1

Electronic Gaming Machine Gambling: Measuring Motivation

Anna C. Thomas, Felicity C. Allen and James Phillips

Journal of Gambling Studies (2009) Vol 25, pp343–355 DOI 10.1007/s10899-009-9133-0 This is an author-created version. The original publication is available at www.springerlink.com with subscription 1 School of Psychology, Psychiatry and Psychological Medicine, Monash University, Caulfield Campus, Caulfield East, Victoria 3145, Australia 2 Faculty of Life and Social Sciences, Swinburne University of Technology, Mail H31, PO Box 218, Hawthorn, Victoria, 3122, Australia email Anna Thomas: tisanna2@gmail.com

2

Abstract Electronic gambling machines (EGMs) are known to be a particularly risky form of gambling (Petry, 2003). It is vital that researchers and clinicians are aware of factors which could lead to people having problems with this form. Gambling motivation is one such factor. This study developed a measure of EGM gambling motivations based on the results of qualitative research conducted with EGM problem gamblers and experienced counsellors (Thomas, Sullivan & Allen, 2009). A community based sample of 232 females (M=29.60 years of age, SD=15.41 years) and 123 males (M=29.64 years of age, SD=12.29 years) participated. Exploratory factor analysis extracted three motivational factors indicating people gambled on EGMs to escape, for its accessibility and for the social environment. Gambling to escape and for its accessibility had substantial positive correlations with frequency of EGM gambling and gambling problems. Social environment correlated less well with these indicators of excessive gambling. Correlations between factors suggested the accessible, social experience offered by EGM venues increases their appeal as a means of escape. The new subscales were internally consistent and demonstrated good evidence of validity. This new measure will facilitate future investigations into the relationships between gambling motivations, other aetiological factors and EGM problem gambling. Introduction EGMs, commonly referred to as slot machines or poker machines, currently generate the highest gaming revenues but are also demonstrably the most damaging form of gambling (e.g., Dickerson, 2004; Petry, 2003), such that legislators seek to limit the numbers or control access to these machines. As the gaming lobby contemplates further technological innovation (e.g. internet, mobile phones, interactive TV) that can enhance personal access to gambling (Griffiths, 2007; Griffiths & Parke, 2002), it is important to understand the factors that currently motivate people to visit gaming venues to play on EGMs. A previous study used grounded theory methodology to investigate the reasons EGM problem gamblers offered for visiting gaming venues (Thomas et al., 2009). The present paper used this qualitative data on self reported reasons to develop motivation scales to determine which motivations were associated with EGM gambling behaviours. A better understanding of people's motivations when visiting gaming venues may offer insights as to why this form is more closely associated with gambling problems than others. Motivation theory has been used to explain why people engage in, and persist at, gambling (Clarke, 2004; McBain & Ohtsuka, 2001). Problem gamblers are more likely to say they gamble for excitement, as a way of escaping from problems or tension, for the challenge and due to boredom/apathy compared to non-problem gamblers (Clarke, 2008; Clarke et al., 2007; Griffiths, 1990; Platz & Miller, 2001; Schrans Schellinck, & Walsh; 2001; Volberg, 2003; Wood, Gupta, Derevensky, & Griffiths, 2004). Some studies have also found problem gamblers play to win (Clarke 2008; Wood et al.) or for the social interaction (Clarke 2008; Clarke et al.; Platz & Miller; Schrans et al.; Wood et al.) more than non-problem gamblers. The evidence for this is less consistent, however, with other studies suggesting problem gamblers are similarly or less likely than others to gamble for

3

monetary (Clarke et al.; Griffiths; Schrans et al.) or social reasons (Griffiths; Lee, Chae, Lee, & Kim, 2007; Volberg). The study of gambling motivation must also consider the potential for different motivations across gambling form. Gambling problems are often restricted to a single form (Petry, 2003), something which may be related to gambling motivation. People motivated by excitement or arousal may be drawn to stimulating games, while those wanting the challenge of beating others will play games of skill. Those wanting to escape problems and negative emotions will be drawn to games providing an effective cognitive distraction (Coventry & Constable, 1999; Sharpe, 2002; Slowo, 1997). Knowing the reasons people gamble may help explain gambling problems on different forms. Two of the authors recently developed a model of EGM problem gambling using grounded theory and qualitative analysis (Thomas, et al., 2009). They found that gambling motivation was central to explaining the transition to problem gambling. Gamblers exposed to situational or emotional stressors were motivated to gamble as a temporary escape from these problems. EGM machines facilitated a cognitive escape and provided something to do, while venues provided an oasis from problems and demands, a source of company for the lonely, were inviting and welcoming and were highly accessible. Reliance on EGM gambling to manage problems was hypothesised to lead to more frequent and problematic gambling. Consistent evidence of a relationship between gambling problems and cognitive escape (Biddle, Hawthorne, Forbes, & Coman, 2005; Getty, Watson, & Frisch, 2000; New Focus Research, 2003; Thomas & Moore, 2003) has led to this factor being given a central platform in various explanatory models (Blaszczynski & Nower, 2002; Ricketts & Macaskill, 2003). Thomas et al. (2009) also found EGM problem gamblers were motivated to continue gambling because of the social environment, the physical retreat that venues provided and their high temporal and geographic accessibility. These broader conceptualisations of motivation and their relationship to gambling problems are not as well understood. Accessibility to gambling, for example, appears in models of problem gambling (e.g., Blaszczynski & Nower, 2002) but is generally hypothesised to have only an indirect impact, via its influence on the initial introduction to gambling. Thomas et al. suggested that the accessibility provided by the long opening hours and numerous venues in Australia plays a major part in influencing continued and excessive EGM gambling. Further, in line with Thomas et al., gambling as a way of connecting socially has been related to gambling problems in some previous studies (Clarke 2008; Clarke et al., 2007; Platz & Miller, 2001; Schrans et al., 2001; Wood et al., 2004). Other studies, however, have failed to find support for this relationship (Griffiths 1990; Lee, et al., 2007; Volberg, 2003). It is possible this inconsistency is due to motivational differences in gamblers attracted to different gambling forms. This study therefore aimed to develop a measure of EGM gambling motivations based on themes  from  Thomas  et  al.’s  (2009)  study.  This  measure  will  facilitate  examination  of  relationships between these motivations and EGM gambling behaviour. An exploratory factor analysis (EFA) was performed to determine the number of distinct motivational

4

factors. It was hypothesised that factors identified by the EFA would be theoretically consistent with themes from the qualitative study. Further, in line with the theoretical expectations of Thomas et al., the newly developed subscales were expected to show substantive positive correlations with measures of EGM gambling frequency and EGM gambling problems. Psychometric characteristics of the new subscales were examined. It was hypothesised that the factors would be internally consistent. A measure of recreational activity motivation which has been used to study gambling motivations tested convergent validity. It was hypothesised the new subscales would show significant moderate-strong, positive correlations with theoretically related subscales measuring gambling to avoid demands and responsibilities; as an oasis from demands and stressors; as a space with constant activity; and as a source of adult company. Thomas et al. (2009) did not find any evidence that motivations differed by either gender or age. Their sample achieved considerable variability according to these variables and the researchers had sought additional insight from experienced counsellors in the field so these variables were seen as appropriate measures of discriminant validity. It was hypothesised that the newly developed subscales would have weak or non-significant correlations with age and gender. Development of the item pool Items were developed to measure motivational themes from the qualitative study (Thomas et al., 2009) as follows: Electronic games provided a cognitive distraction from problems and emotions.

The lights, noises and continual play of machines focussed attention and distracted from internal and external issues.

Venues provided a physical retreat. Participants overwhelmed by problems or the

demands of others saw venues as a private oasis where they could be undisturbed.

EGM problem gamblers who were lonely or socially isolated saw venues as a source of adult company which temporarily alleviated loneliness.

Venues provided a space with constant activity for those who were bored.

Venues were a highly accessible environment with gamblers preferring venues

which were geographically (e.g., close to home or work) and temporally (e.g., open long hours) accessible.

The environment was attractive and welcoming. Staff were friendly, and it was a

safe and acceptable option for women. A total of 35 items were generated in the initial item pool with problem  gamblers’  own  descriptions retained where possible to ensure items had face validity and were accessible

5

to the intended population (Dawis, 1987; Hitchcock et al., 2005). The items were preceded by an introductory statement and rating scale described within the study procedure below. Method Participants Three hundred and fifty-five current EGM gamblers, ranging in age from 18 to 83, participated as part of a larger study. There were 232 females (M = 29.60 years of age, SD = 15.41 years) and 123 males (M = 29.64 years of age, SD = 12.29 years), two females did not report their age. Inclusion criteria restricted recruitment to people who had played EGMs at least twice in the preceding year. This ensured participants could report EGM gambling motivations whilst maximising the range of gambling frequency. The majority of participants lived in a family environment either as a couple with or without children (43%), as a sole parent with children (6%), or with parents (26%). Fifteen percent lived alone and 10% cited alternative living arrangements. The majority of participants (67%) were in full or part time employment. Nine percent were unemployed, 13% were full time students, and 11% were at home/retired. Measures Participants were asked about general demographics such as age, gender and employment status. In addition, participants completed the following measures. The Recreational Experience Preference (REP) Inventory (Driver, 1983) was developed within the context of motivational theory to understand how motivation influences recreational choices. No gambling motivational scales were found which provided comprehensive measurement of the identified EGM gambling motivations so this more general measure of motivation was included for convergent validity. The REP inventory has been used to measure desired outcomes from a range of activities including gambling (e.g., Platz & Millar, 2001; Yoshioka, Nilson, & Simpson, 2002). Introductory statements vary slightly according to the theoretical considerations of the study. Participants rate the importance of each item on a 5-point scale from 1 (Not at all important) to 5 (Extremely important). Scale scores are calculated by summing appropriate items. The full 42 scales are rarely used. Rather, appropriate scales are selected according to need. Five of the two-item scales were used in the current study which have been shown by meta-analysis to have acceptable internal consistency and validity via factor loadings and mean inter-item correlations (Manfredo, Driver, & Tarrant, 1996). The Escape Role Overloads scale was used to measure gambling as an escape from demands and responsibilities (e.g.,  “To  avoid  everyday  responsibilities  for  a  while”). The Escape Physical Stressors scale (e.g.,  “To  get  away  from  noise  at  home”) and the Privacy scale (e.g.,  “To  be  alone”) measured gambling as an oasis from demands and stressors. The Escape Daily Routine scale measured gambling as a space with constant activity to alleviate boredom (e.g.,  “To  have  a  change from  my  daily  routine”). The Meeting New People scale measured gambling as a source of adult company (e.g.,  “To  be  with  other  people  in  the  area”). All scales showed good reliability in the present study (Escape Role

6

Overloads  α=.84; Escape  Physical  Stressors  α=.75; Privacy  α=.84; Escape Daily Routine α=.73; and Meeting  New  People  α=.83). The Problem Gambling Severity Index (PGSI), a component of the Canadian Problem Gambling Index (Ferris & Wynne, 2001) is a nine item measure of the prevalence of gambling problems with two domains: (a) problem gambling behaviour (e.g., “How  often  have  you  bet  more  than  you  could  really  afford  to  lose?”), and (b) consequences of that behaviour (e.g., “How  often  have  you  felt guilty about the way you gamble or what happens  when  you  gamble?”). Questions refer to the last 12 months and responses were scored on a 4-point scale from 0 (Never) to 3 (Almost always). Respondents were also given the option of responding Don’t  know. In the current study participants were asked to focus specifically on their poker machine (EGM) gambling when answering. Scale scores were obtained by summing across the nine items with scores ranging from 0-27 and interpreted as follows: 0 = Non problem gambling, 1-2 = Low risk gambling, 3-7 = moderate risk gambling, 8+ = problem gambling. The scale has demonstrated excellent internal consistency (α  =  .84 - .92; alpha for the present study = .95) and stability (test-retest at 3-4 weeks .78). Validity has been demonstrated with high correlations between the PSGI and other established measures of problem gambling (Ferris & Wynne). Frequency of EGM gambling was measured separately for the city-based casino and venues within clubs and hotels to inform on research questions beyond the scope of this study. Play at the venues outside the casino was used to measure frequency of EGM gambling for this study as the vast majority of EGMs are located outside the casino so this provides the most accurate indication of play frequency. Procedure Ethics approval for the project was obtained from Monash University Standing Committee on Ethics in Research involving Humans. Recruitment occurred through invitations on public and electronic message boards, within local and state-wide newspapers, and on an electronic search engine. Participants could take part by (a) contacting the primary investigator who provided them with a print-based version of the questionnaire and instructions to return it via a pre-paid return envelope, or (b) filling in an electronic copy of the questionnaire available at a university linked website with responses sent to a central database. In both instances respondents completed the questionnaire anonymously in their own time. The vast majority of participants (94%) chose to complete the questionnaire online. No incentive was given for participation. To ensure that participants were not influenced by order of presentation regarding gambling motivations, items from the convergent validity subscales and alternative motivation subscales were randomly intermingled with those of the scale being developed. All questions were preceded with this common introductory statement (note pokies and poker machines are Australian colloquial terms for EGMs): “People  have  many reasons for going to the pokies. Below is a list of reasons people sometimes give for gambling on poker machines. Using the scale provided, please rate each statement in terms  of  how  much  it  applies  to  your  poker  machine  gambling”. All items were positively rated with respondents indicating how much each statement applied to their EGM

7

gambling on 6-point scale as follows: 0 (Doesn’t  apply  to  me), 1 (Occasionally applies to me), 2 (Applies to me sometimes), 3 (Applies to me often), 4 (Applies to me most of the time), and 5 (Applies to me almost always). Results All participants were current EGMs gamblers. Seventy percent reported gambling infrequently on EGMs (less than once a month), 11% gambled a few times a month and 19% gambled on EGMs at least once a week. Measurement of EGM gambling problems showed that 66% of participants were at low or no risk of EGM gambling problems, 12% were at moderate risk, and 23% scored at high risk of gambling problems, i.e., EGM problem gamblers. EFAs were performed with no preconceptions about either factor structure or number, so the maximum likelihood method of factor extraction was used together with oblique rotations to permit correlated factors to emerge (Tabachnick & Fidell, 2007). The first factor analysis  employed  Kaiser’s  criterion  (retaining  only  factors  with  an eigenvalue greater than one) and a Scree plot to gain a preliminary idea of the number of independent factors present in the data. The initial analysis suggested that the data were suitable for factor analysis (KMO =  .95,  Bartlett’s  Test of Sphericity significant p<.001). Five factors had eigenvalues greater than one but the Scree plot suggested a maximum of three distinct factors. As the theoretical model used to underpin this study (Thomas et al., 2009) had shown five theoretical themes, the two, three, four and five factor solutions were all examined to ascertain which solution was the strongest and most meaningful in terms of items and theory. In examining competing solutions the following was considered. Factor loadings greater than .32 were examined as significant, with factor loadings greater than .4 considered substantial (i.e., of practical significance). Communalities less than .3 were considered low (i.e., the solution was only explaining a small proportion of the variance in the variable). Factors with significant loadings on more than one factor were considered potentially problematic, particularly if they failed to show discriminability (i.e., they loaded at similar strengths on more than one factor). Finally, but most importantly, factors were considered in terms of their meaningfulness, in other words that the variables factored together in meaningful ways (Hair, 2005; Tabachnick & Fidell, 2007). In the initial five factor solution, three factors appeared reasonable but the fourth and fifth factors were problematic. These factors contained only three and four variables respectively, the vast majority of which also loaded significantly on other factors. In the four factor solution three factors appeared promising but all variables in the final factor loaded significantly on other factors suggesting this was not an independent factor. The three factor solution showed promise with each factor attracting a reasonable number of items and the majority of variables loading significantly on only one factor; an approximation  of  Thurstone’s  Simple  Structure. Initial inspection of the factors indicated they were meaningful and related well to the theoretical premise of the model from which the items were derived. In the two factor solution, the majority of items loaded onto the

8

first factor, reducing interpretability. The second factor was made up of only a few, low loading variables. The three factor solution was therefore deemed to show the most potential. While the three factors appeared meaningful in terms of the model, five items remained problematic in this solution. Three items failed to load substantially on any factor (below .40). One of those items plus an additional item failed to show discriminability (loading at similar, significant strength on more than one factor). A further item loaded significantly on more than one factor in this and several earlier solutions suggesting a lack of independence. A three-factor solution omitting these five items was undertaken. This solution approached Thurstone’s  simple  structure more closely (i.e., each factor having a number of substantial loadings and all variables loading significantly on only one factor). A final item was dropped at this point as it failed to show discriminability and the three-factor solution  recalculated.  This  resulted  in  a  three  factor  solution  attaining  Thurstone’s  simple structure which explained 57% of the variance in item responses. As expected, factors were  consistent  with  themes  from  Thomas  et  al.’s  (2009)  study. One item in the third factor loaded slightly below .4 but as it cohered with other items in the factor, had a commonality greater than .4 and did not cross load it was retained. Condensed examples of item content have been presented together with factor loadings above .30 in Table 1 below. Factor one, Escape Problems consisted of 14 items and accounted for 47.84% of the variance pre-rotation. This factor related to EGM gambling as a cognitive and physical escape from responsibilities, demands and problems of life. Factor two, Accessibility contained 10 items and accounted for 5.24% of the variance pre-rotation. The factor consisted of items relating to the easy accessibility of venues as somewhere to go. Factor three, Social Environment consisted of five items and accounted for 3.97% of the variance pre-rotation. This factor related to the warm, welcoming environment with the potential for social interaction. Items loaded positively on each factor; high scores indicated that these gambling motivations were relevant to an individual while low scores indicated that they were not relevant.

9

Table 1 Factor Loadings for Items

Escape Accessibility Social Item Number and Condensed Description Problems Environment

To stop thinking about problems .978 Provides a break from worrying .925 Don’t  think  about  responsibilities   .885 To distract from demands of life .839 Machines provide a focal point .804 Distracts from things that bother me .789 Somewhere to escape alone .775 Go when overwhelmed by demands .736 Distracts from issues outside the venue .729 No  one  knows  I’m  there .578 Somewhere to go alone .572 Be around people without talking .557 A place to unwind .501 Somewhere to go after an argument .468 Somewhere  to  go  when  there’s  nothing  to  do .838 Somewhere to go when you need it .752 I  go  when  I’ve  got  time to spare .702 Somewhere to go with something to do .688 Long opening hours .664 Somewhere to go when there is nowhere else .661 Venues are close .636 Can call in when passing (a venue) .613 Reduces boredom .542 An alternative to visiting friends .509 To be around people. .588 To meet new people .571 Can talk to someone .546 A welcoming atmosphere .525 Somewhere to go and feel safe .381 Unrotated Eigenvalues 13.872 1.520 1.152 Rotated Eigenvalues 12.959 11.911 2.933 Factors one and two were strongly correlated (r=.78), indicating that the factors were closely related. This degree of correlation may indicate the factors are dimensions of a higher order factor but inspection of the EFA solution which combined these two factors (the two factor solution) showed that a single factor lacked meaning. The two factors are conceptually distinct and cohere well with different themes from the theoretical model which provided the items. The first factor related to gambling as a cognitive and physical escape from problems while the second related to the attractions of venues as somewhere

10

to go which was geographical and temporal accessibility. There were weak correlations between factors two and three (r=.30) and factors one and three (r=.22). Both the Escape Problems and Accessibility subscales had a relatively large number of items and displayed internal consistencies over .90. A balance should be struck between reliability and conciseness, and an alpha coefficient above .90 may indicate an overly long scale (DeVellis, 2003). Item analysis was conducted on these subscales to identify and then remove redundant or ill-fitting items whilst maintaining high reliability. Eight items on Escape Problems and two items on Accessibility were removed on the basis of one or more of the following issues: low variability in responses, strong positive skews in response, low coherence with other items in the subscale, similarity in item wording, lower on corrected item total correlation and squared multiple correlations than other items, and removal having a smaller effect on alpha compared to remaining items. Escape Problems subsequently retained six items and Accessibility retained eight items. See Table 2 for the final subscales together with their respective alpha scores. Item analysis of the Social Environment subscale suggested that removal of items would lower alpha reliability. As reliability for this subscale was below .90 all five items were retained. Table 2

Final Subscale Items and Associated Alpha Reliability Scores Subscale Items and Condensed Description Alpha Escape Problems .94 1. To stop thinking about problems 2. Provides a break from worrying 3. Machines provide a focal point 4. Somewhere to escape alone 5. No  one  knows  I’m  there 6. Somewhere to go alone Accessibility .91 1. Somewhere to go when there’s  nothing  to do 2. Somewhere to go when you need it 3. I  go  when  I’ve  got  time to spare 4. Somewhere to go with something to do 5. Long opening hours 6. Venues are close 7. Call in when passing (a venue) 8. Reduces boredom Social Environment safe welcoming .71 1. To be around people 2. To meet new people 3. Can talk to someone 4. A welcoming atmosphere 5. Somewhere to go and feel safe

11

Once the final set of items had been identified, subscale scores were created by summing the scores on the relevant items and calculating the average. Each subscale score ranged from 0-5. High scores indicated the subscale was very relevant to the individual, low scores indicated that it was not relevant. Psychometric characteristics were examined. As hypothesised,  Chronbach’s  alpha  scores  indicated  the new subscales were internally consistent (see Table 2). The convergent validity was evaluated by examining correlations between subscales and theoretically relevant measures (see Table 3). As expected, the Escape Problems subscale showed moderate to strong positive correlations with REP subscales measuring escaping demands and responsibilities and oasis from stressors (Escape Role Overloads, Privacy and Escape Physical Stressors). Also as expected, the Accessibility subscale showed a moderate, positive correlation with the REP subscale measuring a place with constant activity (Escape Daily Routine). Finally, the Social Environment subscale, as expected, had a moderately strong, positive correlation with the REP subscale measuring a source of adult company (Meeting New People). Table 3

Correlations between EGM Gambling Motivations and Variables of Interest

Escape Social Measures Problems Accessibility Environment Convergent Validity Escape Role Overloads .85** Privacy .74** Escape Physical Stressors .59** Escape Daily Routine .62** Meeting New People .69** Discriminant Validity Age .40** .28** .21** Gender .12* .20** .05 Gambling Behaviour Frequency of EGM Gambling .56** .52** .18** EGM Problem Gambling (PSGI) .73** .67** .23** N=355, *p<.05, **p<.01 In line with findings of Thomas et al. (2009), the subscales were hypothesised to have weak or non-significant correlations with gender and age. The pattern of correlations in general supported this hypothesis (see Table 3). There was a stronger than expected correlation between Escape Problems and age but this remained substantially weaker than correlations between Escape Problems and measures of convergent validity. The strength of this correlation may be partially due to the tendency for older people to gamble more frequently on EGMs compared to younger people (Grant, Kim, & Brown, 2001). A

12

multiple regression showed that when frequency of gambling was controlled, the relationship between age and Escape Problems was substantially weakened (b=.022, β=.23, t=5.02, partial r=.26, p<.001). Finally, the new subscales were examined in terms of their relationships to gambling behaviour. Thomas et al.’s (2009) model of EGM problem gambling suggested that reliance on gambling as an accessible, attractive and easy means of escaping problems would lead to frequent and problematic gambling. As expected, Escape Problems and Accessibility had moderate to strong positive correlations with measures of EGM gambling frequency and gambling problems. Contrary to expectations Social Environment had only weak, positive correlations with EGM gambling frequency and problem gambling (see Table 3). Discussion Within the context of forthcoming technological innovations and the inherent risk that EGM gambling poses, it is important to understand why people go to an EGM gaming venue. This study developed a scale measuring motivations important to EGM problem gamblers with the EFA identifying three distinct motivational factors. This paper indicates that people go to EGM venues to escape cognitively and physically from problems, and because it is an accessible and social place to visit. There were appreciable correlations between measures of gambling behaviour and both gambling as an escape and due to its accessibility. The social environment factor correlates less well with gambling behaviour indicating this factor may be less associated with risk. Consistent with Thomas et al., (2009), the Escape Problems factor related to using EGM gambling as a physical as well as cognitive escape from problems and demands. As expected, the subscale had substantial positive relationships with both frequency of EGM gambling and EGM gambling problems. Gambling as a cognitive escape has become a central feature of recent models of problem gambling (e.g., Sharpe, 2002) and numerous studies have found evidence that problem gamblers tend to avoid or escape their problems (e.g., Getty et al., 2000; Thomas & Moore, 2003; Wood & Griffiths, 2007). The uniformity of findings across different types of research suggests that this motivational factor is important in explaining multiple forms of problem gambling, particularly continuous forms (Wood & Griffiths). The physical retreat offered by EGM venues, however, is not as clearly articulated in the literature. The present study showed that there were appreciable correlations between escape as a motivator and measures of escape from roles, stressors and privacy. Although other forms of gambling may offer a similar cognitive distraction they may not supply the privacy, escape from role or demands within the home that gamblers desire. The Accessibility factor related to the geographical proximity of venues and the long opening hours which provided somewhere to go and something to do when bored. This factor had substantial relationships with a measure of gambling to escape the daily routine as well as with frequency of EGM gambling and EGM gambling problems. Models of problem gambling which include accessibility tend to show it as a peripheral

13

factor, primarily related to initial gambling decisions (e.g., Blaszczynski & Nower, 2002). These results support theoretical contentions of Thomas et al. (2009) and suggest that the high accessibility to EGM gambling in countries such as Australia is integral in explaining ongoing gambling behaviour including gambling problems (Breen & Zimmerman, 2002). The strong correlation between Escape Problems and Accessibility showed that these two factors were closely related. An EFA solution which combined the two factors was examined to determine whether they were better conceptualised as a single factor but this factor was broad and lacked meaning. Thomas et al.’s (2009) model suggested that EGM problem gamblers were fundamentally motivated to gamble to escape from problems and that different aspects of the games and venue combined to facilitate this escape. It makes sense, therefore, that the motivations would be correlated to some extent. Examination of the Escape Problems and Accessibility factors in light of the theoretical model suggests that they are related but conceptually distinct factors. One describes the retreat that gambling offers while the other describes the attraction of its accessibility. The Social Environment factor related to EGM venues as welcoming and safe environments which facilitated social interaction. This factor supported prior qualitative research (Surgey, 2000) which has also found a welcoming and social atmosphere was important to female EGM problem gamblers. The factor which emerged emphasised a desire for active socialisation and was substantially correlated with a measure of gambling to meet new people. The factor did not attract items measuring passive social interaction (e.g., being around people without talking) or loneliness (e.g., can go when feeling lonely), aspects found to be important to EGM problem gamblers (Surgey; Thomas et al. 2009). The presence of items measuring active socialisation may have minimised the impact of the more subtle and complex items measuring passive social interaction. It is relatively easy to think about visiting a venue to socialise with other people but more complex to conceptualise wanting to be around other people without active interaction. The relationship between Social Environment and measures of gambling behaviour were significant but less strong than those reported for the other two gambling motivations. The comparative lack of strength in these relationships may be due to the fact that social motivations for gambling can be important to both non-problem and problem gamblers (e.g., Lee et al., 2007; Volberg, 2003). Although the factor analysis resulted in fewer factors than conceptual themes in the qualitative study (Thomas et al., 2009), the factors were consistent with theoretical contentions providing good evidence for the validity of constructs (Hitchcock et al., 2005). The consolidation of particular themes enhanced elements of the theoretical model without contradicting the underlying premise. The Escape Problems factor unified the cognitive and physical facets of escaping from problems into a single factor. The Accessibility factor combined items relating to the availability and accessibility of the venues  with  the  desire  for  something  to  do  when  bored,  reflecting  the  gamblers’  attraction to an easy option to assuage boredom. The third factor combined gambling as a means of social interaction with the warm, welcoming atmosphere of the venue.

14

Examination of psychometric characteristics showed these subscales were reliable and valid measures of gambling motivation. Subscales showed acceptable to excellent internal consistency. Further, they displayed good convergent and discriminant validity, having moderate to strong correlations with scales measuring related concepts and weak or non-significant relationships with variables which were conceptually unrelated. The relationship between Escape Problems and age was stronger than expected but remained substantially lower than correlations between this subscale and measures of convergent validity. The process used to develop this scale meant they articulated on motivations relevant to EGM problem gamblers. This limits its usefulness in research investigating differences in motivation across preferred gambling form. It would be valuable for future research to delineate motivations important to other types of problem gamblers. Findings could be used in conjunction with the scale developed here to clarify how motivations differ according to the type of gambling preferred. Those who prefer continuous forms including EGMs and roulette may be more motivated to gamble as an escape while people preferring forms which gradually build excitement, such as horse race or sports betting, may be more motivated by sensation seeking (Griffiths & Delfabbro, 2001; Sharpe, 2002). Further reliability and validity testing on this measure is required to ascertain whether findings are reliable over time and valid for other samples of gamblers. It is unknown, for example, whether the scale is valid for ethnic minorities. In conclusion, this study developed a measure of EGM gambling motivations based on a theoretical model of EGM problem gambling. It indicates that people go to EGM venues to escape, because they are accessible and for the social environment. People who gambled on EGMs to escape and for its accessibility tended to gamble more frequently and experience more gambling problems than those who do not gamble for these reasons. Social motivations were less strongly related to EGM gambling frequency and gambling problems, but the relationships were significant suggesting this motivation has some influence on gambling behaviour. The correlations between these factors suggest that gambling motivations are interrelated, providing additional insight into why this gambling form is more closely associated with gambling problems than others. Although other forms may provide a similar cognitive distraction and be equally accessible, the fact that EGM venues can provide an accessible physical oasis from problems and a warm social experience increases their appeal as a way to avoid problems. This measure can be used in future research to investigate the complex relationships between EGM gambling motivations and other aetiological factors and their combined ability to predict EGM problem gambling. These findings are also of value to clinicians as they confirm that motivations for EGM gambling are multifaceted and likely to interact.

15

References

Biddle, D., Hawthorne, G., Forbes, D., & Coman, G. (2005). Problem gambling in Australian PTSD treatment-seeking veterans. Journal of Traumatic Stress, 18(6), 759-767.

Blaszczynski, A., & Nower, L. (2002). A pathways model of problem and pathological gambling. Addiction, 97(5), 487-499.

Breen, R. B., & Zimmerman, M. (2002). Rapid onset of pathological gambling in machine gamblers. Journal of Gambling Studies, 18(1), 31-43.

Clarke, D. (2004). Impulsiveness, locus of control, motivation and problem gambling. Journal of Gambling Studies, 20(4), 319-345.

Clarke, D. (2008). Older adults' gambling motivation and problem gambling: A comparative study. Journal of Gambling Studies, 24(2), 175-192.

Clarke, D., Tse, S., Abbott, M., Townsend, S., Kingi, P., & Manaia, W. (2007). Reasons for starting and continuing gambling in a mixed ethnic community sample of pathological and non-problem gamblers. International Gambling Studies, 7(3), 299-313.

Coventry, K. R., & Constable, B. (1999). Physiological arousal and sensation-seeking in female fruit machine gamblers. Addiction, 94(3), 425-430.

Dawis, R. V. (1987). Scale construction. Journal of Counseling Psychology, 34(4), 481-489.

DeVellis, R. (2003). Scale development theory and applications. London: Sage Publications.

Dickerson, M. (2004). Analysis of clients presenting to problem gambling counselling services, July 2001 to June 2002, client and services analysis report No. 8. Melbourne: Victorian Government, Department of Human Resources.

Driver, B. L. (1983). Master list of items for Recreation Experience Preference scales and domains. USDA Forest Service, Fort Collins, CO: Rocky Mountains Forest and Range Experiment Station: Unpublished document.

Ferris, J., & Wynne, H. J. (2001). The Canadian Problem Gambling Index: Final Report, from www.ccsa.ca/pdf/ccsa-00805-200.pdf

Getty, H. A., Watson, J., & Frisch, G. R. (2000). A comparison of depression and styles of coping in male and female GA members and controls. Journal of Gambling Studies, 16(4), 377-391.

Grant, J. E., Kim, S. W., & Brown, E. (2001). Characteristics of Geriatric Patients Seeking Medication Treatment for Pathologic Gambling Disorder. Journal of Geriatric Psychiatry and Neurology, 14(3), 125-130.

Griffiths, M. (1990). The acquisition, development, and maintenance of fruit machine gambling in adolescents. Journal of Gambling Studies, 6(3), 193-204.

Griffiths, M. (2007). Interactive television quizzes as gambling: A cause for concern? Journal of Gambling Issues, 20, 269-276, from http://www.camh.net/egambling/issue20/index.htm

Griffiths, M., & Delfabbro, P. (2001). The biopsychosocial approach to gambling: Contextual factors in research and clinical interventions. The Electronic Journal of Gambling Issues: eGambling, 5, no pagination specified, from http://www.camh.net/egambling/issue5/index.html

16

Griffiths, M. & Parke, J. (2002). The social impact of internet gambling. Social Science Computer Review, 20(3), 312-320.

Hair, J. F. (2005). Multivariate data analysis (6 ed.). Upper Saddle River, N.J.: Pearson Prentice Hall.

Hitchcock, J. H., Nastasi, B. K., Dai, D. Y., Newman, J., Jayasena, A., Bernstein-Moore, R., et al. (2005). Illustrating a mixed-method approach for validating culturally specific constructs. Journal of School Psychology, 43(3), 259-278.

Lee, H., Chae, P. K., Lee, H., & Kim, Y. (2007). The five-factor gambling motivation model. Psychiatry Research, 150(1), 21-32.

Manfredo, M. J., Driver, B. L., & Tarrant, M. A. (1996). Measuring leisure motivation: A meta-analysis of the recreation experience preference scales. Journal of Leisure Research, 28(3), 188-213.

McBain, N., & Ohtsuka, K. (2001). Predicting problem gambling among poker machine players from coping styles and motivational factors. In A. Blaszczynski (Ed.) Culture and the gambling phenomenon. Proceedings of the 11th National Association of Gambling Studies conference, Sydney: National Association of Gambling Studies.

New Focus Research. (2003). Study of Clients of Problem Gambling Services Stage One Report: The experiences of problem gamblers, their families and service providers. Melbourne: Gambling Research Panel.

Petry, N. (2003). A comparison of treatment-seeking pathological gamblers based on preferred gambling activity. Addiction, 98(5), 645-655.

Platz, L., & Millar, M. (2001). Gambling in the context of other recreational activity: A quantitative comparison of casual and pathological student gamblers. Journal of Leisure Research, 33(4), 383-395.

Ricketts, T., & Macaskill, A. (2003). Gambling as emotion management: Developing a grounded theory of problem gambling. Addiction Research and Theory, 11(6), 383-400.

Schrans, T., Schellinck, T., & Walsh, G. (2001). 2000 regular VL players follow up: A comparative analysis of problem development and resolution. Nova Scotia: Nova Scotia Department of Health, Addiction Services.

Sharpe, L. (2002). A reformulated cognitive-behavioral model of problem gambling a biopsychosocial perspective. Clinical Psychology Review, 22(1), 1-25.

Slowo, D. (1997). Are all gamblers the same? An exploration of personality and motivational characteristics of individuals with different gambling preferences. In G. Coman, B. Evans, & R. Woottoon (Eds.) Responsible gambling: A future winner. Proceedings of the 8th National Association of Gambling Studies conference, Melbourne: National Association of Gambling Studies.

Surgey, D. (2000). Playing for time: Exploring the impacts of gambling on women. Melbourne: Victorian Government, Department of Human Services.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston: Pearson, Allyn and Bacon.

Thomas, A., & Moore, S. (2003). The interactive effects of avoidance coping and dysphoric mood on problem gambling for female and male gamblers. The Electronic Journal of Gambling Issues: eGambling, 8, no pagination specified, from www.camh.net/egambling/issue8/.

17

Thomas, A. C., Sullivan, G. B., & Allen, F.C.L. (2009). A theoretical model of EGM problem gambling: more than a cognitive escape. International Journal of Mental Health and Addiction, 7, 97-107.

Volberg, R. (2003). Gambling and problem gambling in Arizona. Northampton: Gemini Research.

Wood, R. T. A., & Griffiths, M. D. (2007). A qualitative investigation of problem gambling as an escape-based coping strategy. Psychology and Psychotherapy: theory, research and practice, 80, 107-125.

Wood, R. T. A., Gupta, R., Derevensky, J. L., & Griffiths, M. (2004). Video game playing and gambling in adolescents: Common risk factors. Journal of Child and Adolescent Substance Abuse, 14(1), 77-100.

Yoshioka, C. F., Nilson, R., & Simpson, S. (2002). A cross-cultural study of desired psychological benefits to leisure of American, Canadian, Japanese and Taiwanese college students. LARNet; the Cyber Journal of Applied Leisure and Recreational Research, 4, 1-1, from www.nccu.edu/larnet/2002-4.html.