Casino proximity, visit frequency, and gambling problems

Casino proximity, visit frequency, and gambling problems

Kahlil S. Philander1,2 (corresponding author)

Nassim Tabri3

Richard T. Wood4

Michael Wohl3

1Washington State University, Carson College of Business

2University of Sydney, Department of Psychology

3Carleton University, Department of Psychology

4GamRes Limited

Corresponding address:
915 N. Broadway
Everett, WA 98201
kahlil.philander@wsu.edu
425-405-1659

Abstract

The geospatial impact of casinos on gambling problems is poorly understood, despite its

importance to policy decisions. In this study, we propose a conceptual model to describe how

access relates to gambling problems and we test whether access convenience increases risk. We

collect a large sample of Canadian gamblers (n=6,234) and geolocate each individual relative to

domestic casino locations (N=110), using their home addresses. Our analysis suggests that

nearby casinos increase risk for residents. We further find that frequency of play mediates the

relationship, implying an indirect link between access convenience and gambling problems. The

results are robust to several estimation strategies that address endogeneity issues found in the

empirical literature.

Keywords: casino; gambling exposure; access convenience; travel time; adaptation;
gambling disorder

Casino proximity, visit frequency, and gambling problems

INTRODUCTION

The prospect of a new casino in a community is often a source of controversy for nearby

residents. Economic benefits of casino expansion, including tax revenue, local employment, and

tourism activity are typically welcomed by an array of stakeholders (Eadington, 1998, 1999;

Henderson, 2006; Ishihara, 2017; Philander et al., 2015). Conversely, social costs are often the

focus of parties opposed to gambling. Categories of perceived costs are numerous, including a

loss of community values, increased risk of crime, and increased traffic congestion, but the most

cited concern is typically the harms from gambling-related addiction (Ishizaka et al., 2013; Korn

et al., 2003; C. K. Lee et al., 2010; C. K. Lee & Back, 2003, 2006; Philander et al., 2017; Wan,

2012).

Some of the smallest and largest casino projects involve debates about gambling-related

addiction in the local population. For example, Japan’s casino resorts, which are expected to

involve capital investments in excess of $10 billion per property, have generated considerable

debate about future gambling problems (Murase, 2018; Philander et al., 2017; Roberts &

Johnson, 2016). Resident attitudes about casinos are closely shaped by the direct and indirect

impacts of gambling addiction on the community (C. K. Lee et al., 2010; Vong, 2009; Wu &

Chen, 2015) and policy makers therefore attempt to minimize the former and maximize the latter

by geographically restricting property locations to island or border communities, in an attempt to

concentrate consumption among out-of-jurisdiction visitors (Eadington, 1999).

Despite the importance of understanding casino impacts on nearby residents, the

influence of access convenience on gambling problems remains an open question. There was a

significant expansion of legal gambling in the late-20th and early-21st centuries, but the

prevalence of problem gambling in related populations has tended to remain unchanged (M. W.

Abbott et al., 2014; Welte et al., 2015). This seemingly paradoxical phenomenon has led to

conflicting narratives about the influence of access by residents. Some research posited that

expansion of gambling increases risk for local populations (Korn, 2000b), while other research

proposes that individuals will adapt to the risks after an adjustment period (LaPlante & Shaffer,

2007; Prentice & Zeng, 2018). To thoughtfully develop host community plans for casinos, a

more rigorous understanding of how casino access will contribute to the local populations’

addiction risk is needed.

In the current study, we develop a conceptual model of casino patronage that helps

explain the conflicting narratives about the casino access and local gambling problems. Using a

large sample (n=6,234) of active gamblers from Canada, we then test our hypothesized pathway

using geolocation coordinates for each of the survey respondents’ homes, which is merged into a

database containing the location and age of every casino in Canada. We address potential biases

in our identification strategy by also providing robustness tests in the form of an instrumental

variable model, which we used to test assumptions important to the underlying validity of our

approach.

LITERATURE
There is a well-developed literature on the mechanisms by which casinos can contribute

to local economies (Eadington, 1998). Empirical evidence suggests that casinos positively

contributed to economic output in Macau (Zheng & Hung, 2012), Korea (C. K. Lee & Kwon,

1997), Australia (T. J. Lee, 2011), and a diverse range of U.S. jurisdictions (Evans & Topoleski,

2002; Walker & Jackson, 2007). Studies of resident perceptions find that locals recognize

positive tourism-based impacts of casinos, but have mixed attitudes to casinos overall due to the

perceived social harms (Giacopassi et al., 1999; C. K. Lee et al., 2010; Nichols et al., 2002;

Vong, 2009; Wu & Chen, 2015).

The most salient casino-related cost is problem gambling (e.g. Tan et al., 2017). Problem

gambling is a behavioral addiction, characterized by continued gambling despite negative

personal, social, and/or financial outcomes (Hodgins et al., 2011). Among other criteria,

impacted individuals are preoccupied with gambling, have unsuccessful efforts to control or stop

gambling, and return to gambling another day to recoup losses (Potenza, 2014). Individuals tend

to have excessive involvement in gambling, and this may manifest as a significant amount of

time and/or money spent at gambling outlets, and/or participation in many different forms of

gambling (LaPlante et al., 2014).

According to the Pathways Model, a prominent theory about the etiology of gambling

problems (Blaszczynski & Nower, 2002; Milosevic & Ledgerwood, 2010; Moon et al., 2017;

Nower & Blaszczynski, 2017), increased availability and accessibility of gambling is a factor in

the development of gambling related problems. It is proposed in the model that access for

individuals will enable increased frequency of play, which subsequently can lead to behavioral

conditioning and development of problems (Blaszczynski & Nower, 2002).

Empirical studies of population-wide prevalence rates fail to align with the intuition

provided by the Pathways Model. In general, prevalence studies find that rates of gambling

disorders tend stay level or fall after a casino is introduced in a community, despite the implied

increase in availability (M. W. Abbott et al., 2014; Welte et al., 2015). Many scholars propose

that this phenomenon occurs because the influence of exposure on gambling disorders can be

reduced through adaptive processes, whereby individuals change their behavior to mitigate risk

(M. Abbott, 2006; LaPlante & Shaffer, 2007; Prentice & Zeng, 2018). However, these proposed

mechanisms are propositional or explanatory. It is unclear if there is an “exposure effect” which

can then be followed by individual “adaptation”. Resolving the nature of this relationship is

important, as it will provide clarity as to the appropriate policy decisions that can be made about

casino availability, expansion, and harm mitigation.

Some empirical work has attempted to provide clarity, but these studies have been

affected by different endogeneity issues. In a meta-analysis of problem gambling prevalence

surveys from Australia and New Zealand, Storer et al. (2009) found that higher density of

electronic gambling machines is correlated with higher rates of gambling problems. Philander

(2019) found similar effects in a large representative sample of Canadians, but both studies were

based on cross-sectional data sets. Two noteworthy papers by Jacques and colleagues (Jacques et

al., 2000; Jacques & Ladouceur, 2006) compared rates of disordered gambling from a survey

conducted before and after a casino opening with rates in a similar control municipality without

expansion. Both municipalities had a decrease in prevalence rates, but the control municipality

had a greater drop. Although this appeared to be a useful natural experiment to inform policy,

because of the length of time required for effects to emerge, the difference-in-difference

approach may have been biased by the opening of a new casino less than 20 miles from the

original casino, and a U.S. tribal resort-casino opened less than 75 miles away (Morrison, 2019).

A bias in sample attrition may also explain the results, as gamblers with problems are more likely

to leave longitudinal research studies (Wohl & Sztainert, 2011).

In addition to methodological challenges observed in the empirical literature, a gap in

understanding relates to an absence of generalizable mechanisms that would explain the

connection between the nearby presence of a casino and development of gambling problems. It

seems likely that a behavioral response must occur. In the Pathways Model, consumption

changes are proposed to result from access changes (Nower & Blaszczynski, 2004).

Intuitively, frequency of visit appears to be the most likely mechanism. Increased

frequency of consumption due to closer access aligns with retail management theory. That

literature suggests convenience of access is a determining factor in purchase behavior

(Beauchamp & Ponder, 2010; Jiang et al., 2013; Seiders et al., 2007). Berry et al., (2002)

describes access convenience as consumers' perceived expenditures of time and effort to initiate

consumption behavior, and this seems particularly important for inseparable services like casino

gambling, which must be consumed while they are produced.

A growing number of observational studies suggest that gambling frequency is positively

correlated with risk of developing a gambling problem (Binde et al., 2017; Gainsbury, 2015;

LaPlante et al., 2011, 2014). Most applicable to our study, an examination resort casino

customers found that frequency of play, rather than any game or set of games played during their

casino visits, predicted a history of gambling-related problems (LaPlante et al., 2013). However,

the authors note that their modeling approach is non-causal. Although the Pathways Model

proposes a causal role for behaviorally conditioned gamblers who develop habits and subsequent

gambling problems (Blaszczynski & Nower, 2002; Nower & Blaszczynski, 2017), reverse

causality also seems plausible, considering one of the defining characteristics of gambling

problems is returning to ‘chase losses’ on a subsequent day (Potenza, 2014).

Overall, we reconcile casino access literature with the retail access literature, in a

conceptual model that explain observed and hypothesized phenomena (see Figure 1). We propose

that increased access convenience to casino locations is associated with an increase in the

prevalence of gambling disorder, and that this increase is indirect. Specifically, we hypothesize

that increased proximity of casino outlets is positively associated with frequency of play, which

in turn, is positively associated with gambling disorder. In addition, individual risk-factors such

as age, gender, and income levels (Johansson et al., 2009), or consumption of other forms of

gambling may influence both visit frequency and disorder risk.

Figure 1 – Proposed model of the indirect role of access convenience in impacting addictive disorder
risk. The hypothesized relationship between casino access and gambling disorder risk is mediated by
frequency of visits to casinos. Individual factors contribute to both visit frequency and personal
vulnerabilities to developing a gambling disorder. Other risk-factors or measures of involvement in
gambling contribute visit frequency and risk of developing a gambling disorder.

Although this is a straightforward linking of ideas from consumer behavior and problem

gambling etiology, a similar conceptual model has not been articulated (or tested) in the tourism

or broader gambling literatures. Validating this conceptual framework provides substantial clarity

to policy discussions. In our first hypothesis, we contend that there is behavioral response,

whereby closer casino access leads to increased frequency of consumption.

H1: Frequency of casino play is positively related to increased casino access convenience

We next operationalize potential for behavioral conditioning described by the Pathways

Model, and test for the role of frequency of play in contributing to gambling-related problems.

H2: Gambling disorders are positively related to frequency of casino play

We anticipate that findings provide evidence of the pathway from casino access

convenience to gambling problems via frequency of play, as well as related effect sizes, which

leads to the final hypothesis:

H3: Frequency of casino play mediates an indirect relationship from casino access

convenience to gambling disorders

We hypothesized that in addition to the direct relation described in H1 and H2, we would

observe an indirect relation that connects casino access convenience to risk of developing a

gambling disorder. We further test whether there is a direct effect of access on risk.
EMPIRICAL APPROACH

Mediation model
We estimated a mediation model like that shown in Figure 1, to test whether greater

casino proximity predicts greater gambling disorder risk indirectly via greater visit frequency.

We followed the procedure outlined by Zhao et al. (2010), and tested for the presence of an

indirect-only mediation effect using a bootstrap test (Preacher & Hayes, 2008). We used average

travel time from respondents’ home to their nearest casino as our proxy of access convenience. In

a meta-study of retail patronage, Pan & Zinkhan (2006) concluded that location convenience is

an important determinant for consumers, noting that distance and drive time meaningfully impact

shopping decisions. All models were estimated using the generalized structural equation model

(gsem) in StataMP 15 (StataCorp LLC, 2017).

Instrumental variable model

As mentioned previously, there are potential endogeneity issues in the mediation model,

as individuals at higher risk of gambling problems may choose to locate nearer casinos, leading

to potential reverse causality. To further support the validity of our findings, we estimated a set

of instrumental variable models that provided a stronger framework for inferring causality

between visit frequency and gambling disorder risk (Angrist et al., 1996; Angrist & Imbens,

1995; Baiocchi et al., 2014; Rassen et al., 2009). We also used this approach to perform a

Hausman test (Hausman, 2006) on our travel time variables, to support our claim that they are

not directly related to gambling disorder risk and are indirectly related through visit frequency.

We used a two-stage least squares (2SLS) approach that estimated the reduced form equations:

(1)

𝐷𝐺 = 𝑓(𝐶𝐹, 𝑂𝐺, 𝐷𝐸, 𝜀)

(2)

𝐶𝐹 = 𝑓(𝑇, 𝑂𝐺, 𝐷𝐸, 𝜇)

Where DG is disordered gambling risk; CF is casino gambling frequency; T is travel time

to the nearest casino; OG is other gambling variables including duration of casino presence

(Shaffer et al., 2004), number of other gambling formats played (LaPlante et al., 2014), highest

frequency of non-casino play (LaPlante et al., 2014); DE are demographic controls, and 𝜀 and 𝜇

are respective error terms. We estimated linear and ordinal versions of equation (2). We report

standard errors that account for clustering of observations by forward sortation area (discussed in

the next section) because that location is implicated in the “assignment” of estimated travel time

(Abadie et al., 2017). Individuals living in the same forward sortation area have the same

estimated travel distance to their nearest casino.

Data

Respondents

This study was determined to be exempt from review by the first author’s Institutional

Review Board. We conducted a secondary analysis of a sample of active gamblers from Canada

(Tabri et al., 2020). Casinos were not legal in Canada until 1985 (Korn, 2000a), but expanded to

110 by the time of data collection. We exploited the variation in casino locations by geocoding

each respondent and each casino through postal code information, and computing travel times to

each property for each respondent.

Since gambling disorders often afflict no more than 1-2% of the adult population at a

given time, a large sample is needed (Welte et al., 2015). The data set includes 7,980 Canadians

recruited by a third-party survey company in 2018 from their established online panel of 500,000

people. Participants were compensated with points that were redeemable for gift cards. Roughly

1,000 participants were recruited from each of the managed casino regions in Canada: British

Columbia, Alberta, Saskatchewan, Manitoba, Ontario, Quebec, Nova Scotia, and a set of

provinces that jointly manage some gambling-related operations (New Brunswick, Prince

Edwards Island, and Newfoundland and Labrador). The firm contacted potential participants and

ended recruitment once the requisite number of individuals were recruited.

Within each gambling jurisdiction, quota sampling was used to recruit an equal number

of men and women. Age groups were sampled to be representative of each province.

Participants were adults aged 18 years or older who gambled in the last year, and we subjectively

determined that 80% must have gambled in the last month. Surveying only gamblers increases

the power of the study, as our estimated effects are not impacted by attenuation bias from non-

gamblers. In a representative survey of four Canadian provinces, 49% of individuals aged 18 or

older reported gambling during the past year (Canadian Community Health Survey - Annual

Component Study Documentation, 2017). Although the non-representative sample may create

concerns that we are omitting part of the demand curve, the sample included non-casino

gamblers. Such participants are likely to be the marginal consumers who may participate in

casino gambling if a property is sufficiently close. That is, it is unlikely that a potential casino

gambler is a non-gambler altogether, and as our sampling strategy only excludes non-gamblers, it

seems likely that our approach captures consumers on the margin. Consistent with best practices,

participants who failed one or more attention check questions (n =1,317) were removed (Abbey

& Meloy, 2017), as were those who had relevant data missing (n = 360).

Of the remaining respondents, 48.4% identified as male (n = 3,049), 51.5% identified as

female (n = 3,248), and 0.1% provided an alternative gender response (n = 6). Participants

ranged in reported age from 18 to 99 years (M = 47.80, SD = 14.94). Reported household income

ranges were (in Canadian dollars) ‘under $25,000’ (6.87%), ‘$25,000 to under $40,000’

(12.47%), ‘$40,000 to under $60,000’ (16.02%), ‘$60,000 to under $80,000’ (16.48%), ‘$80,000

to under $100,000’ (14.25%), ‘$100,000 to under $150,000’ (17.02%), ‘$150,000 or more’

(7.89%), and 9.00% stated they were not comfortable providing income information. In terms of

highest level of completed education, 25.56% completed high school, 36.77% received some

post-secondary, 14.96% completed an undergraduate degree, 21.21% completed graduate-level

education. Reported marital status were ‘married’ (48.26%), ‘never married’ (21.56%), ‘married

but separated’ (2.81%), ‘divorced’ (8.01%), ‘widowed’ (3.06%), ‘common-law’ (15.48%), and

0.81% declined to provide a response.

Problem gambling severity index scores are used to measure gambling disorders (Ferris

& Wynne, 2001). The problem gambling severity index is a nine-question scale used for

measuring the severity of gambling problems in the general population. Scores vary from 0 to

27, and there are four classification categories based on that scale (Currie et al., 2013): non-

gambler/non-problem gambler (0); low-risk gambler (1-4); moderate-risk gambler (5-7); and

problem gambler (8-27).

Respondents were asked about their participation in casino and other forms of offline and
online gambling. Gambling frequency variables were measured on a seven-point ordinal scale

including: ‘never’ (never), ‘once a year’ (once yearly), ‘more than once a year but less than once

a month’ (less than monthly), ‘more than once a month but less than once a week’ (monthly),

‘once a week’ (weekly), ‘most days’ (most days), and ‘everyday’ (everyday). Two variables for

gambling involvement were computed, because these behaviors are believed to be implicated in

gambling risk (LaPlante et al., 2014). The first was the highest frequency of non-casino gambling

(depth), which ranged from 1 to 7 (M = 3.07, SD = 1.57). The second was the count of the

number of gambling formats in which the individual participated (breadth), which ranged from 0

to 6 (M = 1.84, SD = 1.57). Other demographic variables were measured, including age, gender

identity, income level, educational level, and marriage status, as these have been identified as

risk factors among Canadian gamblers (Afifi et al., 2010). Summary statistics for variables used

in the analysis are provided in Table 1.

Table 1: Summary Statistics

Problem gambling severity index
Casino visit frequency
Travel time (hours)
Travel time2
Highest freq. non-casino (depth)
Count of gambling formats (breadth)
Log of years with casino
Age cat.
Gender cat.
Income cat.
Education cat.

Count
6,277
6,299
6,262
6,262
6,303
6,303
6,303
6,303
6,303
6,303
6,303

Mean
1.81
2.15
1.24
9.93
3.07
1.84
2.83
3.79
1.52
4.46
3.89

Std. dev.
3.94
1.14
2.90
42.16
1.57
1.57
0.44
1.47
0.50
2.01
1.66

Min.
.02
.0005
1.10

Max.
15.27
233.04
3.47

Access convenience
Access convenience was estimated as the travel time between respondents’ homes and

their nearest casino. Casino location and opening date information was obtained using data from

the Alberta Gambling Research Institute (2018), provincial records (Local Government Share of

Provincial Casino and Community Gaming Centre Revenue, 2018), and individual property

searches. All locations were cross-validated in Google Earth (Gorelick et al., 2017). In total,

there were 110 properties across Canada. Respondents were asked for the first three digits of

their postal code, which is also known as the forward sortation area. Forward sortation areas are

geographically contiguous areas containing an average of 8,000 households, but can include over

60,000 households depending on density (Graebner, 2018). Forward sortation areas were turned

into geolocations by averaging the latitudes and longitudes of all six-digit postal codes in the

area.

Euclidian distances were computed for every casino and forward sortation area, and

travel times in hours by car to the nearest casino were then computed for each respondent that

provided a forward sortation area, using the georoute module in Stata (Weber & Péclat, 2019). In

Figure 2, we illustrate an example of geographic relationships appearing in this study, using a

market with multiple casinos, the Vancouver metropolitan area (ArcGIS Online, 2020). In that

market, casinos appear in many of the high-density population areas, and we illustrate that a

large proportion of the population is within a ten minute drive of one or more casinos.

Figure 2 – Sample illustration of geographic relationships appearing in this study: Metro Vancouver
area casino locations, overlaid on population density plots and sub-ten minute travel time to casino
regions by car (ArcGIS Online, 2020).

The age of the nearest casino has also been proposed as a potentially important exposure

variable (Shaffer et al., 2004), and therefore that value was computed and converted using the

natural logarithm as a measure of the duration of exposure (log of years with casino). We model

travel time using a quadratic functional form, which is consistent with prior models of retail

demand conveying a negative first-order and positive second-order effect of distance on

patronage (Barrow & Borges, 2014; Converse, 1950; Reilly, 1932).
RESULTS

Mediation model
We estimate standard errors using a bootstrap approach, but note that the use of biascorrected confidence intervals do not change our findings (Hayes & Scharkow, 2013). We

attempted 5,000 replications and all but nine converged. As shown in Figure 2, travel time has a

statistically significant effect on casino visit frequency, with a negative first-order effect, β = -

0.146, z = -7.18, p < .001, and a positive second-order effect, β = 0.006, z = 5.07, p < .001,

supporting H1. Casino visit frequency has a statistically significant effect on problem gambling

severity index scores, β = 0.758, z = 13.20, p < .001, supporting H2. The indirect effects of the

linear, β = -0.111, z = -6.19, p<.001, and squared, β = 0.005, z = 4.62, p <.001, travel time terms

are statistically significant, supporting H3.

As a robustness check, we tested whether the non-linear travel time terms varied across a

meaningful range. To do so, we calculated the marginal indirect effects of travel time, combining

both linear and squared effects of travel time at the 25th percentile = 0.202 hours, and 75th

percentile = 0.817 hours. At the 25th percentile, the marginal indirect effect of travel time was -

0.109, p<0.001, and at the 75th percentile, the marginal indirect effect of travel time was -0.103,

p<0.001, two similar results. These estimates suggest that within common travel times, the linear

term tends to dominate effect sizes.

Figure 3 – Mediation model results. Bootstrap standard errors from 5,000 replication attempts
and 4,991 replications shown in brackets; dotted lines indicate non-significant effects; PGSI =
problem gambling severity index.
*

p < 0.05, ** p < 0.01, *** p < 0.001

The significant indirect effect estimate in our model implies that frequency mediates the

impact of access on problem gambling. Some authors would suggest that the absence of a direct

effect is consistent with a fully mediated effect (Zhao et al., 2010), although others suggest that

focus should remain on magnitude and significance of indirect effects (Rucker et al., 2011). In

either case, our findings are consistent with our proposed mediation model. The related Hausman

test following the process outline by Antonakis et al. (2010) required the exclusion of the

demographic variables in order to enable the model estimator to converge, but despite that

restriction, still supports the mediator as being exogenous: cov(ufrequency, epgsi) = -.317, p=0.073).

A Hausman test using a linear regression model and using all variables also supports exogeneity:

χ2(40) = 5.75, p=1.000). Despite this empirical evidence, we relaxed the assumption of

exogeneity in the next section as a robustness test.

Instrumental variable model

To provide a baseline understanding of effects, Table 2 first presents a series of ordinary

least squares (OLS) models estimating the impacts of casino visit frequency on problem

gambling severity index scores. Models 1-6 all show relatively stable effect sizes, estimating that

a one unit increase in visit frequency along our 7-point scale is related to an average increase of

0.7 to 1.1 in problem gambling severity index score. As we noted, it is possible that these OLS

effect sizes are endogenous, due to reverse causality or omitted variable bias. The estimated

effects of other variables are as expected. Respondents’ depth and breadth of gambling are both

positively related to problem gambling severity index scores, with large effect sizes. For

example, according to our estimate in Model 3, a person that participates in all six forms of

gambling that compose our breadth variable would be projected to have a 3-point higher problem

gambling severity index score on average than a person with no participation. The inclusion of

demographic and province of residence control variables is indicated. Their addition causes our

casino age variable to cease being statistically significant. As expected, our travel time variables

(Models 4-6) were not statistically significant predictors of problem gambling severity index

score.

Table 2: OLS models (Dependent variable: Problem gambling severity index score)

Casino visit freq.

(1)
1.082***
(0.059)

Depth
Breadth
Log casino age

(2)
0.714***
(0.058)
0.330***
(0.043)
0.497***
(0.050)
0.344***
(0.096)

(3)
0.757***
(0.061)
0.310***
(0.043)
0.456***
(0.052)
0.193
(0.099)

Travel time (hours)
Travel time2
Province
Sex
Age
Income
Education
Marital status
Observations
Adjusted R2

No
No
No
No
No
No
6,274
0.098

No
No
No
No
No
No
6,274
0.176

Yes
Yes
Yes
Yes
Yes
Yes
6,274
0.207

(4)
1.107***
(0.060)

0.068
(0.078)
-0.002
(0.005)
No
No
No
No
No
No
6,234
0.099

(5)
0.694***
(0.061)
0.331***
(0.043)
0.516***
(0.052)
0.385***
(0.098)
-0.013
(0.069)
-0.001
(0.005)
No
No
No
No
No
No
6,234
0.177

(6)
0.758***
(0.061)
0.316***
(0.043)
0.455***
(0.053)
0.160
(0.100)
-0.124
(0.068)
0.005
(0.006)
Yes
Yes
Yes
Yes
Yes
Yes
6,234
0.207

Robust standard errors in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001

In Table 3, we provide reduced-form models using a 2SLS approach. Our instruments

appear to be valid. Both travel time variables were significant in all three models, showing first

and second order effects. The joint weak identification F-statistics are all above recommended

values of 10 (Baum et al., 2018; Staiger & Stock, 1997) and also exceed the computed 10%

maximal IV size (19.93) of weak instruments (Stock & Yogo, 2005). These results support H1,

suggesting frequency of casino play is positively related to convenience. Also, the Hansen-J tests

support the use of our instruments, failing to reject their exclusion from the second-stage

equation and further supporting the results of our Hausman tests (Baum et al., 2018; Hansen,

1982).

Importantly, the Hansen-J tests imply that casino proximity did not exert a direct effect

on problem gambling severity index scores in our sample, but more likely influenced gambling

risk indirectly, through frequency of visit. The estimate of the effect of casino visit frequency in

our full model (3) was larger than the equivalent OLS model, but both support H2. The fully

specified 2SLS model suggests a one unit increase in reported casino frequency was associated

with a 1.6-point increase in problem gambling severity index score.

Table 3: Two-stage least squares models (First stage dependent variable: Casino visit frequency; second
stage dependent variable: Problem gambling severity index score)

Travel time (hours)
Travel time2

Model 1
Casino
PGSI
Freq.
Score
***
-0.238
(0.028)
0.012***
(0.002)
0.598**
(0.230)

Casino visit freq.

Depth
Breadth
Log casino age
Province
Sex
Age
Income
Education
Marital status
Weak ID. (F-stat)
Hansen J-stat

Model 2
Casino
PGSI
Freq.
Score
***
-0.241
(0.022)
0.011***
(0.002)

No
No
No
No
No
No

No
No
No
No
No
No
126.89
0.81

Model 3
Casino
PGSI
Freq.
Score
***
-0.146
(0.022)
0.006***
(0.001)

1.003***
(0.169)

1.620***
(0.472)

-0.015
(0.011)
0.308***
(0.011)
0.037
(0.034)

0.337***
(0.044)
0.418***
(0.066)
0.350***
(0.097)

-0.015
(0.011)
0.312***
(0.011)
0.064*
(0.031)

0.329***
(0.045)
0.186
(0.159)
0.105
(0.109)

No
No
No
No
No
No

No
No
No
No
No
No
241.68
1.10

Yes
Yes
Yes
Yes
Yes
Yes

Yes
Yes
Yes
Yes
Yes
Yes
22.74
0.00

Hansen J p-value
Adj. R-Squared
Observations

0.37
0.08
6,234

6,234

6,234

0.29
0.17
6,234

6,234

0.94
0.16
6,234

Robust standard errors in parentheses; PGSI = problem gambling severity index
*
p < 0.05, ** p < 0.01, *** p < 0.001

our instruments, they assume that the casino visit frequency variable is linear. In Table 4, we

relaxed that assumption, and estimated an ordered probit model in the first stage. Ordered probit

models fit onto regression models with ordinal dependent variables. Larger fitted values

correspond to higher outcomes. The results were consistent with our other findings. Travel time

decreased frequency of visits at a decreasing rate, and higher visit frequencies were associated

with higher risk of a gambling disorder. 1

Table 4: Linear regression model with ordered endogenous variable (First stage dependent variable:
Casino visit frequency; second stage dependent variable: Problem gambling severity index score)

Although the models in Table 3 provide several diagnostic tests about the suitability of

Travel time (hours)
Travel time2

Model 1
Casino
PGSI
Freq.
Score
-0.253***
(0.033)
0.012***
(0.002)

Casino visit freq. (Base: Never)
Once yearly
Less than monthly
Monthly
Weekly

0.479**
(0.164)
0.842***
(0.228)
3.418***
(0.394)
4.967***
(0.620)

Model 2
Casino
PGSI
Freq.
Score
-0.285***
(0.031)
0.013***
(0.002)

0.531***
(0.152)
1.114***
(0.220)
3.398***
(0.373)
5.030***
(0.599)

Model 3
Casino
PGSI
Freq.
Score
-0.186***
(0.028)
0.008***
(0.002)

0.347**
(0.133)
0.876***
(0.177)
2.985***
(0.331)
4.400***
(0.565)

We thank a reviewer for identifying that one of the PGSI questions is related to increased
visitation and may produce some recursive results, “did you go back another day to try to in back the
money you lost?” We re-estimated Table 4 without that question contributing to the PGSI score and
produced similar results (not shown but available upon request).

9.309***
(1.419)

Most days/Everyday

Depth
Breadth
Log casino age
Province
Sex
Age
Income
Education
Marital status
Observations

No
No
No
No
No
No
6,234

No
No
No
No
No
No
6,234

9.291***
(1.362)

8.262***
(1.320)

-0.014
(0.012)
0.314***
(0.011)
0.023
(0.037)

0.282***
(0.043)
0.433***
(0.054)
0.333***
(0.096)

-0.012
(0.013)
0.323***
(0.012)
0.061
(0.035)

0.265***
(0.042)
0.435***
(0.054)
0.173
(0.099)

No
No
No
No
No
No
6,234

No
No
No
No
No
No
6,234

Yes
Yes
Yes
Yes
Yes
Yes
6,234

Yes
Yes
Yes
Yes
Yes
Yes
6,234

Robust standard errors in parentheses; PGSI = problem gambling severity index
*
p < 0.05, ** p < 0.01, *** p < 0.001

on visit frequency, and the influence of visit frequency on problem gambling severity index

scores (see Figure 4). As predicted, increased travel time was related to reductions in frequency

of play, while gambling disorder risk and standard errors increased with higher visitation. Risk

increased more quickly above a monthly frequency. At a frequency of ‘most days/everyday’,

estimated problem gambling severity index scores were roughly eight points higher than with no

casino gambling. An eight point score is a meaningful difference, as it is sufficient to be

classified as a ‘problem gambler’ on the index (Ferris & Wynne, 2001).

To illustrate the risk from higher visitation, we plotted the marginal effects of travel time

Figure 4 – Estimated impact of travel time on reported casino visit frequency category (left –
“Probability of casino play frequency”) and estimated impact of casino visit frequency on problem
gambling severity index scores, with vertical bars indicating standard errors (right – “Marginal effect on
PGSI Score”). Increases in travel time lead to a higher probability of reporting ‘never’ as a casino play
frequency (left). Higher frequency of play leads to higher PGSI score estimates, with notable rises from
‘less than monthly’ to ‘less than weekly’ and ‘weekly’ to ‘Most days or everyday’.

DISCUSSION

Casino gambling can make substantial contributions to regional economies by drawing in

visitors, but negative perceptions of casinos are shaped by potential increases in local gambling

problems (Eadington, 1998, 1999). Confounding these worthwhile debates is a set of empirical

findings on the impact of increased access convenience on problem gambling, whose mixed

results are generally unreliable due to methodological limitations. This study used a novel

empirical approach to demonstrate reliable relations among casino proximity, casino gambling,

and problem gambling risk levels, in a large data set from multiple Canadian jurisdictions. We

observed that after controlling for other gambling behavior, demographics, and selection effects,

risk of gambling problems increases for individuals who live closer to casinos. This increase in

risk was mediated by frequency of visit, and the results further suggest that access convenience

does not exert a direct effect on risk.

These findings provide substantially improved clarity to a policy area with frequent

disagreement. Our identification strategy enabled us to provide a stronger argument of a link

between high frequency casino gambling and risk of developing a gambling disorder. Because of

the challenges posed by uncontrollable covariates, past efforts to establish such relations have

been susceptible to statistical bias, and even natural experiments are vulnerable to bias due to the

complexity of contributing environmental factors to gambling disorders. By examining responses

from gamblers using an instrumental variable approach, we were able to estimate effects with

less of a concern about confounding variables. Although we use a cross-sectional survey to

collect respondent data, our variable of interest (travel time) is not identified through subjective

survey responses, but rather location data that has had an impact that precedes our collected

responses.

Noteworthy in interpreting these findings is that gambling disorder prevalence rates in

Canada fell between 2002 and 2014, despite the expansion of casino gambling locations

(Philander, 2019). Accordingly, there is reason to believe that prevention and adaptation

strategies by individuals and organizations, were effective at reducing population-level

prevalence of gambling addiction. Casinos may increase risk to nearby residents, but individual,

public, or industrial mitigation strategies to reduce disorder prevalence may have had a greater

aggregate effect. Our observation that risk rises more sharply above monthly frequencies of play

may assist managers and public health advocates in targeting interventions that can reduce the

probability of developing a gambling disorder by focusing interventions on higher frequency

players. For example, many casinos have introduced entry fees for local residents (Philander,

2017), and the schedule of payments for entrance can be tiered around frequency levels noted to

be related to higher risks.

More broadly, our findings provide an empirical basis for considering alternative location

policies in the placement of casinos. Placing casinos in “tourism zones” that have relatively few

nearby residents may be optimal, depending on policymaker goals. Where residents are

impacted, regional exposure effects should be weighed against positive social outcomes (e.g.

jobs and development) and policymakers should consider interventions that can mitigate harms

from excessive consumption. For instance, strategies such as in-venue responsible gambling

messaging, limit setting features, and employee interventions have all shown some merits in

reducing harms (Ladouceur et al., 2017; Tanner et al., 2017).

Limitations and Future Work

The current research examined a large sample of individuals who gambled, and all were

residents of the same country. The findings may not be generalizable to other cultures, and the

findings do not provide intuition about how non-gamblers may convert to casino gambling.

Similar work in other regions and with other goods would add to our broader understanding of

access convenience and public health risks. Future work should also consider the policy

implications of our findings more thoroughly. For example, research examining how to manage

risks by developing interventions that reduce access convenience for high frequency groups, or

research that assesses differentiated impacts to specific demographic groups.

There is ongoing opportunity to explore with greater detail how access convenience is

implicated in local risk. For example, it remains unclear if population-wide adaptation is related

to spontaneous recovery by individuals, or whether specific policies and programs meaningfully

contribute to protection efforts. Like many jurisdictions, regulated Canadian casino, lottery, and

online operators spend between 0.4% and 1.6% of their gross revenue from gambling on

responsible gambling programs (Canadian Partnership For Responsible Gambling & Responsible

Gambling Council, 2018), but it is unknown how this leads to adaptive effects, if at all

(Ladouceur et al., 2017).

There may be proximity related effects that are unrelated to access convenience. For

instance, increased saliency of the gambling product may occur if the casino venue is in the

individual’s neighbourhood, which may prompt increase visitation that is not directly related to

ease of travel. Future studies could accommodate these effects with instruments that capture

salience-related interventions, such as mail advertisements that are sent to specific postal codes.

Conclusion

Our findings about the mechanism implicated in problem gambling risk provide clarity to

an area of the literature that has received substantial debate, and that has important policy

implications for casino stakeholders. Casinos can have substantial positive impacts on local

economies by way of tourism, but as managers, policymakers, and other stakeholders consider

access in their communities, they should examine how residents’ travel times will change,

particularly among those that may have other risk-factors. To offset potential harms from

increases in gambling frequency, policymakers should consider interventions, including

prevention and treatment programs, if casinos are placed in close proximity to their constituents.

REFERENCES

Abadie, A., Athey, S., Imbens, G., & Wooldridge, J. (2017). When Should You Adjust Standard

Errors for Clustering? (No. w24003; p. w24003). National Bureau of Economic

Research. https://doi.org/10.3386/w24003

Abbey, J. D., & Meloy, M. G. (2017). Attention by design: Using attention checks to detect

inattentive respondents and improve data quality. Journal of Operations Management,

53–56(1), 63–70. https://doi.org/10.1016/j.jom.2017.06.001

Abbott, M. (2006). Do EGMs and problem gambling go together like a horse and carriage?

Gambling Research: Journal of the National Association for Gambling Studies

(Australia), 18(1), 7.

Abbott, M. W., Romild, U., & Volberg, R. A. (2014). Gambling and Problem Gambling in

Sweden: Changes Between 1998 and 2009. Journal of Gambling Studies, 30(4), 985–

999. https://doi.org/10.1007/s10899-013-9396-3

Afifi, T. O., Cox, B. J., Martens, P. J., Sareen, J., & Enns, M. W. (2010). Demographic and social

variables associated with problem gambling among men and women in Canada.

Psychiatry Research, 178(2), 395–400. https://doi.org/10.1016/j.psychres.2009.10.003

Alberta Gambling Research Institute. (2018). Casinos and Racinos in Canada.

Angrist, J. D., & Imbens, G. W. (1995). Two-Stage Least Squares Estimation of Average Causal

Effects in Models with Variable Treatment Intensity. Journal of the American Statistical

Association, 90(430), 431–442. https://doi.org/10.1080/01621459.1995.10476535

Angrist, J. D., Imbens, G. W., & Rubin, D. B. (1996). Identification of Causal Effects Using

Instrumental Variables. Journal of the American Statistical Association, 91(434), 444–

455. https://doi.org/10.1080/01621459.1996.10476902

Antonakis, J., Bendahan, S., Jacquart, P., & Lalive, R. (2010). On making causal claims: A

review and recommendations. The Leadership Quarterly, 21(6), 1086–1120.

https://doi.org/10.1016/j.leaqua.2010.10.010

ArcGIS Online. (2020). Environmental Systems Research Institute. https://arcg.is/HOWP8

Baiocchi, M., Cheng, J., & Small, D. S. (2014). Instrumental variable methods for causal

inference. Statistics in Medicine, 33(13), 2297–2340. https://doi.org/10.1002/sim.6128

Barrow, C. W., & Borges, D. R. (2014). Gravity models and casino gaming: A review, critique,

and modification. UNLV Gaming Research & Review Journal.
Baum, C. F., Schaffer, M. E., & Stillman, S. (2018). Instrumental Variables and GMM:

Estimation and Testing. The Stata Journal: Promoting Communications on Statistics and

Stata. https://doi.org/10.1177/1536867x0300300101

Beauchamp, M., & Ponder, N. (2010). Perceptions of retail convenience for in-store and online

shoppers. Marketing Management Journal, 20(1), 49–65. Business Source Premier.

Berry, L. L., Seiders, K., & Grewal, D. (2002). Understanding Service Convenience. Journal of

Marketing, 66(3), 1–17. https://doi.org/10.1509/jmkg.66.3.1.18505

Binde, P., Romild, U., & Volberg, R. A. (2017). Forms of gambling, gambling involvement and

problem gambling: Evidence from a Swedish population survey. International Gambling

Studies, 17(3), 490–507. https://doi.org/10.1080/14459795.2017.1360928

Blaszczynski, A., & Nower, L. (2002). A pathways model of problem and pathological gambling.

Addiction, 97(5), 487–499. https://doi.org/10.1046/j.1360-0443.2002.00015.x

Canadian Community Health Survey—Annual Component Study Documentation (Issue June).

(2017).

Canadian Partnership For Responsible Gambling, & Responsible Gambling Council. (2018).

Canadian Gambling Digest 2016-2017. https://doi.org/10.11575/PRISM/38271

Converse, P. D. (1950). New Laws of Retail Gravitation. Journal of Marketing, 14(3), 379–384.

https://doi.org/10.2307/1248191
Currie, S. R., Hodgins, D. C., & Casey, D. M. (2013). Validity of the Problem Gambling Severity

Index Interpretive Categories. Journal of Gambling Studies.

https://doi.org/10.1007/s10899-012-9300-6

Eadington, W. R. (1998). Contributions of Casino-Style Gambling to Local Economies. The
ANNALS of the American Academy of Political and Social Science, 556(1), 53–65.
Eadington, W. R. (1999). The Economics of Casino Gambling. Journal of Economic
Perspectives, 13(3), 173–192.
Evans, W. N., & Topoleski, J. H. (2002). The Social and Economic Impact of Native American

Casinos (Working Paper No. 9198; Working Paper Series). National Bureau of Economic

Research. https://doi.org/10.3386/w9198

Ferris, J., & Wynne, H. (2001). The Canadian Problem Gambling Index: User Manual. Ottawa,

ON: Canadian Centre on Substance Abuse. https://doi.org/10.1007/s10899-010-9224-y

Gainsbury, S. M. (2015). Online Gambling Addiction: The Relationship Between Internet

Gambling and Disordered Gambling. Current Addiction Reports, 2(2), 185–193.

https://doi.org/10.1007/s40429-015-0057-8

Giacopassi, D., Nichols, M., & Stitt, B. G. (1999). Attitudes of Community Leaders in New

Casino Jurisdictions Regarding Casino Gambling’s Effects on Crime and Quality of Life.

Journal of Gambling Studies, 15(2), 123–147. https://doi.org/10.1023/A:1022282208218

Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., & Moore, R. (2017). Google

Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of

Environment. https://doi.org/10.1016/j.rse.2017.06.031

Graebner, C. (2018). Statistics: Canadian Census. https://www.lib.sfu.ca/help/researchassistance/format-type/census

Hansen, L. P. (1982). Large sample properties of generalized method of moments estimators.

Econometrica: Journal of the Econometric Society. https://doi.org/10.2307/1912775

Hausman, J. A. (2006). Specification Tests in Econometrics. Econometrica.
https://doi.org/10.2307/1913827

Hayes, A. F., & Scharkow, M. (2013). The Relative Trustworthiness of Inferential Tests of the

Indirect Effect in Statistical Mediation Analysis: Does Method Really Matter?

Psychological Science. https://doi.org/10.1177/0956797613480187

Henderson, J. (2006). Betting on Casino Tourism in Asia: Singapore’s Integrated Resorts.

Tourism Review International. https://doi.org/10.3727/154427206779307196

Hodgins, D. C., Stea, J. N., & Grant, J. E. (2011). Gambling disorders. The Lancet, 378(9806),

1874–1884. https://doi.org/10.1016/S0140-6736(10)62185-X

Ishihara, H. (2017). Japanese Casino Law. Gaming Law Review and Economics, 21(1), 34–35.

Ishizaka, A., Nemery, P., & Lidouh, K. (2013). Location selection for the construction of a casino

in the Greater London region: A triple multi-criteria approach. Tourism Management, 34,

211–220. https://doi.org/10.1016/j.tourman.2012.05.003

Jacques, C., & Ladouceur, R. (2006). A Prospective Study of the Impact of Opening a Casino on

Gambling Behaviours: 2- and 4-Year Follow-ups. The Canadian Journal of Psychiatry,

51(12), 764–773. https://doi.org/10.1177/070674370605101206

Jacques, C., Ladouceur, R., & Ferland, F. (2000). Impact of Availability on Gambling: A

Longitudinal Study. The Canadian Journal of Psychiatry, 45(9), 810–815.

https://doi.org/10.1177/070674370004500904

Jiang, L. (Alice), Yang, Z., & Jun, M. (2013). Measuring consumer perceptions of online

shopping convenience. Journal of Service Management, 24(2), 191–214.

https://doi.org/10.1108/09564231311323962

Johansson, A., Grant, J. E., Kim, S. W., Odlaug, B. L., & Götestam, K. G. (2009). Risk Factors

for Problematic Gambling: A Critical Literature Review. Journal of Gambling Studies,

25(1), 67–92. https://doi.org/10.1007/s10899-008-9088-6

Korn, D. A. (2000a). Expansion of gambling in Canada: Implications for health and social
policy. CMAJ : Canadian Medical Association Journal.
Korn, D. A. (2000b). Expansion of gambling in Canada: Implications for health and social
policy. CMAJ, 163(1), 61–64.
Korn, D. A., Gibbins, R., & Azmier, J. (2003). Framing public policy towards a public health

paradigm for gambling. Journal of Gambling Studies, 19(2), 235–256.

https://doi.org/10.1023/A:1023685416816

Ladouceur, R., Shaffer, P., Blaszczynski, A., & Shaffer, H. J. (2017). Responsible gambling: A

synthesis of the empirical evidence. Addiction Research & Theory, 25(3), 225–235.

https://doi.org/10.1080/16066359.2016.1245294

LaPlante, D. A., Afifi, T. O., & Shaffer, H. J. (2013). Games and Gambling Involvement Among

Casino Patrons. Journal of Gambling Studies, 29(2), 191–203.

https://doi.org/10.1007/s10899-012-9307-z

LaPlante, D. A., Nelson, S. E., & Gray, H. M. (2014). Breadth and depth involvement:

Understanding Internet gambling involvement and its relationship to gambling problems.

Psychology of Addictive Behaviors, 28(2), 396–403. https://doi.org/10.1037/a0033810

LaPlante, D. A., Nelson, S. E., LaBrie, R. A., & Shaffer, H. J. (2011). Disordered gambling, type

of gambling and gambling involvement in the British Gambling Prevalence Survey 2007.

The European Journal of Public Health, 21(4), 532–537.

LaPlante, D. A., & Shaffer, H. J. (2007). Understanding the Influence of Gambling

Opportunities: Expanding Exposure Models to Include Adaptation. American Journal of

Orthopsychiatry. https://doi.org/10.1037/0002-9432.77.4.616

Lee, C. K., & Back, K. J. (2003). Pre- and post-casino impact of residents’ perception. Annals of
Tourism Research. https://doi.org/10.1016/S0160-7383(03)00060-4

Lee, C. K., & Back, K. J. (2006). Examining structural relationships among perceived impact,

benefit, and support for casino development based on 4 year longitudinal data. Tourism

Management. https://doi.org/10.1016/j.tourman.2004.11.009

Lee, C. K., Kang, S. K., Long, P., & Reisinger, Y. (2010). Residents’ perceptions of casino

impacts: A comparative study. Tourism Management.

https://doi.org/10.1016/j.tourman.2009.02.011

Lee, C. K., & Kwon, K. S. (1997). The Economic Impact of the Casino Industry in South Korea.

Journal of Travel Research, 36(1), 52–58. https://doi.org/10.1177/004728759703600108

Lee, T. J. (2011). Research Note: The Economic Impact of Opening a Gaming Venue in

Australia. Tourism Economics, 17(2), 457–464. https://doi.org/10.5367/te.2011.0043

Local Government Share of Provincial Casino and Community Gaming Centre Revenue. (2018).

Milosevic, A., & Ledgerwood, D. M. (2010). The subtyping of pathological gambling: A

comprehensive review. In Clinical Psychology Review.

https://doi.org/10.1016/j.cpr.2010.06.013

Moon, M., Lister, J. J., Milosevic, A., & Ledgerwood, D. M. (2017). Subtyping Non-treatment-

seeking Problem Gamblers Using the Pathways Model. Journal of Gambling Studies,

33(3), 841–853. https://doi.org/10.1007/s10899-016-9658-y

Morrison, K. (2019). World Casino Directory. https://www.worldcasinodirectory.com/

Murase, H. (2018). Casino legalization and gambling disorder: A comprehensive review and

recommendations. Journal of Health Psychology Research.

https://doi.org/10.11560/jhpr.170601075

Nichols, M., Stitt, B. G., & Giacopassi, D. (2002). Community Assessment of the Effects of

Casinos on Quality of Life. Social Indicators Research, 57(3), 229–262.

https://doi.org/10.1023/A:1014777323534

Nower, L., & Blaszczynski, A. (2004). The Pathways Model as harm minimization for youth

gamblers in educational settings. In Child and Adolescent Social Work Journal.

https://doi.org/10.1023/B:CASW.0000012347.61618.f7

Nower, L., & Blaszczynski, A. (2017). Development and validation of the Gambling Pathways

Questionnaire (GPQ). Psychology of Addictive Behaviors, 31(1), 95–109.

https://doi.org/10.1037/adb0000234

Pan, Y., & Zinkhan, G. M. (2006). Determinants of retail patronage: A meta-analytical

perspective. Journal of Retailing. https://doi.org/10.1016/j.jretai.2005.11.008

Philander, K. S. (2017). Entry fees as a responsible gambling tool: An economic analysis. UNLV
Gaming Research & Review Journal, 21(1), 4.

Philander, K. S. (2019). Regional impacts of casino availability on gambling problems: Evidence

from the Canadian Community Health Survey. Tourism Management, 71.

https://doi.org/10.1016/j.tourman.2018.10.017

Philander, K. S., Abarbanel, B. L. L., Bernhard, B. J., & Cho, R. (2017). Socio-Economic
Impacts of Japanese Integrated Resorts: Review & Recommendations.

Philander, K. S., Bernhard, B. J., Wimmer, B. S., Singh, A. K., & Eadington, W. R. (2015). U.S.

casino revenue taxes and short-run labor outcomes. Journal of Policy Modeling, 37(1).

https://doi.org/10.1016/j.jpolmod.2015.01.008

Potenza, M. N. (2014). The neural bases of cognitive processes in gambling disorder. Trends in

Cognitive Sciences, 18(8), 429–438. https://doi.org/10.1016/j.tics.2014.03.007

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and

comparing indirect effects in multiple mediator models. Behavior Research Methods.

Prentice, C., & Zeng, Z. (2018). From gambling exposure to adaptation: Implications for casino

sustainability. Journal of Retailing and Consumer Services, 41(September 2017), 31–36.

https://doi.org/10.1016/j.jretconser.2017.11.004

Rassen, J. A., Brookhart, M. A., Glynn, R. J., Mittleman, M. A., & Schneeweiss, S. (2009).

Instrumental variables I: instrumental variables exploit natural variation in

nonexperimental data to estimate causal relationships. Journal of Clinical Epidemiology.

https://doi.org/10.1016/j.jclinepi.2008.12.005

Reilly, W. J. (1932). The Law of Retail Gravitation. American Journal of Sociology.

Roberts, J., & Johnson, T. (2016). Problem Gambling: How Japan Could Actually Become the

Next Las Vegas. UNLV Gaming Law Journal, 6(2), 3.

Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E. (2011). Mediation Analysis in

Social Psychology: Current Practices and New Recommendations. Social and Personality

Psychology Compass. https://doi.org/10.1111/j.1751-9004.2011.00355.x

Seiders, K., Voss, G. B., Godfrey, A. L., & Grewal, D. (2007). SERVCON: Development and

validation of a multidimensional service convenience scale. Journal of the Academy of

Marketing Science, 35(1), 144–156. https://doi.org/10.1007/s11747-006-0001-5

Shaffer, H. J., LaBrie, R. A., & LaPlante, D. (2004). Laying the Foundation for Quantifying

Regional Exposure to Social Phenomena: Considering the Case of Legalized Gambling

as a Public Health Toxin. Psychology of Addictive Behaviors, 18(1), 40–48.

https://doi.org/10.1037/0893-164X.18.1.40

Staiger, D., & Stock, J. H. (1997). Instrumental Variables Regression with Weak Instruments.
Econometrica, 65(3), 557–586. https://doi.org/10.2307/2171753

StataCorp LLC. (2017). Stata (No. 15).

Stock, J. H., & Yogo, M. (2005). Testing for weak instruments in Linear Iv regression. In

Identification and Inference for Econometric Models: Essays in Honor of Thomas

Rothenberg. https://doi.org/10.1017/CBO9780511614491.006

Storer, J., Abbott, M., & Stubbs, 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(3), 225–244.

https://doi.org/10.1080/14459790903257981

Tabri, N., Wood, R. T. A., Philander, K., & Wohl, M. J. A. (2020). An examination of the validity

and reliability of the Positive Play Scale: Findings from a Canadian national study.

International Gambling Studies, 20(2), 282–295.

https://doi.org/10.1080/14459795.2020.1732442

Tan, A.-L., Lee, T. J., & Kim, J. J. (2017). Resident perceptions of casinos in a newly developed

casino destination. International Journal of Tourism Sciences, 17(1), 15–31.

https://doi.org/10.1080/15980634.2016.1274175

Tanner, J., Drawson, A. S., Mushquash, C. J., Mushquash, A. R., & Mazmanian, D. (2017). Harm

reduction in gambling: A systematic review of industry strategies. Addiction Research &

Theory, 25(6), 485–494. https://doi.org/10.1080/16066359.2017.1310204

Vong, F. (2009). Changes in Residents’ Gambling Attitudes and Perceived Impacts at the Fifth

Anniversary of Macao’s Gaming Deregulation. Journal of Travel Research, 47(3), 388–

397. https://doi.org/10.1177/0047287508322787

Walker, D. M., & Jackson, J. D. (2007). Do Casinos Cause Economic Growth? American

Journal of Economics and Sociology, 66(3), 593–607. https://doi.org/10.1111/j.1536-

7150.2007.00528.x

Wan, Y. K. P. (2012). The social, economic and environmental impacts of casino gaming in

Macao: The community leader perspective. Journal of Sustainable Tourism.

https://doi.org/10.1080/09669582.2011.636818

Weber, S., & Péclat, M. (2019). A Simple Command to Calculate Travel Distance and Travel

Time. The Stata Journal: Promoting Communications on Statistics and Stata.

https://doi.org/10.1177/1536867x1801700411

Welte, J. W., Barnes, G. M., Tidwell, M.-C. O., Hoffman, J. H., & Wieczorek, W. F. (2015).
Gambling and Problem Gambling in the United States: Changes Between 1999 and 2013.

Journal of Gambling Studies, 31(3), 695–715. https://doi.org/10.1007/s10899-014-9471-

Wohl, M. J. A., & Sztainert, T. (2011). Where Did All the Pathological Gamblers Go? Gambling

Symptomatology and Stage of Change Predict Attrition in Longitudinal Research.

Journal of Gambling Studies, 27(1), 155–169. https://doi.org/10.1007/s10899-010-9186-

Wu, S.-T., & Chen, Y.-S. (2015). The social, economic, and environmental impacts of casino

gambling on the residents of Macau and Singapore. Tourism Management, 48, 285–298.

https://doi.org/10.1016/j.tourman.2014.11.013

Zhao, X., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and Truths

about Mediation Analysis. Journal of Consumer Research, 37(2), 197–206.

https://doi.org/10.1086/651257

Zheng, V., & Hung, E. P. W. (2012). Evaluating the Economic Impact of Casino Liberalization in

Macao. Journal of Gambling Studies, 28(3), 541–559. https://doi.org/10.1007/s10899-

011-9251-3