Self-generated motives for gambling in two population-based samples of gamblers

23
This article was downloaded by: [Dalhousie University] On: 18 July 2013, At: 10:09 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK International Gambling Studies Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rigs20 Self-generated motives for gambling in two population-based samples of gamblers Daniel S. McGrath a , Sherry H. Stewart a , Raymond M. Klein a & Sean P. Barrett a a Department of Psychology, Dalhousie University, Halifax, Nova Scotia, Canada Published online: 11 Aug 2010. To cite this article: Daniel S. McGrath , Sherry H. Stewart , Raymond M. Klein & Sean P. Barrett (2010) Self-generated motives for gambling in two population-based samples of gamblers, International Gambling Studies, 10:2, 117-138, DOI: 10.1080/14459795.2010.499915 To link to this article: http://dx.doi.org/10.1080/14459795.2010.499915 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions

Transcript of Self-generated motives for gambling in two population-based samples of gamblers

This article was downloaded by: [Dalhousie University]On: 18 July 2013, At: 10:09Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Gambling StudiesPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rigs20

Self-generated motives for gamblingin two population-based samples ofgamblersDaniel S. McGrath a , Sherry H. Stewart a , Raymond M. Klein a &Sean P. Barrett aa Department of Psychology, Dalhousie University, Halifax, NovaScotia, CanadaPublished online: 11 Aug 2010.

To cite this article: Daniel S. McGrath , Sherry H. Stewart , Raymond M. Klein & Sean P. Barrett(2010) Self-generated motives for gambling in two population-based samples of gamblers,International Gambling Studies, 10:2, 117-138, DOI: 10.1080/14459795.2010.499915

To link to this article: http://dx.doi.org/10.1080/14459795.2010.499915

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Self-generated motives for gambling in two population-based samplesof gamblers

Daniel S. McGrath*, Sherry H. Stewart, Raymond M. Klein and Sean P. Barrett

Department of Psychology, Dalhousie University, Halifax, Nova Scotia, Canada

In the present study, self-generated responses to a question regarding reasons forgambling from two epidemiological surveys were combined and placed into anotherearlier motivational model for alcohol use, adapted for gambling. Of the 3601 reasons,954 could be categorised into the model’s categories: (a) coping motives (internal,negative reinforcement); (b) enhancement motives (internal, positive reinforcement);and (c) social motives (external, positive reinforcement). Results indicate that copinggamblers experienced greater gambling severity and psychopathology, enhancementgamblers were most likely to gamble while intoxicated and social gamblers were morelikely to choose socially-related gambling. An examination of remaining motivessuggests additional categories may be warranted – specifically financial and charitablereasons. These findings offer some support for the model; however, it may need to beexpanded to account for other motives. The study highlights the advantages andlimitations of using self-generated reasons to study gambling motivation.

Keywords: gambling; motives; epidemiology; comorbidity; subtyping

Introduction

Motivational models of addiction have proven useful in understanding certain addictive

behaviours, such as alcohol and other substance use/abuse (e.g. Cooper, Russell, Skinner

& Windle, 1992; Cooper, 1994; Conrod, Pihl, Stewart & Doniger, 2000). Suggestions

abound regarding the contribution of motivations involving emotional self-regulation in

the etiology and maintenance of pathological gambling. For example, some suggest that

problem gamblers engage in gambling either for its tension- or dysphoria-reducing effects

(Beaudoin & Cox, 1999), or for its euphoric consequences (Hickey, Haertzen &

Henningfield, 1986). Many motivational models of addiction argue that desires for mood

alteration underlie addictive behaviours (Cox & Klinger, 1988). Recent research in the

drug abuse literature supports the contention that there are subgroups of substance abusers

who use drugs/alcohol for different reasons, and that these differing underlying

motivations are useful in predicting drug preferences and co-morbid psychopathology

(Conrod et al., 2000). Despite frequent suggestions of the importance of motives to

gambling behaviour (Beaudoin & Cox, 1999; Hickey et al., 1986; Blaszczynski & Nower,

2002), little empirical research has examined this construct in the gambling area.

One example of a motivational model that has proven extremely useful in the broader

addictions area is the model developed by Cooper and colleagues (1992) to explain

excessive and problem drinking. Cooper contends that people drink to obtain certain desired

rewards (Cooper et al., 1992; Cooper, 1994). She argues that the various outcomes desired

ISSN 1445-9795 print/ISSN 1479-4276 online

q 2010 Taylor & Francis

DOI: 10.1080/14459795.2010.499915

http://www.informaworld.com

*Corresponding author. Email: [email protected]

International Gambling Studies

Vol. 10, No. 2, August 2010, 117–138

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

from drinking can be organised along two dimensions. The first pertains to the valence of the

desired outcome. People may drink to obtain a positive incentive (positive reinforcement) or

to avoid a negative incentive (negative reinforcement). The second dimension pertains to

the source of the outcome desired from drinking: drinkers may drink to change something

within themselves such as a mood state (i.e. internal) or to change something in their social

environment (i.e. external). When these two dimensions are crossed, three specific drinking

motives applicable to adults result: coping motives (COP) (internal, negative

reinforcement; i.e. to reduce or avoid negative emotions); enhancement motives (ENH)

(internal, positive reinforcement; i.e. to increase positive emotions); and social motives

(SOC) (external, positive reinforcement motives; i.e. to increase social affiliation).

Cooper et al.’s model also seems readily applicable to gambling behaviour (Stewart &

Zack, 2008). However, research has only recently begun to examine the utility of this model

as applied to gambling. In a study by Stewart, Zack, Collins, Klein and Fragopoulos (2008),

158 pathological gamblers who typically drink while gambling completed self-report

measures including the Inventory of Gambling Situations (IGS; Littman-Sharp, Turner &

Toneatto, 2009). Three clusters of pathological gamblers were identified based on IGS

scores: ‘ENH gamblers’ who showed high scores on a positive situations IGS factor (e.g.

gambling in response to pleasant emotions), and low scores on a negative situations IGS

factor (e.g. gambling in response to unpleasant emotions) (59% of the sample); ‘COP

gamblers’ who showed high scores on both IGS factors, but particularly high scores on the

negative factor (23%); and ‘low emotion regulation gamblers’ who showed low scores on

both factors (18%). Gamblers in the first two clusters gamble for emotional regulation

reasons and appear to correspond closely to Blaszczynski and Nower’s (2002) ‘antisocial

impulsivist’ and ‘emotionally vulnerable’ subgroups, respectively. Motivations for

gambling in the third cluster appear to pertain little to emotional regulation; this group

corresponds most closely to Blaszczynski and Nower’s (2002) ‘behaviourally conditioned’

subgroup. Stuart, Stewart, Wall and Katz (2008) aimed to extend this motivation-based sub-

typing scheme to 180 undergraduate gamblers. Participants completed the IGS, and again

the same three clusters emerged: Low emotion regulation gamblers (45% of the sample),

ENH gamblers (42%) and COP gamblers (13%). Only the distribution of gamblers’

subtypes differed from the original study with pathological gamblers (i.e. more low emotion

regulation and fewer ENH). Both COP- and ENH-motivated undergraduates scored higher

on the Problem Gambling Severity Index (PGSI; Ferris & Wynne, 2001), reported greater

gambling frequency and time spent gambling, than low emotion regulation gamblers.

Lastly, ENH gamblers reported spending more money gambling than low emotion

regulation gamblers. Taken together, these results suggest that sub-classification of

gamblers according to motives for gambling might be a theoretically useful way of

understanding the diversity among problem and non-problem gamblers.

In both of the above-mentioned studies (Stewart et al., 2008; Stuart et al., 2008),

gambling motives were assessed using the IGS (Littman-Sharp et al., 2009) which assesses

motives indirectly. More specifically, gambling motives are inferred on the basis of

responses regarding contexts in which gambling takes place. For example, when a gambler

strongly endorses gambling in response to situations involving conflict with others (e.g. to

deal with interpersonal problems with family/loved ones), it is inferred that the underlying

motive for gambling is to cope with negative emotions. Of course, this is an assumption. One

alternative method would be to ask gamblers more directly about their primary reason(s)

for gambling, using a questionnaire. This was the approach taken in the development

of the Gambling Motives Questionnaire (GMQ; Stewart & Zack, 2008) modelled after

the Drinking Motives Questionnaire (DMQ; Cooper et al., 1992). The DMQ is based on

118 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Cooper et al.’s three-factor model of drinking motives mentioned previously. The GMQ

was designed to assess a gambler’s relative frequency of gambling across 15 items

comprised of three subscales: social, coping and enhancement motives. It has been shown to

possess strong psychometric properties among community-recruited gamblers (Stewart &

Zack, 2008). Another method to capture reasons for gambling would be to ask gamblers to

self-generate motives (see Neighbors, Lostutter, Cronce & Larimer, 2002; Pantalon,

Maciejewski, Desai & Potenza, 2008). Rather than constraining their responses as in

a questionnaire, gamblers describe in an open-ended fashion their primary reason(s)

for gambling. While using open-ended queries can pose challenges (e.g. cost, coding,

lengthening questionnaires), this method may be advantageous for obtaining information

that is uncontaminated by the researcher’s point of view and providing information

about each gambler’s most readily accessible motives as these are typically generated first

(Stacy, Leigh & Weingardt, 1994). It is this latter approach that was the focus of the current

investigation.

The goal of the current study was to examine the utility of categorising gamblers

according to their own self-generated motives for gambling as elicited from two

epidemiological gambling surveys conducted in the Canadian provinces of Alberta (Smith

& Wynne, 2002) and Newfoundland and Labrador (Market Quest Research Group Inc.,

2005). Respondents’ first-generated motives were initially coded into four categories based

on Cooper et al.’s (1992) motivational model, adapted for gambling. The motives categories

included: coping (COP) (internal, negative reinforcement), enhancement (ENH) (internal,

positive reinforcement), social (SOC) (external, positive reinforcement), and other (OTH).

The ‘other’ motives were those that failed to fit within the context of Cooper et al.’s model.

Gambling motives categories were then compared according to a number of criterion

variables. We expected to observe: (1) a greater rate of endorsement of COP by women

gamblers and ENH by men gamblers given previous suggestions that men and women

gamble for different reasons (e.g. Burger, Dahlgren & MacDonald, 2006; Stewart & Zack,

2008). Also, consistent with what has been observed in the alcohol literature (see Cooper

et al., 1992; Cooper, 1994), we hypothesised that: (2) COP and ENH gamblers would

be characterised by greater levels of gambling involvement and a greater severity of

gambling problems than SOC gamblers. It was also predicted that: (3) SOC would be related

to involvement in those activities practiced in a social context (e.g. card games); COP would

be related to those activities involving greater absorption/attentional distraction (e.g. slots);

and ENH would be related to more ‘exciting’ gambling activities (e.g. sports betting). These

exciting forms of gambling could be considered ‘active’ forms of gambling as they involve

some degree of skill, planning, and increased arousal, while activities involving attentional-

distraction could be considered more ‘passive’ (Bonnaire, Bungener & Varescon, 2006). It

was also predicted that COP and ENH would be strongly associated with particularly ‘risky’

gambling activities in terms of their addictive potential such as electronic gambling (Ellery,

Stewart & Loba, 2005; Focal Research, 1998). We also predicted that: (4) COP would be

associated with greater endorsement of suicidality and treatment for stress-related illnesses

given the similarity of COP to Blaszczynski and Nower’s (2002) emotionally vulnerable

gambler subtype who are said to be prone to mood and anxiety-related psychopathology.

Group differences on these indices of psychopathology may have important implications for

the treatment of gambling problems, especially among COP gamblers. Finally, it was

predicted that: (5) ENH gamblers would be more likely to gamble when drunk/high, as ENH

gamblers are more sensitive to heart rate increases associated with combined drinking and

gambling (an index of susceptibility to psychostimulation), suggesting greater risk of using

alcohol while gambling (Stewart, 2009).

International Gambling Studies 119

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Method

Survey design

Data for the current investigation were compiled from both the Newfoundland and

Labrador-Gambling Prevalence Survey (Market Quest Research Group Inc., 2005) and the

Alberta Canadian Problem Gambling Index Questionnaire (Smith & Wynne, 2002). Both

datasets were previously integrated by the Ontario Problem Gambling Research Centre

(OPGRC) as part of a secondary analysis of population-based surveys that employed

the use of the CPGI (Ferris & Wynne, 2001). The Newfoundland and Labrador Survey

contained responses from a province-wide sample of 2596 adult respondents 19 years of

age and older (52% women), with 83% (n ¼ 2154) reporting at least some gambling in the

past 12 months. The overall response rate was not available. Sampling was stratified by

region but was otherwise random. The Alberta Survey had a response rate of 64% and

contained responses from a province-wide sample of 1804 adult respondents 18 years of

age and older (50% women), with 82% (n ¼ 1480) reporting at least some gambling in the

past 12 months. Sampling was stratified by region and by gender but was otherwise

random. While the data included weights at the provincial level, the current study used

unweighted data. Only those reporting that they had gambled in the previous 12 months

were included in our analysis. Thus the combined sample included in this report was

comprised of a total of 3634 adult gamblers.

Measures

Open-ended gambling motives assessment

Both surveys contained a common open-ended question asking respondents to give their

main reasons for why they gamble. Only individuals who gambled in at least one activity

over the previous 12 months were able to respond to the question. No reasons data were

collected from the non-gamblers. For Newfoundland and Labrador, participants were

asked to respond to the question “What are the main reasons why you gamble?” For

Alberta, participants were asked to respond to the stem “What are the main reasons why

you participate in ________ [name of gambling activity]?” In the Alberta survey only, this

question was repeated for each gambling activity endorsed. Only the first response (i.e.

most top of mind) per gambling activity was included in our analyses because the first self-

generated motive was presumed to be the most accessible and have the greatest influence

on behaviour (Stacy et al., 1994; Wiers et al., 2002). In both surveys, the first self-

generated gambling motive was then coded into one of ten answer categories by the

individual conducting the survey, such as “In order to do things with your friends” or “for

excitement or as a challenge”. Participant responses that could not be coded directly into a

category had their verbatim answer recorded separately. The first responses from both

surveys were subsequently combined by our team and placed in the context of Cooper and

colleagues’ (1992) model, adapted for gambling. For instance, if a participant’s first self-

generated reason for gambling was “for excitement or as a challenge”, it was inferred that

their principal motivation for gambling was ENH. For the Alberta survey, a respondent

was placed into a motives category based on their most commonly endorsed motive for all

of their gambling activities (e.g. when an ENH reason was offered more often than any

other reason). For cases in which more than one motive was equally endorsed, a composite

score was used to identify the most relevant gambling activity. Participants were not asked

to indicate their ‘favourite’ form of gambling; therefore the most relevant activity was

inferred based on level of involvement. The composite score was comprised of the average

120 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

standardised values (each variable was converted to z-scores) of three variables for each

gambling activity: the amount of dollars spent, the number of minutes played, and the

largest amount gambled in the past 12 months. The motive associated with the gambling

activity that had the highest average score was inferred to be the most relevant reason for

gambling and was used to place the respondent into a motive category from Cooper et al.’s

model. Specifically, all reasons for gambling for the combined datasets were coded as

reflecting primarily COP motives, primarily ENH motives, primarily SOC motives,

or OTH motives. The uncategorised ‘other’ responses were then further examined in

an effort to identify additional gambling motives based on the frequency of which they

were mentioned. Three additional broad reasons were identified: financial-related (FIN),

charity-related (CHA), and recreation-related (REC) reasons for gambling (see Table 1 for

coding scheme). Inter-rater reliability was computed for the full coding scheme following

the recoding of a random subset (20%) of responses by a second independent rater and was

found to be very high, k ¼ 0.99 ( p ¼ 0.000), 95% CI (0.976, 1.00).

Demographic variables

The demographic items included in both surveys were: age; marital status; education;

current employment status; number of children in home; sex; and total household income.

Most demographic questions were identical across the two surveys; however, in some

Table 1. Categorisation of reasons for gambling into Cooper’s model adapted for gambling andadditionally identified gambling categories.

Motives’ groups NResponses to the common question onreasons for gambling

Cooper’s model Coping 9 . I can forget about my problems. To be alone. To distract yourself from everyday problems

Social 253 . It is an opportunity to socialize. Group thing at work. In order to do things with your friends

Enhancement 692 . It is exciting/fun. It decreases my boredom. Entertainment. For excitement or as a challenge. For entertainment or fun

Newly identifiedgroups

Financial 1655 . I can win money. If jackpot is high. Try luck/take a chance/ hope to win. To win money

Charitable 694 . To support worthy causes/charities. Gifts

Recreation 65 . It’s a hobby. As a hobby. Because I am good at it. Because you are good at it

‘Other’ 233 . Out of curiosity. Change leftover after buying something. It is always right in front of you at stores. Other. For some other reason. Do not know

International Gambling Studies 121

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

instances (e.g. income) answer categories were collapsed for the purpose of creating a

common metric.

Gambling activities and gambling involvement

Both surveys included questions on past 12-month involvement in specific gambling

activities, the frequency of involvement with each activity, and money spent on each

activity. Most questions were worded identically; however, when not worded identically,

categories were combined to create variables that were equivalent whenever possible.

Although there is no information on psychometric validation of these items, the wording is

similar to items appearing on other gambling surveys and similar items have been shown

to be sensitive to changes induced by problem gambling interventions, for example,

attesting to validity (e.g. Doiron & Nicki, 2007).

Canadian problem gambling index (CPGI; Ferris & Wynne, 2001)

The CPGI was designed to measure problem gambling within the general population.

Embedded in the CPGI is the 9-item PGSI. It is designed to differentiate between non-

problem, low and moderate risk gamblers, and problem gamblers, and has good internal

consistency (a ¼ 0.84), test–retest reliability over a 3–4-week period (r ¼ 0.78), and

good convergent validity with the South Oaks Gambling Screen (Lesieur & Blume, 1987)

and the DSM-IV (American Psychiatric Association [APA], 1994) pathological gambling

criteria (both r ¼ 0.83) (Ferris & Wynne, 2001).

Co-morbid psychopathology

Two indices of psychopathology were common to the two surveys. The first pertains to

medical care for stress-related illness in the past 12 months. Secondly, both questionnaires

asked about serious suicidal ideation or attempts in the past 12 months. Both questions

were answered in a forced choice ‘yes/no’ format. These questions were used to test

predictions regarding psychopathology as a function of primary gambling motives.

Intoxication while gambling

An index of intoxication while gambling was common to the two surveys. The question

asked respondents if they had gambled while drunk or high in the past 12 months. Answers

were in a forced choice ‘yes/no’ format. This question was used to test the predictions

regarding substance misuse while gambling across gambling motives groups.

Results

Participant characteristics

The sample was comprised of 3634 gamblers (50.2% women, mean age ¼ 44.0 years,

SD ¼ 13.9). All respondents were 18 years of age or over. The majority of individuals in

the sample reported being married or living with a common law partner (70.5%), with

17.0% reporting being single and 12.5% being widowed or divorced. Most respondents

were employed either full or part-time (62.5%); the rest were retired (18.1%), were

homemakers (7.1%), or were unemployed (8.3%). The majority of respondents (60.2%)

reported completing/attending post-secondary schooling with 24.2% completing high

122 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

school, and 15.6% reporting some high school or less. Lastly, the average reported annual

household income of the combined sample was CAN$51,557.

The average PGSI (Ferris & Wynne, 2001) score for the gamblers was 0.50 (SD ¼ 2.13)

with 85.7% classified as ‘non-problem gamblers’, 8.9% ‘low-risk gamblers’, 3.7%

‘moderate-risk gamblers’ and 1.7% ‘problem gamblers’ (i.e. a PGSI score of 8 or above).

The average gambler spent money on 2.9 (SD ¼ 1.7) gambling activities over the past 12

months with the highest rates of participation being: lottery (81.9%), raffle tickets (56.8%),

scratch/break-open tickets (45.0%) and video lottery terminals (VLTs)/slot machines

(19.0%). (The values do not add to 100% as gambling activities are not mutually exclusive.)

Categorisation of first-generated reasons for gambling

The first-generated reasons for gambling were placed into the context of Cooper et al.’s

(1992) model. Of the 3601 reasons given, 26.5% (N ¼ 954) were judged to be a fit for the

three categories of motives in the model. These reasons for gambling were categorised as

‘ENH motives’ (N ¼ 692), ‘SOC motives’ (N ¼ 253) and ‘COP motives’ (N ¼ 9). It was

determined that the remaining 2647 reasons could not clearly be placed into the three

categories of motives.

A large proportion (73.5%,N ¼ 2647) of the first-generated reasons for gambling failed

to fit within the context of Cooper et al.’s (1992) model. Thus, we attempted to identify

further categories that could capture other motivations for gambling based on the relative

frequency of responses. For instance, 46.0% (N ¼ 1655) of the responses described

‘winning money’ as the primary motivation for gambling. These were subsequently

categorised as ‘Financial Motives’ (FIN). The large proportion of gamblers motivated by

the possibility of financial gain is consistent with previous reports (e.g. Neighbors et al.,

2002). Additionally, 19.3% (N ¼ 694) described charity (e.g. charity raffle tickets) as the

primary motivation for gambling. Such responses were placed into a category termed

‘Charitable Motives’ (CHA). Also, 1.8% (N ¼ 65) of reasons described recreation (e.g. it is

a hobby) as the motivation for gambling and were placed into a ‘Recreational Motives’

category (REC). The remaining ‘Other’ reasons (6.5%,N ¼ 233) did not warrant placement

into additional categories of motives given the individual low frequency of these motives.

For instance, responses such as ‘left-over change’ or ‘something to do’ were too infrequent,

ambiguous, or lacked sufficient meaning to warrant classification.

Reasons for gambling in each provincial survey

A x 2 analysis revealed that the two surveys (Alberta and Newfoundland) differed on

representation of the seven motive groups, x 2 (6) ¼ 153.02, p ¼ 0.000. An equal

proportion of COP motives were present in both surveys (, 0.1%). Both provinces also

had an equal proportion of FIN motives (46.0%) and a roughly equal proportion of CHA

motives (19.2% in Alberta vs 19.3% in Newfoundland). However, there were a higher

proportion of SOC motives in the Newfoundland survey (10% vs 2.7%) and ENH motives

in the Alberta survey (21.8% vs 17.4%). And a larger proportion of REC motives were

reported in the Newfoundland survey (2.9% vs 0.3%) and more OTH motives were found

in the Alberta sample (9.9% vs 4.1%).

Data analyses

Given the small sample associated with the COP group (N ¼ 9), each of the hypotheses

was tested by comparing across motives groups (e.g. COP vs ENH vs SOC) using

International Gambling Studies 123

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

non-parametric statistics for continuous (e.g. gambling problems) and categorical

(e.g. types of gambling) outcome measures. As a result of heterogeneity of variance

within the dataset, Kruskal–Wallis tests were used for comparison in the case of

continuous measures, with Mann–Whitney U tests performed to examine pairwise

comparisons when significant main effects were found. In the case of categorical

variables, groups of motives were compared using Fisher’s exact tests across dichotomous

outcome measures. Lastly, some of the variables that were originally categorical variables

(e.g. age and income) were re-coded as continuous measures using a method described in

Wicki, Gmel, Kuntsche, Rehm and Grichting (2006). This transformation involves using

the mid-points of categories as the mean for that category. For instance, the first age

category (18–24 years) was recoded as 21 years. Lastly, to determine if additional

categories were warranted, we conducted analyses comparing gamblers who reported

FIN, CHA, REC or OTH reasons for gambling to the Cooper et al.’s (1992) model

categories in terms of: PGSI score, number of gambling activities played and money spent

gambling in the previous 12 months.

Gender, age, and income composition of motives groups

We also conducted Fisher’s exact tests to determine whether there were differences

between the groups of gambling motives on gender or age. The COP, SOC and ENH

groups of motives were comprised of 67%, 58% and 49% women, respectively. Using

Fisher’s exact tests, no significant differences in gender composition were found for COP

vs SOC ( p ¼ 0.739); and COP vs ENH ( p ¼ 0.336); however, significant differences

were found between SOC vs ENH gamblers ( p ¼ 0.019) with a larger proportion of

women endorsing SOC motives (58.1%) compared to ENH (49.3%). While our

predictions of a greater rate of endorsement of COP motives by women gamblers and of

ENH motives by men were not supported, the greater rate of endorsement of SOC motives

among women is consistent with previous findings using a questionnaire method

(e.g. Stewart & Zack, 2008). However, given the larger proportion of women in the COP

group (67%), it is conceivable that low power owing to the limited number of COP

gamblers (N ¼ 9) in the sample may have contributed the null gender findings for the

COP category. Next, we examined the gender composition of the remaining groups of

motives. The REC, FIN, CHA and OTH groups of motives were comprised of 48%, 47%,

57% and 52% women, respectively. Fisher’s exact tests revealed three significant

differences: a greater proportion of women in the SOC vs the FIN group ( p ¼ 0.001), the

CHA vs the ENH group ( p ¼ 0.003), and the CHA vs the FIN group ( p ¼ 0.000).

A Kruskal–Wallis test conducted to compare the categories of motives by age

revealed significant differences, x 2 (6) ¼ 79.51, p ¼ 0.000. The mean ages for the groups

of motives were: ENH (M ¼ 40.4 years, SD ¼ 15.0), SOC (M ¼ 43.0 years, SD ¼ 15.2),

COP (M ¼ 42.8 years, SD ¼ 11.1), REC (M ¼ 47.7 years, SD ¼ 14.3), FIN (M ¼ 45.1

years, SD ¼ 13.2), CHA (M ¼ 45.7 years, SD ¼ 13.0) and OTH (M ¼ 42.1 years,

SD ¼ 14.1). Mann–Whitney U tests revealed that ENH gamblers were younger than SOC;

however, both groups were younger than REC, FIN and CHA. The OTH group was also

younger on average than the REC, FIN and CHA groups.

Lastly, a Kruskal–Wallis test comparing all seven categories of motives by income

(see Table 2) revealed significant differences, x 2 (6) ¼ 22.80, p ¼ 0.001. Mann–Whitney

U tests revealed that REC gamblers earned less than ENH, FIN and OTH gamblers. The

OTH group also earned more than the SOC group. Finally, the CHA group earned more

money on average than the ENH, SOC, REC and FIN groups.

124 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Tab

le2

.M

ann

–W

hit

ney

Ure

sult

sfo

ran

nu

alh

ou

seh

old

inco

me

(in

do

llar

s)b

yse

ven

gro

up

so

fg

amb

lin

gm

oti

ves

Mo

tiv

es’

gro

up

sM

ean

ran

kM

ean

(SD

)

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

UP

r

CO

PS

OC

10

5.8

69

8.2

35

0,7

14

(24

,39

8)

49

,07

4(2

9,2

37

)6

10

.00

.72

50

.03

CO

PE

NH

27

2.8

62

70

.47

50

,71

4(2

4,3

98

)5

1,8

67

(30

,58

3)

18

49

.00

.96

80

.01

CO

PR

EC

35

.29

28

.12

50

,71

4(2

4,3

98

)4

1,1

00

(27

,11

0)

13

1.0

0.2

79

0.1

4C

OP

FIN

59

4.8

66

16

.62

50

,71

4(2

4,3

98

)5

3,3

06

(29

,60

5)

41

36

.00

.87

10

.01

CO

PC

HA

24

4.4

32

73

.88

50

,71

4(2

4,3

98

)5

6,5

49

(29

,69

0)

16

83

.00

.62

10

.02

CO

PO

TH

82

.86

91

.33

50

,71

4(2

4,3

98

)5

5,7

47

(29

,94

2)

55

2.0

0.6

73

0.0

3

SO

CE

NH

34

9.2

03

65

.86

49

,07

4(2

9,2

37

)5

1,8

67

(30

,58

3)

48

,04

4.5

0.3

42

0.0

4S

OC

RE

C1

24

.02

10

4.8

04

9,0

74

(29

,23

7)

41

,10

0(2

7,1

10

)3

96

5.0

0.0

78

0.1

1S

OC

FIN

65

5.0

97

15

.59

49

,07

4(2

9,2

37

)5

3,3

06

(29

,60

5)

10

,58

57

.00

.05

60

.05

SO

CC

HA

32

4.2

13

78

.63

49

,07

4(2

9,2

37

)5

6,5

49

(29

,69

0)

43

,32

1.5

0.0

02

*0

.11

SO

CO

TH

17

0.0

41

94

.99

49

,07

4(2

9,2

37

)5

5,7

47

(29

,94

2)

14

,18

3.0

0.0

23

*0

.12

EN

HR

EC

29

7.0

52

38

.17

51

,86

7(3

0,5

83

)4

1,1

00

(27

,11

0)

10

,63

3.5

0.0

17

*0

.10

EN

HF

IN8

58

.32

88

8.7

15

1,8

67

(30

,58

3)

53

,30

6(2

9,6

05

)3

1,5

17

6.0

0.2

45

0.0

3E

NH

CH

A5

10

.63

56

2.0

85

1,8

67

(30

,58

3)

56

,54

9(2

9,6

90

)1

2,9

85

7.0

0.0

06

*0

.08

EN

HO

TH

34

6.6

53

76

.51

51

,86

7(3

0,5

83

)5

5,7

47

(29

,94

2)

42

,45

4.5

0.0

92

0.0

6

RE

CF

IN4

88

.65

64

4.1

04

1,1

00

(27

,11

0)

53

,30

6(2

9,6

05

)2

3,1

57

.50

.00

3*

0.0

8R

EC

CH

A2

12

.49

30

2.6

54

1,1

00

(27

,11

0)

56

,54

9(2

9,6

90

)9

34

9.5

0.0

00

*0

.15

RE

CO

TH

86

.93

11

9.8

54

1,1

00

(27

,11

0)

55

,74

7(2

9,9

42

)3

07

1.5

0.0

01

*0

.21

FIN

CH

A8

65

.73

92

0.6

05

3,3

06

(29

,60

5)

56

,54

9(2

9,6

90

)3

09

,60

0.0

0.0

36

*0

.05

FIN

OT

H6

95

.46

73

1.9

65

3,3

06

(29

,60

5)

55

,74

7(2

9,9

42

)1

01

,01

3.5

0.2

62

0.0

3

CH

AO

TH

35

7.8

63

54

.33

56

,54

9(2

9,6

90

)5

5,7

47

(29

,94

2)

46

,42

9.0

0.8

43

0.0

1

Res

po

nd

ents

per

gro

up:

CO

7,

SO

18

9,

EN

53

3,

RE

50

,F

IN¼

12

25,

CH

53

9,

OT

17

4.

*p,

0.0

5.

International Gambling Studies 125

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Relations of gambling motives to gambling problems and gambling involvement

Gambling problems

The three Cooper et al.’s (1992) groups of motives were compared on severity of gambling

problems on the PGSI (Ferris & Wynne, 2001). A Kruskal–Wallis test revealed a

significant difference between the groups of motives, x 2 (2) ¼ 11.92, p ¼ 0.003. Mean

values are presented in Figure 1a. Mann–Whitney U tests revealed that, as predicted, COP

Figure 1. (A) Mean PGSI scores, (B) Mean number of gambling activities, and (C) Mean moneyspent gambling for Cooper et al.’s (1992) gambling motives groups. (*Indicates a significantdifference, p , 0.05)

126 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

had more severe gambling problems than SOC gamblers, U ¼ 607.50, p ¼ 0.001.

However, contrary to predictions, ENH did not have higher PGSI scores than SOC

gamblers, U ¼ 83,974.50, p ¼ 0.181. Finally, COP gamblers had more severe gambling

problems than ENH gamblers, U ¼ 1756.00, p ¼ 0.002.

Lastly, all seven groups of motives were compared on PGSI scores. A Kruskal–Wallis

test revealed a significant difference between groups of motives, x 2 (6) ¼ 106.91,

p ¼ 0.000. Mann–Whitney U tests (see Table 3) revealed that FIN gamblers had

significantly lower PGSI scores when compared to the three Cooper et al.’s (1992) model

groups. However, FIN gamblers had significantly higher PGSI scores than the CHA

motives group. The FIN group did not differ from the REC group. The CHA group had

significantly lower PGSI scores than the other six groups of motives (COP, SOC, ENH,

FIN, REC and OTH). Lastly, the REC group had significantly lower PGSI scores than the

COP group. However, no differences were found on PGSI scores between the REC group

when compared to SOC, ENH, FIN and OTH groups of motives.

Level of gambling involvement

The extent to which levels of gambling involvement (frequency, number of activities

played, and money spent) differ between groups of motives was also explored. The

gambling frequency variable consisted of seven categories ranging from ‘daily’ gambling

to ‘between 1 and 5 times per year’ in the past 12 months. Respondents were asked to rate

their gambling frequency for each activity they participated in. For the purpose of using

Fisher’s exact tests, responses to all gambling activities were collapsed into one

dichotomous variable: ‘gambled at least once a week’ for at least one activity (yes/no). Of

the total 7095 responses, 1531 (21.6%) were at least once a week. The proportion of COP,

SOC and ENH gamblers who gambled at one activity at least once a week was 88.9%,

56.5% and 44.5%. Significant differences were found for COP . ENH ( p ¼ 0.013); SOC

. ENH ( p ¼ 0.001) and a trend was found for COP . SOC ( p ¼ 0.083).

For the number of gambling activities in which the participant had been involved over

the previous 12 months, a Kruskal–Wallis test did not reveal significant differences

between the three Cooper et al.’s (1992) groups of motives, x 2 (2) ¼ 3.77, p ¼ 0.152.

Contrary to predictions, neither COP nor ENH gamblers endorsed a greater number of

gambling activities than SOC gamblers (see Figure 1b). However, when all seven groups

of motives were compared on the number of gambling activities, a statistically significant

difference emerged, x 2 (6) ¼ 307.66, p ¼ 0.000. A series of Mann–Whitney U tests (see

Table 4) showed that the FIN group participated in significantly fewer gambling activities

than the SOC, ENH, REC and CHA groups. CHA gamblers participated in significantly

fewer gambling activities than all of the other motives groups. REC gamblers participated

in significantly more gambling activities than FIN, but fewer than SOC.

Lastly, the amount of money spent on gambling over the past 12 months did

significantly differ between the Cooper et al.’s (1992) groups of motives, x 2 (2) ¼ 31.14,

p ¼ 0.000. It was predicted that COP gamblers would have a higher average rank than the

SOC gamblers; however, despite relatively large differences (see Figure 1c), this was not

the case, U ¼ 676.50, p ¼ 0.503. Similarly, COP gamblers did not differ significantly

from ENH gamblers, U ¼ 1605.00, p ¼ 0.175. Unexpectedly, SOC gamblers spent

significantly more money than ENH, U ¼ 56,079.00, p ¼ 0.000. When all seven groups of

motives were compared on money spent gambling during the past 12 months, a significant

difference between groups of motives emerged, x 2 (6) ¼ 240.59, p ¼ 0.000 (see Table 5).

The CHA gamblers were found to spend less money gambling than the COP, SOC, ENH,

International Gambling Studies 127

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Tab

le3

.M

ann

–W

hit

ney

Ure

sult

sfo

rP

GS

Isc

ore

sb

yse

ven

gro

up

so

fg

amb

lin

gm

oti

ves

Mo

tiv

es’

Gro

up

sM

ean

Ran

kM

(SD

)

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

UP

r

CO

PS

OC

19

0.5

01

29

.49

8.0

0(9

.58

)0

.51

(1.6

7)

60

7.5

0.0

01

*0

.21

CO

PE

NH

50

1.8

93

49

.04

8.0

0(9

.58

)0

.84

(2.6

1)

17

56

.00

.00

2*

0.1

2C

OP

RE

C5

1.8

93

5.5

18

.00

(9.5

8)

1.0

9(4

.11

)1

63

.00

.00

3*

0.3

5C

OP

FIN

12

36

.44

83

0.3

08

.00

(9.5

8)

0.3

8(1

.50

)3

81

2.0

0.0

00

*0

.10

CO

PC

HA

53

4.6

73

49

.63

8.0

0(9

.58

)0

.21

(1.7

7)

14

79

.00

.00

0*

0.2

5C

OP

OT

H1

75

.50

11

9.4

18

.00

(9.5

8)

0.7

5(3

.15

)5

62

.50

.00

0*

0.23

SO

CE

NH

45

8.9

24

78

.15

0.5

1(1

.67

)0

.84

(2.6

1)

83

,97

4.5

0.1

81

0.0

4S

OC

RE

C1

59

.90

15

7.9

30

.51

(1.6

7)

1.0

9(4

.11

)8

12

0.5

0.8

19

0.0

1S

OC

FIN

99

8.0

09

47

.85

0.5

1(1

.67

)0

.38

(1.5

0)

19

8,3

52

.00

.02

6*

0.0

5S

OC

CH

A5

20

.31

45

7.1

20

.51

(1.6

7)

0.2

1(1

.77

)7

6,0

75

.00

.00

0*

0.2

1S

OC

OT

H2

46

.70

24

0.0

30

.51

(1.6

7)

0.7

5(3

.15

)2

8,6

65

.00

.42

70

.04

EN

HR

EC

38

0.6

83

61

.15

0.8

4(2

.61

)1

.09

(4.1

1)

21

,33

0.0

0.3

41

0.0

4E

NH

FIN

12

50

.66

11

41

.94

0.8

4(2

.61

)0

.38

(1.5

0)

51

9,5

78

.00

.00

0*

0.1

1E

NH

CH

A7

53

.26

63

3.9

10

.84

(2.6

1)

0.2

1(1

.77

)1

98

,76

9.5

0.0

00

*0

.25

EN

HO

TH

47

0.8

04

39

.84

0.8

4(2

.61

)0

.75

(3.1

5)

75

,22

2.5

0.0

31

*0

.07

RE

CF

IN8

92

.21

85

9.2

51

.09

(4.1

1)

0.3

8(1

.50

)5

1,7

26

.50

.37

90

.02

RE

CC

HA

42

0.7

73

76

.18

1.0

9(4

.11

)0

.21

(1.7

7)

19

,90

5.0

0.0

00

*0

.14

RE

CO

TH

15

1.0

51

49

.07

1.0

9(4

.11

)0

.75

(3.1

5)

74

71

.50

.79

70

.02

FIN

CH

A1

20

3.4

81

10

7.0

90

.38

(1.5

0)

0.2

1(1

.77

)5

27

,15

5.0

0.0

00

*0

.12

FIN

OT

H9

41

.59

96

5.1

40

.38

(1.5

0)

0.7

5(3

.15

)1

87

,99

9.5

0.3

03

0.0

2

CH

AO

TH

45

1.6

55

00

.77

.21

(1.7

7)

.75

(3.1

5)

72

28

2.5

.00

0*

0.1

7

Res

po

nd

ents

per

gro

up:

CO

9,

SO

25

3,

EN

69

2,

RE

65

,F

IN¼

1,6

55

,C

HA¼

69

4,

OT

23

3.

*p,

0.0

5.

128 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Tab

le4

.M

ann

–W

hit

ney

Ure

sult

sfo

rn

um

ber

of

gam

bli

ng

acti

vit

ies

inth

ep

ast

12

mo

nth

sb

yse

ven

gro

up

so

fg

amb

lin

gm

oti

ves

Mo

tiv

es’

gro

up

sM

ean

ran

kM

ean

(SD

)

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

UP

r

CO

PS

OC

12

4.7

81

31

.74

3.7

8(2

.17

)3

.81

(1.7

4)

10

78

.00

.78

30

.02

CO

PE

NH

36

0.1

13

50

.88

3.7

8(2

.17

)3

.69

(2.1

2)

30

32

.00

.89

00

.01

CO

PR

EC

42

.83

36

.76

3.7

8(2

.17

)3

.08

(1.5

9)

24

4.5

0.4

19

0.0

9C

OP

FIN

11

24

.28

83

0.9

13

.78

(2.1

7)

2.6

0(1

.57

)4

82

1.5

0.0

60

0.0

5C

OP

CH

A5

06

.22

35

0.0

03

.78

(2.1

7)

2.3

4(1

.19

)1

73

5.0

0.0

16

*0

.09

CO

PO

TH

15

5.8

91

20

.17

3.7

8(2

.17

)2

.79

(1.7

1)

73

9.0

0.1

24

0.1

0

SO

CE

NH

50

1.1

54

62

.71

3.8

1(1

.74

)3

.69

(2.1

2)

80

41

6.5

0.0

52

0.0

6S

OC

RE

C1

67

.04

13

0.1

53

.81

(1.7

4)

3.0

8(1

.59

)6

31

4.5

0.0

03

*0

.16

SO

CF

IN1

22

9.8

49

01

.71

3.8

1(1

.74

)2

.60

(1.5

7)

12

1,9

86

.00

.00

0*

0.2

5S

OC

CH

A6

51

.66

40

9.2

33

.81

(1.7

4)

2.3

4(1

.19

)4

2,8

42

.00

.00

0*

0.4

0S

OC

OT

H2

85

.91

19

7.4

53

.81

(1.7

4)

2.7

9(1

.71

)1

8,7

45

.50

.00

0*

0.3

2

EN

HR

EC

38

3.5

53

30

.52

3.6

9(2

.12

)3

.08

(1.5

9)

19

,33

8.5

0.0

58

0.0

7E

NH

FIN

14

38

.77

10

63

.29

3.6

9(2

.12

)2

.60

(1.5

7)

38

9,4

09

.50

.00

0*

0.2

6E

NH

CH

A8

32

.01

55

5.3

93

.69

(2.1

2)

2.3

4(1

.19

)1

44

,27

3.0

0.0

00

*0

.35

EN

HO

TH

49

3.8

73

71

.32

3.6

9(2

.12

)2

.79

(1.7

1)

59

,25

6.0

0.0

00

*0

.20

RE

CF

IN1

01

7.4

58

54

.34

3.0

8(1

.59

)2

.60

(1.5

7)

43

,58

6.0

0.0

08

*0

.06

RE

CC

HA

47

2.8

63

71

.30

3.0

8(1

.59

)2

.34

(1.1

9)

16

,51

9.0

0.0

00

*0

.14

RE

CO

TH

16

5.0

01

45

.18

3.0

8(1

.59

)2

.79

(1.7

1)

65

65

.00

.09

40

.10

FIN

CH

A1

19

4.7

41

12

7.9

22

.60

(1.5

7)

2.3

4(1

.19

)5

41

,61

3.5

0.0

25

*0

.05

FIN

OT

H9

37

.21

99

6.2

92

.60

(1.5

7)

2.7

9(1

.71

)1

80

,73

9.5

0.1

11

0.0

4

CH

AO

TH

44

9.1

85

08

.16

2.3

4(1

.19

)2

.79

(1.7

1)

70

,56

2.5

0.0

02

*0

.10

Res

po

nd

ents

per

gro

up:

CO

9,

SO

25

3,

EN

69

2,

RE

65

,F

IN¼

1,6

55

,C

HA¼

69

4,

OT

23

3.

*p,

0.0

5.

International Gambling Studies 129

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Tab

le5

.M

ann

–W

hit

ney

Ure

sult

sfo

rm

on

eysp

ent

gam

bli

ng

(in

do

llar

s)in

the

pas

t1

2m

on

ths

by

sev

eng

rou

ps

of

gam

bli

ng

mo

tiv

es

Mo

tiv

es’

gro

up

sM

ean

ran

kM

ean

(SD

)

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

Gro

up

1G

rou

p2

UP

r

CO

PS

OC

13

4.3

61

16

.98

12

64

.57

(17

25

.9)

80

4.3

6(3

34

0.7

)6

76

.50

.50

30

.04

CO

PE

NH

42

7.7

13

29

.46

12

64

.57

(17

25

.9)

70

6.4

9(3

21

1.1

)1

60

5.0

0.1

75

0.0

5C

OP

RE

C4

0.6

43

1.5

01

26

4.5

7(1

72

5.9

)2

98

.74

(48

1.8

)1

42

.50

.22

00

.15

CO

PF

IN1

06

8.0

77

90

.27

12

64

.57

(17

25

.9)

55

1.1

6(3

43

0.5

)3

57

6.5

0.1

08

0.0

4C

OP

CH

A5

07

.79

33

1.1

41

26

4.5

7(1

72

5.9

)1

35

.99

(51

4.9

)1

07

9.5

0.0

15

*0

.09

CO

PO

TH

16

6.9

31

14

.41

12

64

.57

(17

25

.9)

60

6.1

6(2

92

3.2

)4

27

.50

.04

1*

0.1

3

SO

CE

NH

51

9.9

64

12

.88

80

4.3

6(3

34

0.7

)7

06

.49

(32

11

.1)

56

,07

9.0

0.0

00

*0

.18

SO

CR

EC

14

7.4

31

22

.86

80

4.3

6(3

34

0.7

)2

98

.74

(48

1.8

)5

35

0.0

0.0

43

*0

.12

SO

CF

IN1

12

9.3

78

68

.66

80

4.3

6(3

34

0.7

)5

51

.16

(34

30

.5)

12

,70

35

.50

.00

0*

0.1

7S

OC

CH

A6

22

.20

38

1.1

88

04

.36

(33

40

.7)

13

5.9

9(5

14

.9)

34

05

.50

.00

0*

0.4

1S

OC

OT

H2

77

.68

17

3.6

38

04

.36

(33

40

.7)

60

6.1

6(2

92

3.2

)1

3,6

92

.50

.00

0*

0.4

0

EN

HR

EC

35

2.9

33

84

.99

70

6.4

9(3

21

1.1

)2

98

.74

(48

1.8

)1

69

,29

.50

.25

70

.04

EN

HF

IN1

13

9.4

31

10

4.1

67

06

.49

(32

11

.1)

55

1.1

6(3

43

0.5

)4

97

,95

9.5

0.2

39

0.0

3E

NH

CH

A7

59

.82

55

2.9

77

06

.49

(32

11

.1)

13

5.9

9(5

14

.9)

14

7,0

45

.00

.00

0*

0.2

7E

NH

OT

H4

67

.17

35

6.8

77

06

.49

(32

11

.1)

60

6.1

6(2

92

3.2

)5

4,7

38

.50

.00

0*

0.1

9

RE

CF

IN9

17

.18

81

2.8

62

98

.74

(48

1.8

)5

51

.16

(34

30

.5)

39

,14

8.5

0.1

00

0.0

4R

EC

CH

A4

94

.54

34

6.0

92

98

.74

(48

1.8

)1

35

.99

(51

4.9

)1

0,9

13

.50

.00

0*

0.2

0R

EC

OT

H1

78

.55

13

1.4

42

98

.74

(48

1.8

)6

06

.16

(29

23

.2)

42

43

.50

.00

0*

0.2

3

FIN

CH

A1

21

5.5

88

81

.05

55

1.1

6(3

43

0.5

)1

35

.99

(51

4.9

)3

62

,91

7.5

0.0

00

*0

.24

FIN

OT

H9

26

.14

71

6.1

85

51

.16

(34

30

.5)

60

6.1

6(2

92

3.2

)1

35

,22

3.5

0.0

00

*0

.13

CH

AO

TH

43

6.6

84

55

.65

13

5.9

9(5

14

.9)

60

6.1

6(2

92

3.2

)7

0,5

27

.50

.33

60

.03

Res

po

nd

ents

per

gro

up:

CO

7,

SO

22

7,

EN

65

3,

RE

57

,F

IN¼

1,5

75

,C

HA¼

65

8,

OT

22

4.

*p,

0.0

5.

130 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

REC and FIN groups. No differences were found between the CHA group and OTH

groups. The OTH group was found to spend less than COP, SOC and ENH gamblers, but

spent more than the REC and FIN groups. Similarly, SOC gamblers also spent more

gambling than REC and FIN gamblers.

Relations of gambling motives to types of gambling activities

For associations of groups of motives with choice of gambling activities, we compared

Cooper et al.’s (1992) groups of motives on involvement with several forms of gambling.

The activities included in the original surveys were combined into six broad categories:

cards, races, sports, bingo, lottery and slots/VLTs. Of these, three activities that involve

some level of skill (cards and casino games, races, and sports) were considered to be

‘active’ or exciting activities, while three activities that are essentially games of chance

(bingo, lottery and slots/VLTs) could be seen as more ‘passive’ or absorbing/distracting

gambling activities (Bonnaire et al., 2006). First, we used a series of Fisher’s exact tests

to investigate whether certain motives for gambling were associated with the ‘active’

gambling activities. A larger proportion of SOC gamblers (39.9%) compared to ENH

(28.3%) gamblers reported gambling on cards and casino games ( p ¼ 0.001). Also, ENH

gamblers (6.1%) were more likely to have participated in other ‘exciting’ forms of

gambling such as races than SOC gamblers (2.0%) ( p ¼ 0.010). No other between-group

differences were found.

Next, we used a series of Fisher’s exact tests to examine possible associations between

certain motives for gambling and ‘passive’ gambling activities (i.e. bingo, lottery and

slots/VLTs). As predicted, a larger proportion of SOC (33.6%) compared to ENH (19.2%)

gamblers reported playing bingo in the past 12 months ( p ¼ 0.000). There was also a trend

for increased bingo participation among COP (44.4%) compared to ENH (19.2%) gamblers

( p ¼ 0.078), but no differences were noted between COP and SOC gamblers ( p ¼ 0.494).

For past 12-month lottery participation, no significant differences were found between the

groups of motives. However, the majority of individuals in the COP (88.9%), ENH (91.3%)

and SOC groups (92.1%) reported playing the lottery, possibly placing ceiling effects on

the analyses. Finally, significant differences were found among the groups of motives for

slots/VLTs. As hypothesised, a larger proportion of COP (66.7%) compared to SOC

(29.2%) gamblers reported playing slots/VLTs ( p ¼ 0.025). A trend was also found for

greater slots/VLTs participation among COP (66.7%) compared to ENH (35.8%) gamblers

( p ¼ 0.079). Lastly, there was also a trend for greater participation by ENH gamblers

(35.8%) compared to SOC gamblers (29.2%) ( p ¼ 0.063) for slots/VLTs participation.

Relations of gambling motives to suicidal thoughts/attempts and treatment forstress-related illness

Responses to questions surrounding health issues that participants attributed to their

gambling behaviour were compared across Cooper et al.’s (1992) groups of motives.

Fisher’s exact tests revealed trends for increased suicide thoughts/attempts among COP

when compared to both SOC ( p ¼ 0.068) and ENH ( p ¼ 0.102) gamblers. As predicted, a

larger proportion of COP gamblers (20.0%), in comparison to SOC gamblers (0.9%) and

ENH gamblers (1.9%), reported suicidal thoughts or attempts as a result of their gambling.

No significant differences were found between SOC and ENH gamblers ( p ¼ 0.498).

Analyses also revealed a trend for higher endorsement of seeking a doctor’s help for stress-

related illnesses in the previous year among COP (40.0%) compared to SOC gamblers

International Gambling Studies 131

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

(10.1%), ( p ¼ 0.091). A trend was also found for higher endorsement among ENH

(15.1%) compared to SOC gamblers (10.1%), ( p ¼ 0.084). No differences were found

between COP and ENH gamblers ( p ¼ 0.171).

Relations of gambling motives to intoxication while gambling

We investigated whether gambling while drunk or high differed across Cooper et al.’s

(1992) motives groups using Fisher’s exact tests. It was initially predicted that ENH

gamblers would be more likely to gamble while intoxicated than either SOC or COP

gamblers; however, this was not the case. No significant differences were found between

COP (20%) vs SOC (11.8%), ( p ¼ 0.475); nor COP (20%) vs ENH (22.1%), ( p ¼ 1.000).

The expected significant difference was found, however, between ENH (22.1%) and SOC

(11.8%) gamblers ( p ¼ 0.002).

Discussion

The purpose of this study was to examine the utility of classifying gamblers into groups of

gambling motives based on their own initial self-generated responses to an open-ended

interview question about their primary reason(s) for gambling. This investigation was

completed by analysing responses within an integrated dataset comprised of gambling

epidemiological surveys conducted in two Canadian provinces. Response categories for

reasons for gambling were placed in the context of Cooper and colleagues’ (1992)

motivational model for alcohol use, adapted for gambling behaviour. Although reasons for

gambling that reflect primarily COP motives, primarily ENH motives, and primarily SOC

motives were identified, they represented only 26.5% of the total 3601 reasons provided by

gamblers. ENH motives comprised 19.2% of the total reasons for gambling, SOC motives

comprised 7.0% of reasons, and COP motives comprised less that 1.0% of the total

motives. Despite the use of non-parametric statistics for our analyses, the statistical power

in comparisons involving the COP gamblers is low and may lead to spurious conclusions

in some instances. Therefore, the reader should use discretion in interpreting the results of

direct comparisons which involve the COP group.

Prior to analysing the gambling criterion variables of interest, the groups of motives

were compared across the two provinces and according to demographic characteristics.

In both surveys (i.e. Alberta and Newfoundland and Labrador) COP gamblers were

quite rare (. 1%). Newfoundland did, however, have a greater representation of SOC

gamblers, while Alberta was found to have a greater proportion of ENH gamblers. These

provincial variations may be a result of cultural differences surrounding gambling or could

potentially be due to differences in access to gambling (e.g. less availability of certain

forms of gambling in rural communities). Differences were also found between the

motives groups on age and income. ENH and CHA gamblers in the sample were found to

be younger than gamblers who gambled for other reasons, while CHA in particular tended

to be older. Lastly, the REC gamblers reported the lowest average household income while

the CHA gamblers reported the highest.

Although relatively few motives were categorised according to Cooper et al.’s (1992)

model, several of our a priori hypotheses were supported among the COP, ENH and

SOC motivated-gamblers identified in the dataset, while others were not. First, it was

hypothesised that: (1) COP motives would be endorsed to a greater extent by women and

ENH motives would be endorsed more often by men. This hypothesis was not supported

as COP and ENH gamblers were equally endorsed by men and women. SOC motives,

132 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

however, were more frequently endorsed by women consistent with some previous

research (e.g. Stewart & Zack, 2008). It was also hypothesised that: (2) COP and ENH

gamblers would display higher gambling severity scores than SOC gamblers. We found

partial support as COP gamblers, but not ENH gamblers, had greater problem-gambling

severity scores as measured by the CPGI (Ferris & Wynne, 2001). It was also predicted

that: (3) the three motives groups would differ in their level of involvement and

choice of gambling activities. A trend was found with the COP group gambling on a more

frequent basis than SOC gamblers and significantly more than ENH gamblers. However,

unexpectedly, COP gamblers did not spend more money gambling than the other groups.

COP gamblers were also more likely than SOC and ENH to endorse passive (and

theoretically highly risky) activities such as slots/VLTs. However, with regards to ENH

gamblers, our hypothesis was only partially supported as the ENH group endorsed race

track betting more than SOC gamblers (6.1% vs 2.0%) but not cards and casino games

(28.3% vs 39.9%) or sports betting (18.1% vs 22.5%). Given the social nature of certain

forms of sports betting (e.g. betting on games of skill), this latter finding should not

be completely unexpected. It was also hypothesised that: (4) COP motives would be

associated with greater psychopathology. This prediction was partially supported as

relative to SOC gamblers, COP gamblers showed trends for more suicidal

thoughts/attempts and for greater help-seeking for stress-related illness. Finally, with

regards to alcohol/drug use, it was predicted that: (5) ENH gamblers would be more likely

than the other groups to gamble when drunk/high. Again, this hypothesis was only partially

supported as the ENH gamblers were more likely than SOC gamblers to report they

gambled while intoxicated, but no differences were found between the ENH and COP

gamblers. Overall, these findings are generally consistent with what has been seen using

other methods for identifying COP gamblers (e.g. Stewart et al., 2008; Stewart & Zack,

2008). Even though COP gamblers were quite rare in this sample when using the open-

ended methodology, those who were identified as COP gamblers were more likely to be

problem gamblers, to engage in higher risk gambling activities and to display certain

emotional vulnerability characteristics (e.g. elevated suicidality).

The categorisation of groups of motives in the current study revealed a considerable

number of reasons for gambling that failed to fit within the framework of Cooper and

colleagues’ (1992) motivational model adapted for gambling. Although these motives

were not specifically addressed in our a priori hypotheses, their potential importance

in furthering our understanding of motivations for gambling warranted investigation.

Specifically, we conducted post-hoc analyses comparing the FIN, REC, CHA and OTH

groups on demographic characteristics and across measures of gambling severity and

gambling involvement. Several notable findings were revealed in our analyses. First, the

FIN, REC and CHA gamblers tended to be older than those categorised into Cooper et al.’s

motivational groups (with the exception of COP). In particular, distinct demographic

differences were associated with the CHA group as they tended to be older, more likely

to be female, and to have higher incomes on average than most of the other types of

gamblers. In terms of gambling severity scores, several interesting differences emerged

among the groups. For instance, FIN-motivated gamblers had significantly lower scores

than most other gamblers. Differences between the CHA group and the other gamblers

were especially pronounced as this group had significantly lower scores than all others.

Lastly, with regards to overall involvement in gambling, CHA gamblers participated in

fewer gambling activities and spent less money gambling than most other groups. These

results may suggest that CHA is the lowest risk motivation for gambling; alternatively

these findings could also indicate that people who are highly sensitive to external

International Gambling Studies 133

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

judgement tend to portray themselves in a more socially positive light (e.g. by endorsing

fewer problem gambling items and claiming to gamble for CHA reasons). Further

exploration of the influence of FIN and CHA motives on gambling behaviour is needed.

While distinct differences between COP, ENH and SOC gamblers across criterion

variables were found, the large proportion of respondents who could not be classified

according to Cooper et al.’s (1992) model raises concern about the comprehensiveness of

the model for classifying gambling motives. Undoubtedly, the categorisation of self-

generated open-ended responses revealed relatively few motives that could be concretely

placed within the model’s framework. This finding may suggest that: (1) Cooper et al.’s

(1992) motivational model adapted for gambling, in its current form, does not adequately

capture the full spectrum of motives applicable to gambling; or (2) using top-of-mind

responses may be less effective for identifying gambling motives than forced-choice

questionnaire methods such as the GMQ (Stewart & Zack, 2008); each of these

possibilities is discussed in turn.

It is conceivable that Cooper et al.’s (1992) model does not adequately encapsulate all of

the theoretically relevant motives associated with gambling. While there is substantial

evidence highlighting the importance of COP, ENH and SOC motives in the drinking

literature, the present results suggest that other motives associated specifically with

gambling are more prevalent. Indeed, the majority of motives (73.5%) existed outside of the

model, with FIN (46.0%) and CHA (19.3%) being the two most frequent reasons endorsed

by respondents. It is perhaps unsurprising that a substantial portion of individuals would

self-generate one of these motives as their main reason for gambling. For example, much of

the gambling that takes place in Canada is charity-related. Endorsement of CHA motives

for gambling may be much lower in other jurisdictions where gambling is privately

operated (i.e. not directed towards charitable purposes). However, despite its inability to

capture all relevant motives in the current study, outright rejection of the utility of Cooper

et al.’s model in identifying gambling motives may also be unwarranted. For instance,

empirical support from previous investigations using questionnaire based approaches

(e.g. Stewart & Zack, 2008; Stewart et al., 2008; Stuart et al., 2008) provides strong

evidence for the inclusion of COP, ENH and SOC in the discussion of gambling motives. In

addition, the results of our current investigation in many ways also support the predictive

validity of Cooper et al.’s model. In most cases, the predicted differences between the

Cooper et al.’s motives groups on the available criterion variables were found. For example,

although the COP group was small and thus tests involving this group were underpowered,

respondents who endorsed COP motives were more likely to report problems with gambling

and other emotional vulnerabilities. Our findings are in line with expectations based on

previous work as well as the theory upon which the model is based. Rather, we argue that the

current 3-factor model may need to be expanded to reflect the broader range of motives

potentially associated with gambling, many of which may not be applicable to drinking.

Alternatively, the distribution of motives in the current study could be due to

limitations associated with using open-ended top-of-mind responses to categorise motives.

The small proportion of COP gamblers identified may be indicative of a flaw in classifying

gambling motives based on initial responses to an open-ended query about motives. For

instance, other questionnaires that employ continuous scales allow for a more nuanced

assessment of reasons for gambling, whereas the present method limits responses to a

single top-of-mind motive. Accessibility to this top-of-mind response may be less a

function of one’s true motivation, and more a function of social norms or social

desirability. A large proportion of the motives identified in this study (e.g. FIN, CHA and

REC) could be considered socially acceptable reasons to gamble, whereas COP motives

134 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

would not. It is conceivable that socially appropriate reasons could be offered as a first

response even if one’s own true primary reason is COP-related. However, the open-ended

method of classifying reasons for gambling did identify potential motives that have

received little attention in the literature to date. In one exception, Neighbors et al. (2002)

employed a mixed qualitative and quantitative methodology to identify reasons for

gambling in a sample of undergraduate gamblers. Using an open-ended question they

identified 16 individual motives, several of which were found in the present study. For

instance, motives that Neighbors et al. termed: ‘enjoyment/fun’, ‘excitement’, ‘occupy

time/boredom’ and ‘challenge’ describe ENH motives. Similarly, they identified both

‘coping’ and ‘social’ reasons for gambling. Also, ‘winning’ and ‘luck’ overlap with our

FIN motives category and their ‘skill’ category resembles our REC motives category.

Lastly, ‘conformity’ reasons for gambling were identified in their student sample,

however; conformity motives did not appear to emerge among our adult sample, offering

further support to the suggestion conformity motives may be more relevant for youth

addictive behaviours (Cooper, 1994). Interestingly, in many respects the proportion of

motives identified by Neighbors et al. resembles that of the current study. For instance,

both studies identified FIN motives as being the most reported and COP motives as being

among the least cited reasons for gambling. Furthermore, a large proportion of gamblers in

both studies gambled for ENH or SOC reasons. While the findings of Neighbors et al.

corroborate the results of the current investigation, questions remain regarding the role of

social desirability in producing these top-of-mind responses.

The current study suffers from a problem that plagues all secondary analyses – the

survey was not originally designed to answer the questions proposed in this investigation.

It would have been preferable to have information on each gambler’s activity of choice,

rather than using a composite score of gambling involvement to determine this in the

Alberta survey. It would also have been preferable to have included the 63 IGS items

(Littman-Sharp et al., 2009) or the GMQ (Stewart & Zack, 2008) in the surveys so that

the self-generated motives could be directly compared to previous sub-typing schemes

(e.g. Stuart et al., 2008). A further limitation is that a combined dataset with responses

from two separate surveys was used. While there are noted strengths associated with using

two different datasets (e.g. greater external validity and geographic diversity) there are

also weaknesses. For instance, sometimes questions were not asked in the same order, and

at times wording differed slightly across the surveys. Nonetheless, the items chosen for

analysis were either identical across the two surveys (in the majority of cases) or were very

similar, helping to minimise this limitation. Lastly, it may have been preferable to examine

gambling motives for each individual gambling activity. Indeed, recent work has begun to

focus on the motives behind specific forms of gambling (e.g. measuring motives among

electronic gaming machine gamblers; Thomas, Allen & Phillips, 2009). Unfortunately, the

type of data necessary to conduct this analysis was only available in one of the two datasets

and thus is not presented herein.

To our knowledge, this study is the first to categorise self-generated reasons for

gambling into the context of the Cooper et al. (1992) motivational model for alcohol use

adapted for gambling. This study may have important implications for further gambling

motivation classification efforts as well as significance for the treatment and prevention of

problem gambling. Overall, it was found that relatively few self-generated reasons could

successfully be categorised within Cooper et al.’s model. The majority of motives that

could not be placed into the model were identified as being financial and charitable in

nature. Although most of the a priori predictions surrounding the three Cooper et al.’s

motives groups were supported, the proportionally few motives that could be included are

International Gambling Studies 135

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

a cause for concern. Future work could focus on exploring the utility of Cooper et al.’s

model adapted for gambling by expanding its framework to include other potentially

relevant motives for gambling including those identified in this report. In addition,

potential issues of concern surround the use of self-generated methods for identifying

gambling motives. Despite previous work suggesting that top-of-mind responses are the

most readily accessible (e.g. Stacy et al., 1994), the vast majority of responses offered

could be labelled as being more socially appropriate/acceptable. The usefulness of self-

generated reasons for gambling could be further evaluated through a direct comparison of

this method with other psychometrically validated measures such as the IGS (Littman-

Sharp et al., 2009) or GMQ (Stewart & Zack, 2008).

Acknowledgements

This research was supported by a generous grant funded by the Ontario Problem Gambling ResearchCentre. Daniel S. McGrath was supported through doctoral awards from the Ontario ProblemGambling Research Centre, the Nova Scotia Health Research Foundation, and the Nova ScotiaGaming Foundation during the completion of this work. Sherry H. Stewart was supported through aKillam Research Professorship from the Dalhousie University Faculty of Science. The authors wouldlike to acknowledge Pamela Collins, Adrienne Girling, and Lyndsay Bozec for their assistance withdata formatting and coding.

Notes on contributors

Daniel S. McGrath MSc is a senior graduate student in the Department of Psychology at DalhousieUniversity. His main research area focuses on examining the effects of tobacco smoke and nicotineon gambling behaviour. Much of his work is currently funded by the Ontario Problem GamblingResearch Centre (OPGRC), the Nova Scotia Health Research Foundation (NSHRF), and the NovaScotia Gaming Foundation (NSGF).

Sherry H. Stewart is a Killam Research Professor in the Departments of Psychiatry and Psychologyat Dalhousie University. She is currently the coordinator of the Doctoral Training Program inClinical Psychology at Dalhousie. She is presently the Associate Editor of the international journalsCognitive Behaviour Therapy and Current Drug Abuse and was recently appointed by former HealthMinister Tony Clements to the Board of Directors of the Canadian Centre on Substance Abuse.

Raymond M. Klein is a University Research Professor at Dalhousie University, chairperson of theDepartment of Psychology and past president of the Canadian Society for Brain, Behavior andCognitive Science. Professor Klein is a cognitive scientist who, since his first sabbatical at BellTelephone Laboratories, has a palpable interest in applying the methods of experimental psychologyto help solve real-world problems. His basic and applied research strategies are strongly influencedby those of Donald Hebb, Donald Broadbent and Michael Posner.

Sean P. Barrett is an Assistant Professor in the Department of Psychology at Dalhousie Universityand his research interests span a variety of areas of addiction including the co-morbidity of variousaddictive behaviours (e.g. drinking and smoking with gambling) and the role of personality factorssuch as sensation seeking in risk for addictive disorders.

References

American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders,fourth edition – revised (DSM-IV-R). Washington, DC: APA.

Beaudoin, C.M., & Cox, B.J. (1999). Characteristics of problem gambling in a Canadian context:A preliminary study using a DSM-IV-based questionnaire. The Canadian Journal of Psychiatry,44, 483–487. http://publications.cpa-apc.org/browse/sections/0

Blaszczynski, A., & Nower, L. (2002). A pathways model of problem and pathological gambling.Addiction, 97(5), 487–499. doi:10.1046/j.1360-0443.2002.00015.x.

136 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Bonnaire, C., Bungener, C., & Varescon, I. (2006). Pathological gambling and sensation seeking –how do gamblers playing games of chance in cafes differ from those who bet on horses at theracetrack? Addiction Research & Theory, 14, 619–629, doi: 10.1080/16066350600964296.

Burger, T.D., Dahlgren, D., & MacDonald, C.D. (2006). College students and gambling: Anexamination of gender differences in motivation for participation. College Student Journal, 40,704–714. http://www.projectinnovation.biz/csj_2006.html

Conrod, P.J., Pihl, R.O., Stewart, S.H., & Dongier, M. (2000). Validation of a system of classifyingfemale substance abusers on the basis of personality and motivational risk factors for substanceabuse. Psychology of Addictive Behaviors, 14, 243–256. doi:10.1037/0893-164X.14.3.243.

Cooper, M.L. (1994). Motivations for alcohol use among adolescents: Development and validationof a four-factor model. Psychological Assessment, 6, 117–128. doi:10.1037/1040-3590.6.2.117.

Cooper, M.L., Russell, M., Skinner, J.B., & Windle, M. (1992). Development and validation ofa three-dimensional measure of drinking motives. Psychological Assessment, 4, 123–132.doi:10.1037/1040-3590.4.2.123.

Cox, W.M., & Klinger, E. (1988). A motivational model of alcohol use. Journal of AbnormalPsychology, 97, 168–180. doi:10.1037/0021-843X.97.2.168.

Doiron, J.P., & Nicki, R.M. (2007). Prevention of pathological gambling: A randomized controlledtrial. Cognitive Behaviour Therapy, 36, 74–84. doi:10.1080/16506070601092966.

Ellery, M., Stewart, S.H., & Loba, P. (2005). Alcohol’s effects on video lottery terminal (VLT) playamong probable pathological and non-pathological gamblers. Journal of Gambling Studies, 21,299–324. doi:10.1007/s10899-005-3101-0.

Ferris, J., & Wynne, H. (2001). Canadian problem gambling index: Final report. Toronto, ON:Centre for Addiction and Mental Health.

Focal Research. (1998). Nova Scotia video lottery players’ survey 1997/98. Halifax, NS: NovaScotia Department of Health, Problem Gambling Services.

Hickey, J., Haertzen, C., & Henningfield, J. (1986). Simulation of gambling responses on theAddiction Research Center Inventory. Addictive Behaviors, 11, 345–349. doi:10.1016/0306-4603(86)90062-6.

Lesieur, H.R., & Blume, S.B. (1987). The South Oaks Gambling Screen (SOGS): A new instrumentfor the identification of pathological gamblers. The American Journal of Psychiatry, 144(9),1184–1188. http://ajp.psychiatryonline.org/

Littman-Sharp, N., Turner, N., & Toneatto, T. (2009). Inventory of gambling situations (IGS):User’s guide. Toronto, ON: Centre for Addiction and Mental Health.

Market Quest Research Group Inc. (2005). 2005 Newfoundland and Labrador gambling prevalencestudy. St. John’s, NL: Department of Health and Community Services, Government ofNewfoundland and Labrador.

Neighbors, C., Lostutter, T.W., Cronce, J.M., & Larimer, M.E. (2002). Exploring college studentgambling motivation. Journal of Gambling Studies, 18, 361–370. doi:10.1023/A:1021065116500.

Pantalon, M.V., Maciejewski, P.K., Desai, R.A., & Potenza, M.N. (2008). Excitement-seekinggambling in a nationally representative sample of recreational gamblers. Journal of GamblingStudies, 24, 63–78. doi:10.1007/s10899-007-9075-3.

Smith, G.J., & Wynne, H.J. (2002). Measuring gambling and problem gambling in Albertausing the Canadian problem gambling index. Edmonton, AB: Alberta Gambling ResearchInstitute.

Stacy, A.W., Leigh, B.C., & Weingardt, K.R. (1994). Memory accessibility and association ofalcohol use and its positive outcomes. Experimental and Clinical Psychopharmacology, 2,269–282. doi:10.1037/1064-1297.2.3.269.

Stewart, S.H. (2009, April). Sub-typing gamblers on the basis of affective motivations forgambling. Paper presented to Nova Scotia Department of Health Promotion and Protection.Halifax, NS.

Stewart, S.H., & Zack, M. (2008). Development and psychometric evaluation of a three-dimensionalgambling motives questionnaire. Addiction, 103, 1110–1117. doi:10.1111/j.1360-0443.2008.02235.x.

Stewart, S.H., Zack, M., Collins, P., Klein, R.M., & Fragopoulos, F. (2008). Subtyping pathologicalgamblers on the basis of affective motivations for gambling: Relations to gambling problems,drinking problems, and affective motivations for drinking. Psychology of Addictive Behaviors,22, 257–268. doi: 10.1037/0893-164X.22.2.257.

International Gambling Studies 137

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013

Stuart, S., Stewart, S.H., Wall, A.M., & Katz, J. (2008). Gambling among university undergraduatestudents: An investigation of gambler subtypes varying in affective motivations for gambling.In M.J. Esposito (Ed.), Psychology of gambling (pp. 83–110). Hauppauge, NY: NovaBiomedical Books.

Thomas, A.C., Allen, F.C., & Phillips, J. (2009). Electronic gaming machine gambling: Measuringmotivation. Journal of Gambling Studies, 25, 343–355. doi:10.1007/s10899-009-9133-0.

Wicki, M., Gmel, G., Kuntsche, E., Rehm, J., & Grichting, E. (2006). Is alcopop consumption inSwitzerland associated with riskier drinking patterns and more alcohol-related problems?Addiction, 101, 522–533. doi:10.1111/j.1360-0443.2006.01368.x.

Wiers, R., Stacy, A., Ames, S., Noll, J., Sayette, M., Zack, M., et al. (2010). Implicit and explicitalcohol-related cognitions. Alcoholism: Clinical and Experimental Research, 26(1), 129–137.doi:10.1097/00000374-200201000-00018.

138 Daniel S. McGrath et al.

Dow

nloa

ded

by [

Dal

hous

ie U

nive

rsity

] at

10:

09 1

8 Ju

ly 2

013