditure
on, electronic gaming machines in Adelaide’,South Australian Geographical Journal
98, pp. 1929.
M
ARSHALL
, D.C. (2005) ‘The gambling environment and gambler behaviour: evidence from
Richmond-Tweed, Australia’,International Gambling Studies 5(13), pp. 6383.
M
ARSHALL
, D.C. (2009) ‘Gambling as a public health issue: the critical role of the local
environment’,Journal of Gambling Issues 23, pp. 6680.
M
ARSHALL
, D.C. & B
AKER
, R.G.V. (2000) ‘No chance: electronic gaming machines and the
disadvantaged in Melbourne’,Geodate 13(43), pp. 58.
M
ARSHALL
, D.C. & B
AKER
, R.G.V. (2001a) ‘Clubs, spades, diamonds and disadvantage: the
geography of electronic gaming machines in Melbourne’,Australian Geographical Studies
39(13), pp. 1733.
M
ARSHALL
, D.C. & B
AKER
, R.G.V. (2001b) ‘Unfair odds? Factors influencing the
distribution of electronic gaming machines in Melbourne’,Urban Policy and Research
19(13), pp. 7792.
M
ARSHALL
, D.C. & B
AKER
, R.G.V. (2002) ‘The evolving market structures of gambling:
case studies modelling the socioeconomic assignment of gaming machines in Melbourne
and Sydney, Australia’,Journal of Gambling Studies 18(33), pp. 27390.
442 M. Young et al.
M
ARSHALL
, D.C., M
C
M
ILLEN
, J., N
IEMEYER
,S.&D
ORAN
, B. (2004) Gaming machine
accessibility and use in suburban Canberra: a detailed analysis of the Tuggeranong Valley, Centre
for Gambling Research, Australian National University, Canberra.
M
C
M
ILLEN
,J.&D
ORAN
, B. (2006) ‘Problem gambling and gaming machine density:
socio-spatial analysis of three Victorian localities’,International Gambling Studies 6(13),
pp. 529.
N
EAL
, P., D
ELFABBRO
, P.H. & O’N
EIL
, M. (2005) Problem gambling and harm: towards a
national definition, report prepared for the National Gambling Research Program Working
Party, Melbourne.
P
EARCE
, J., M
ASON
, K., H
ISCOCK
,R.&D
AY
, P. (2008) ‘A national study of neighbourhood
access to gambling opportunities and individual gambling behaviour’,Journal of
Epidemiology and Community Health 62, pp. 8628.
P
OSEY
, A. (1998) ‘In the absence of casinos: a look at video poker’, in Meyer-Arendt, K. &
Hartmann, R. (eds) Casino gambling in America: origins, trends and impacts, Cognizant
Communication, New York, pp. 22738.
P
RODUCTIVITY
C
OMMISSION
(1999) Australia’s gambling industries: inquiry report No. 10,
Productivity Commission, Melbourne.
P
RODUCTIVITY
C
OMMISSION
(2010) Gambling, report No. 50, Productivity Commission,
Canberra.
PSMA A
USTRALIA
(2010) Product description: G-NAF version 1.11, PSMA Australia,
Canberra.
Q
UEENSLAND
G
OVERNMENT
(2011) Australian gambling statistics 198384 to 200809 (27th
edn), Office of Economic and Statistical Research, Brisbane.
R
OBITAILLE
,E.&H
ERJEAN
, P. (2008) ‘An analysis of the accessibility of video
lottery terminals: the case of Montreal’,Inter national Journal of Health Geographics
7(23), pp. 115.
R
OUSSEEUW
, P. (1987) ‘Silhouettes: a graphical aid to the interpretation and validation of
cluster analysis’,Journal of Computational and Applied Mathematics 20, pp. 5365.
R
USH
, B., V
ELDHUIZEN
,S.&A
DLAF
, E. (2007) ‘Mapping the prevalence of problem
gambling and its association with treatment accessibility and proximity to gambling
venues’,Journal of Gambling Issues 20, pp. 193214.
S
OUTH
A
USTRALIAN
C
ENTRE FOR
E
CONOMIC
S
TUDIES
(2005) Community impacts of
electronic gaming machine gambling (Part A), Department of Justice, State of Victoria,
Melbourne.
S
TORER
, J., A
BBOTT
,M.&S
TUBBS
, J. (2009) ‘Access or adaptation? A meta-analysis of
surveys of problem gambling prevalence in Australia and New Zealand with respect to
concentration of electronic gaming machines’,International Gambling Studies 9(33), pp.
22544.
V
ICTORIAN
D
EPARTMENT OF
J
USTICE
(2008) Destination gaming: evaluating the benefits for
Victoria, Department of Justice, Melbourne.
W
ELTE
, J.W., W
IECZOREK
, W.F., B
ARNES
, G., T
IDWELL
, M.C. & H
OFFMAN
, J. (2004) ‘The
relationship of ecological and geographic factors to gambling behavior and pathology’,
Journal of Gambling Studies 20(43), pp. 40523.
W
HEELER
, B.W., R
IGBY
, J.E. & H
URIWAI
, T. (2006) ‘Pokies and poverty: problem gambling
risk factor geography in New Zealand’,Health and Place 12(13), pp. 8696.
W
ILSON
, D., G
ILLILAND
, J., R
OSS
, N., D
EREVENSKY
,J.&G
UPTA
, R. (2006) ‘Video lottery
terminal access and gambling among high school students in Montreal’,Canadian Journal
of Public Health 97(33), pp. 2026.
Y
OUNG
, M. (2011) ‘Towards a critical geography of gambling spaces: the Australian
experience’,Human Geography 4, pp. 3347.
Y
OUNG
, M. (2012) ‘Statistics, scapegoats and social control: a critique of pathological
gambling prevalence research, Addiction Research and Theory, available online: http://dx.
doi.org/10.3109/16066359.2012.680079 (Accessed 1 October 2012).
Y
OUNG
, M., L
AMB
,D.&D
ORAN
, B. (2009) ‘Mountains and molehills: a spatiotemporal
analysis of poker machine expenditure patterns in the Northern Territory’,Australian
Geographer 40(33), pp. 24969.
Placing Bets 443
Y
OUNG
, M., L
AMB
,D.&D
ORAN
, B. (2011) ‘Gambling, resources, and racial economy: an
analysis of EGM expenditure patterns in three remote Australian towns’,Geographical
Research 49(13), pp. 5971.
Y
OUNG
, M., M
ARKHAM
,F.&D
ORAN
, B. (2012) ‘Too close to home? The relationships
between residential distance to venue and gambling outcomes’,International Gambling
Studies 12(2), pp. 257273.
Y
OUNG
,M.&T
YLER
, W. (2008) ‘Mediating markets: gambling venues, communities and
social harm’,Gambling Research 20(13), pp. 5065.
444 M. Young et al.