Skip to main content
PMC home page
The European Journal of Public Health logoLink to The European Journal of Public Health
. 2010 Jan 27;22(2):273–278. doi: 10.1093/eurpub/ckp232

How do gamblers start gambling: identifying behavioural markers for high-risk internet gambling

Julia Braverman 1,2,✉, Howard J Shaffer 1,2
PMCID: PMC3324590  PMID: 20110273

Abstract

Background: The goal of this study is to identify betting patterns displayed during the first month of actual Internet gambling on a betting site that can serve as behavioural markers to predict the development of gambling-related problems. Methods: Using longitudinal data, k-means clustering analysis identified a small subgroup of high-risk gamblers. Results: Seventy-three percent of the members of this subgroup eventually closed their account due to gambling-related problems. The characteristics of this high-risk subgroup were as follows: (i) frequent and (ii) intensive betting combined with (iii) high variability across wager amount and (iv) an increasing wager size during the first month of betting. Conclusion: This analysis provides important information that can help to identify potentially problematic gamblers during the early stages of gambling-related problems. Public health workers can use these results to develop early interventions that target high-risk Internet gamblers for prevention efforts. However, one study limitation is that the results distinguish only a small proportion of the total sample; therefore, additional research will be necessary to identify markers that can classify larger segments of high-risk gamblers.

Introduction

The availability of Internet gambling services raises public health and public policy concerns about their potential to influence the development of gambling-related addiction.1,2 Among all people who have ever been exposed to gambling activities worldwide, ∼5% experience some gambling-related problems and between ∼0.2% and 2.1% become ‘pathological gamblers’ during their lifetime.3–10 Early intervention efforts might successfully prevent, or at least diminish, the likelihood of developing gambling-related disorders.11,12 Therefore, it is essential to identify gamblers at higher risk of developing gambling-related problems as early as possible.13 Just as biomarkers indicate the presence of pathogenic biological processes, behavioural markers can indicate high probability of developing disordered gambling or more generally addiction.

A growing body of published articles has described the actual gambling behaviour of subscribers to the Internet gambling service, ‘bwin’ Interactive Entertainment, AG (‘bwin’).14–22 However, only one article23 identified specific behavioural markers to predict gambling-related problems among this group. Xuan and Shaffer compared the gambling behaviours of self-identified live-action Internet bettors during the month prior to closing their accounts because of gambling-related problems with the gambling behaviours of a matched sample of live-action gamblers who did not close their accounts. Live-action betting allows gamblers to follow a particular sporting event and to bet on an immediate proposition within the event while the event is occurring (e.g. betting on who will be the next man out during a cricket match). Xuan and Shaffer23 found that those who closed their accounts due to gambling-related problems experienced increasing money loss, increasing stakes per bet, and increasingly shorter odds bets as the time of account closure approached. In this new study, instead of using the last segment of prospective betting patterns, we used the initial betting patterns from the same sample of gamblers used by Xuan and Shaffer.23

The syndrome model of addiction24 suggests an aetiological approach to the emergence of addiction. From this perspective, antecedent distal (e.g. genetic, post-traumatic stress disorder, etc.) and proximal (e.g. reward value, psychiatric morbidity, socio-economic status, etc.) events can influence the likelihood of developing addiction. These aetiological antecedents include individual vulnerability factors (e.g. biological, psychological and social), object exposure and repeated object interaction. There is growing evidence that many distal (e.g. neurogenetic and cultural) and proximal (e.g. psychiatric morbidity) influences contribute to the development of addiction in general and gambling-related problems in particular.10,25–29

The purpose of this study is to examine some proximal features associated with gambling that might influence or relate to the emergence of addiction. Specifically, we investigate whether several gambling characteristics cluster in a reliable way during the first month of Internet gambling to identify live-action sports betters who will later close their accounts due to gambling-related problems.

We considered four characteristics of first-month gambling to be important candidate variables that might distinguish between account closers and other gamblers: gambling frequency; gambling intensity; gambling trajectory; and gambling variability. Research has shown that gamblers vary in their gambling frequency (i.e. the number of betting days during the period of observation) and their betting intensity (i.e. the average number of bets per day). For example, LaBrie et al.30 observed high frequency and high intensity of gambling among the characteristics of a distinct group of most involved Internet casino bettors. Another potentially important characteristic that might identify high-risk gambling behaviour early in a sequence is gambling trajectory, i.e. the tendency to increase or decrease the amount of wagered money. LaPlante et al.17 observed that most gamblers tend to decrease the amount of money they wager after the first 8-day period of gambling. However, the most involved sports bettors in this study (∼1% of the sample) did not show this ‘adaptation’; instead, they increased their betting activity. The American Psychiatric Association has identified a need to increase the amount of wagers to achieve the desired excitement previously experienced at lower levels of wagering (e.g. tolerance) or to try to make back previous losses as a criterion associated with pathological gambling.31 Finally, the variability of betting might distinguish prospectively high-risk gamblers from their lower risk counterparts. Previous studies showed that a uniform, stable and consistent gambling pattern characterizes the majority of Internet gamblers.15,22,30 Deviation from this prototypical behaviour pattern could be a marker that distinguishes high-risk gamblers from recreational gamblers.

Consequently, we hypothesize that gambling frequency, intensity, variability and the tendency to increase or decrease wagers (trajectory) during the first month of Internet live-action gambling, will identify reliable and meaningfully different subgroups of gamblers. In addition, we examine whether members of any of the identified subgroups are more likely to develop gambling-related problems than others.

Method

Participants

We selected the analytic sample from the full research cohort that included 48 114 people who opened an account with the Internet betting service provider ‘bwin’ during February 2005. For a complete description of this sample, interested readers should see LaBrie et al.15 About half of the full sample (21 996 participants) engaged in live-action gambling more than three times during the 2-year observation period. Of those, 1758 formally closed an account after 1 month and before the end of a 2-year period, and 599 reported the reason for closing by selecting one of three available choices: (i) having no further interest in gambling; (ii) being unsatisfied with the service; or (iii) due to gambling-related problems. This last choice did not specify a particular range or intensity of problems. Finally, we excluded 69 participants who had less than two active gambling days during the first month. We excluded these participants because it was impossible to calculate some variables (e.g. standard deviation and trajectory) for less than two data points. Nineteen of those 69 participants reported closing their account because of gambling-related problems. The final analytic cohort consisted of 530 participants; 176 (33%) of those reported closing their account for gambling-related problems, 98 (19%) were unsatisfied with the service provided and 256 (48%) reported having no interest in gambling.

Measures

Our gambling behaviour measures represented daily aggregates of betting activity records during the first 30 days, starting with the first live-action betting day. The daily aggregates included number of bets, amount of money wagered and winnings credited to the bettors’ account for live-action betting on that day. From this available information, we calculated four variables that describe a pattern of gambling activity: (i) frequency—total number of active days (i.e. days on which a participant placed at least one live-action bet) during the first 30 days, starting from the first day of live-action gambling; (ii) intensity—total number of live-action bets divided by frequency; (iii) variability—standard deviation of wagers; and (iv) trajectory—the trajectory of first month wagers. To calculate trajectory, we coded the active betting days sequentially (i.e. the first betting day was coded 1; the next betting day was coded 2, and so on). We then computed a linear regression model with wager as a dependent variable and a sequence number as a predictor. We used the slope coefficient of the regression model to describe the trajectory of wager. A positive slope value indicated increasing wager size; a negative slope value reflected a decreasing pattern of wager size.

In addition to these early predictors, we calculated the following variables that summarize betting behaviour for the entire period of gambling (i.e. from first live-action betting day to account closing): total wagers—sum of total stakes wagered; total number of bets—sum of bets for the whole period; average bet size—total nember of bets divided by the total wagers; period of gambling—number of days from first to last betting days; and total losses—total wagers minus total winnings.

Statistical analysis

We used a k-means cluster analysis to identify subgroups (clusters) of users with similar first-month gambling behaviours. We created the k-means clusters by associating every observation with the nearest mean. The final partition method minimizes the distances between observed scores and the cluster centres.32,33 Before clustering the Internet gamblers, to assure comparability, we standardized all variables using z transformations. After identifying clusters, we conducted a chi-square test to identify meaningful associations between cluster membership and reason for closing an account (i.e. gambling-related problems versus other reasons)

We used the k-means cluster analysis to partition observations into k relatively homogeneous subgroups or clusters. All cases that belong to a single cluster demonstrate similar patterns of behaviours, as defined by the variables included in the cluster analysis. A major drawback of k-means cluster analysis is its potential instability. Consequently, to ensure the reliability of the results, we split the sample randomly into two halves and repeated the same clustering procedure for each subgroup independently. We also calculated a Kappa degree of concordance in cluster membership by comparing memberships of both subsamples separately with the memberships of the total sample. Following this procedure for 3–10 clusters, we identified a four-cluster solution as stable and reliable. The subgroup total membership concordance for subgroup one was good (k = 0.658, P < 0.001) and for subgroup two it was perfect (k = 0.925, P < 0.001).

Results

Demographics

The total analytic sample consisted of 44 (8%) women and 486 (92%) men. The mean age at the time of registration was 28.4 (SD = 8.8), ranging from 17 to 62. The users reported that they were residents of 21 countries; the most frequent residency was Germany (53%), followed by Turkey (11%), Poland (6%) and Spain (5%). Men in the sample were younger than women [mean = 28 versus 32 years old, t(528) = 3.16, P < 0.05] and placed more bets per day during the first month [6.8 versus 5.3, t(528) = 1.98, P < 0.05] [We conducted a logarithmic transformation (natural logarithm) of all variables prior to conducting a t-test to ensure normality of the distribution]. There were no other gender differences.

Intercorrelations

Spearman’s correlations between frequency, intensity, variability and trajectory were low to moderate (all r’s < 0.60). We found the highest correlation (r = 0.571) between variability and frequency.

Description of clusters

As table 1 shows, the k-means cluster analysis identified four clusters of Internet gamblers distinguished by high versus low z-scores for each variable. Cluster 1 (high intensity, high variability, n = 15) contained gamblers who played frequently (mean = 20 days during the first month) and intensively (mean = 13.5 bets/day), with high variability of the wager sizes (mean of SD is 722 Euro) and positive trajectory (mean = 0.19). Cluster 2 (low first month activity, n = 22) contained gamblers who played very rarely (mean = 2.2 days). Cluster 3 (high intensity, low variability, n = 115) consisted of gamblers who played as frequently as the members of Cluster 1 (mean = 19 days) and intensively (mean = 14 bets/day), but bet about the same amount of money each day that they gambled [i.e. relatively low SD of wagers (mean = 82 Euro) with almost no increase in the waging size (mean = 0.03)]. Finally, Cluster 4 (moderate betting, n = 378) included the majority of gamblers: those who played rarely (mean = 7 days), not intensively (mean =4 bets/day), with low variability in their wager size [M(SD) =34 Euro] and the trajectory close to zero (mean = 0.09).

Table 1.

Standardized scores of cluster centres on gambling behaviour characteristics

Cluster 1 High activity, high variability (n = 15) Cluster 2 Low first- month activity (n = 22) Cluster 3 High activity, low variability (n = 115) Cluster 4 Moderate betting (n = 378)
Frequency (days) 2.63489 −0.54361 2.39258 0.27904
Intensity (bets) 1.78653 0.03928 1.89973 0.00430
Variability (Euro) 4.40874 0.15649 0.26157 −0.04460
Slope 0.26706 −2.48611 0.14316 0.22323
Open in a new tab

We conducted a series of post hoc one-way analysis of variance (ANOVAs) tests to examine the differences between the four clusters. All differences for the four primary measures were significant with F’s (3526) ranging from 61.6 (variability) to 149.8 (frequency) (P < 0.001 for all comparisons).

Cluster groups’ characteristics and association with gambling-related problems

A chi-square analysis revealed a significant relationship between cluster membership and reason for closing an account (χ2 = 13.58, P < 0.01). We contrasted the group that closed their account for gambling-related problems with the other two account closing groups (i.e. those who reported being unsatisfied with the service and those with no interest in gambling). The vast majority (73%; N = 11) of members of the high activity, high variability cluster reported closing their accounts for gambling-related problems. This compared to only 45% (N = 10, low first-month activity), 29% (N = 33, high activity, low variability) and 32% (N = 122, moderate betting), of members from the other three clusters. The other three clusters did not differ significantly from each other.

Post hoc one-way ANOVAs [All the variables except trajectory were not normally distributed. Consequently, we performed a logarithmic transformation (natural logarithm) on these variables to ensure normality.] revealed statistically significant differences between the members of each of the four clusters for the following variables that describe the entire period of gambling: total wagers [F(3526) = 59.45, P < 0.001], total bets [F(3526) = 91.53, P < 0.001], average bet per day [F(3526) = 75.91, P < 0.001) and total losses [F(3526) = 5.05, P < 0.001]. Table 2 presents the mean values of these variables for each subgroup. There was no significant difference for the period of gambling. These analyses further revealed that the members of the high-intensity, high-variability gambler subgroup wagered significantly more money than did members of other clusters (P < 0.01). Also, the member of both high intensity subgroups made more bets and placed them more frequently than the members of low first-month activity and moderate betting clusters (P < 0.01). Finally, members of the High Intensity, High Variability cluster lost more money than did members of the other clusters, P < 0.05).

Table 2.

Mean values of variables that describe gambling behaviours of different clusters for the entire period of gambling

Cluster 1 High intensity, high variability Cluster 2 Low first- month activity Cluster 3 High intensity, low variability Cluster 4 Moderate betting
Period of gambling (days) 447.33 337.86 425.78 360.33
Total wagers (Euro) 74 085.80 6317.81 23 660.03 5444.43
Total bets (N) 1165.66 165.91 1995.76 317.58
Bets per day (N) 10.65 5.42 12.64 4.92
Total losses (Euro) 4308.62 809.87 1705.85 408.42
Open in a new tab

To determine whether demographic characteristics were associated with subgroup membership, we conducted a series of chi-square analyses for categorical variables and ANOVA for the continuous age variable. These analyses revealed no significant gender, nationality or primary language differences among members of the four clusters (P > 0.4 for all variables). However, ANOVA demonstrated that the members of both high intensity clusters were slightly older (mean age = 30) than members of low first-month activity (mean age = 27) and moderate betting clusters (mean age = 28) (P < 0.02).

Finally, a series of t-tests revealed that no single variable (frequency, intensity, variability or trajectory) was associated with closing an account due to gambling-related problems [t(528) ranged from 0.60 to 1.15, with P > 0.25 in all instances).

Discussion

The k-means cluster analysis identified four meaningful subgroups of Internet live-action gamblers based on their actual first 30 days of live-action betting. Gamblers characterized by high-intensity and frequency of gambling and by high variability of wager sizes were at higher risk than other gamblers to report gambling-related problems upon closing their accounts.

To our knowledge, this study is the first to use cluster analysis to identify a group of gamblers at higher risk for reporting gambling-related problems. This also is the first study to examine the first segment of a prospective period of actual betting behaviours. These results complement the findings of Xuan and Shaffer,23 who investigated the last segment of a prospective period of gambling before gamblers closed their account because of gambling-related problems. Analysing the first segment of a longitudinal sequence is more critical than a later segment if the purpose of the research is to identify gamblers who might develop problems in the future, thereby providing public health workers with the opportunity to intervene at an earlier time.

Only about half of a percent of all gamblers who registered at the ‘bwin’ site during the 2-year period ever closed their account for gambling-related problems. The challenge, therefore, is to use these few cases to correctly identify others who might have similar problems or might develop such difficulties in the future. The high-risk subgroup that we identified in this study included only ∼3% of all those gamblers who closed their accounts because of gambling-related problems. Nevertheless, identifying even a small portion of the total gamblers at-risk for developing gambling-related problems is an important step towards understanding these risks and towards identifying the larger subgroup at risk. For example, during early studies of seizure disorders, neurologists could classify the aetiology of very few accurately and most seizures remained of unknown origin. Aided by growing clinical and empirical evidence, the majority of seizures now can be classified accurately. As the evidence grows for gambling-associated risks, we expect to observe a similar phenomenon. More research will provide scientists with the opportunity to better identify the risks of developing gambling-related problems. Perhaps even more important is that although only a minority of those who self-reported gambling-related problems demonstrated the pattern of high-intensity and high-variability gambling, those who did evidence this pattern are at very high risk to close their account for gambling-related problems later. More than 70% of the members of the distinguished cluster reported closing their account due to gambling-related problems. Being a member of this high-risk subgroup also predicted losing more money during the entire period of gambling. These findings seem to clearly identify one group of high-risk gamblers.

Although the results of this study do not reveal why high wager intensity and variability predict gambling-related problems, we can speculate that external factors (e.g. social relationships, availability of time, or monetary resources) influence gamblers with problems more than ‘social gamblers’. Variability of wager sizes might reflect gamblers’ desire to stop or limit their gambling or to control their impulses. Frequent gamblers who keep their daily betting sums constant presumably have more control over their gambling behaviour than do those players who dramatically change the sums they bet. This explanation is consistent with the clinical definition of excessive gambling as an impulse control disorder31 and the results of several studies that demonstrated problematic gamblers often evidence periods of intensive gambling interrupted by the periods of remission.16,34 However, we will need more research to clarify these issues.

An examination of the distribution of cluster centres raises the question of whether high variability might be a single variable capable of predicting account closing because of gambling-related problems. However, the results demonstrate that neither variability nor other variables taken separately are sufficient to predict gambling-related problems.

The syndrome model of addiction24 suggests an aetiological approach to addiction where there are distal and proximal antecedent commonalities (e.g. neurogenetics and psychopathology), across different expressions of addiction (e.g. gambling, substance misuse and excessive eating). Therefore, the same behavioural markers that are antecedent to Internet gambling-related problems might be relevant to predict other expressions of addiction. Further studies should test this assumption by examining the early behavioural patterns of alcohol drinkers, smokers or individuals with other addiction problems.

Limitations

This study has some limitations. We analysed only the behaviours of those who closed their accounts during the 2-year period, a very small proportion of the total sample. Although we employed account closers’ self-reported gambling-related problems as an indication of actual gambling-related problems, we did not have clinical or collateral evidence about the participants’ mental status (e.g. pathological gambling). Participants’ self-report of having experienced gambling-related problems does not permit us to distinguish between levels 2 (e.g. subclinical) and 3 (e.g. pathological) gamblers within the public health nosology. Consequently, we conservatively suggest that the participants in this study represent level 2 gamblers; that is, these are people who report some gambling-related adverse symptoms, but an insufficient cluster of symptoms to warrant a clinical diagnosis.35 Furthermore, gamblers who report symptoms insufficient to call for a clinical diagnosis are likely more diverse than those whose symptom patterns warrant a diagnosis of pathological gambling. Consequently, more research is necessary to clarify the extent of psychopathology among customers who close their accounts and select ‘gambling-related problems’ as the reason for their account closing. Similarly, we require more research to better understand the initial behaviours of gamblers who do not close their account but still report either gambling-related problems or demonstrate particularly dangerous gambling behaviours.

In our analysis, we used four characteristics of initial betting behaviour. However, other factors might also be candidates for identifying at-risk gamblers. These other variables might include total amount of wagers, number of winning or losing days, number of deposits made during the first month, etc. Future studies can use the predictors found by the present analysis and other variables to build other models that can predict high-risk gambling.

Our analysis identified only a small portion of individuals with gambling-related problems. The population of problematic and pathological gamblers is likely more heterogeneous.36 Levels 2 (e.g. problem) and 3 (e.g. pathological) gamblers vary in their socio-demographic characteristics, motivation, personality types as well as their patterns of betting behaviour.12 Therefore, it is likely that we can generalize our results to only a small subgroup of gamblers. It is important to conduct further studies that will identify risk factors or subgroups for the non-identified 97% of gamblers with gambling-related problems.

This study used Internet gamblers from only one gambling service and these results might not generalize to other populations. Moreover, we only considered live-action bettors in our analysis. We found that there were a substantial number of users who reported closing their account for gambling-related problems who demonstrated little live-action betting activity during their first month or during the whole 2-year period of gambling. It is likely that these gamblers participated in other games (e.g. fixed odds betting) and this other betting activity was contributing to account closing.

In sum, this study advances the field of gambling research by providing a logical continuation of a series of investigations of actual betting behaviours. Our previous publications on this topic provided reports about overall gambling patterns14,15 and, in particular, about the last segment of betting activities before closing an account.23 In this study, we identified a subgroup of Internet gamblers at high risk for reporting gambling-related problems by using k-means cluster analysis. This method could be used in the future to investigate other high-risk groups using different variables that describe gambling behaviours. Further studies could confirm the findings for other types of gambling (e.g. poker and casino).

Funding

Bwin.com, Interactive Entertainment, AG provided primary support for this study. The Division on Addictions also receives support from the National Center for Responsible Gambling, National Institute on Alcohol and Alcohol Abuse, National Institute of Mental Health, Venetian Casino Resort, LLC., The Massachusetts Council on Compulsive Gambling, St. Francis House, and the University of Nevada.

Conflicts of interest: None declared.

Key points.

  • This is the first study to use actual gambling behaviour to identify specific behavioural markers to predict gambling-related problems.

  • This study used cluster analysis to identify a group of gamblers at higher risk for reporting gambling-related problems based on the characteristics of their initial stage of gambling.

  • The study revealed that gamblers characterized by high intensity and frequency of gambling and by high variability of wager sizes during their first month of gambling were at higher risk than other gamblers to report gambling-related problems upon closing their accounts.

Acknowledgements

The authors are thankful to Sarah Nelson, Debi LaPlante, Richard LaBrie, Heather Gray and John Kleschinsky for their support and thoughtful comments on previous drafts of the article. None of the supporters or any of the authors has personal interests in bwin.com and its associated companies that would suggest a conflict of interest.

References

  • 1.Ladd GT, Petry NM. Disordered gambling among university-based medical and dental patients: A focus on Internet gambling. Psychol Addictive Behav. 2002;16:76–9. doi: 10.1037//0893-164x.16.1.76. [DOI] [PubMed] [Google Scholar]
  • 2.Watson S, Liddell P, Jr, Moore RS, et al. The legalization of Internet gambling: a consumer protection perspective. J Public Policy Market. 2004;23:209–13. [Google Scholar]
  • 3.Petry NM, Stinson FS, Grant BF. Comorbidity of DSM-IV pathological gambling and other psychiatric disorders: results from the National Epidemiologic Survey on Alcohol and Related Conditions. J Clin Psych. 2005;66:564–74. doi: 10.4088/jcp.v66n0504. [DOI] [PubMed] [Google Scholar]
  • 4.Kessler RC, Irving H, Labrie RA, et al. DSM-IV pathological gambling in the National Comorbidity Survey Replication. Psychol Med. 2008;38:1351–60. doi: 10.1017/S0033291708002900. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Welte JW, Barnes G, Wieczorek W, et al. Alcohol and gambling pathology among U.S. adults: prevalence, demographic patterns and comorbidity. J Stud Alcohol. 2001;62:706–12. doi: 10.15288/jsa.2001.62.706. [DOI] [PubMed] [Google Scholar]
  • 6.Welte JW, Barnes GM, Tidwell MC, et al. The prevalence of problem gambling among U.S. adolescents and young adults: results from a national survey. J Gambling Stud. 2008;24:119–33. doi: 10.1007/s10899-007-9086-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Shaffer HJ, Hall MN. Updating and refining meta-analytic prevalence estimates of disordered gambling behaviour in the United States and Canada. Can J Public Health. 2001;92:168–72. doi: 10.1007/BF03404298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Shaffer HJ, Hall MN, Vander Bilt J. Estimating the prevalence of disordered gambling behavior in the United States and Canada: a research synthesis. Amer J Public Health. 1999;89:1369–76. doi: 10.2105/ajph.89.9.1369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Shaffer HJ, Korn DA. Gambling and related mental disorders: a public health analysis. In: Fielding JE, Brownson RC, Starfield B, editors. Annual review of public health. Palo Alto: Annual Reviews Inc.; 2002. pp. 171–212. [DOI] [PubMed] [Google Scholar]
  • 10.Shaffer HJ, LaBrie RA, LaPlante DA, Nelson SE, Stanton MV. The road less traveled: moving from distribution to determinants in the study of gambling epidemiology. Can J Psych. 2004;49:504–16. doi: 10.1177/070674370404900802. [DOI] [PubMed] [Google Scholar]
  • 11.Tolchard B, Thomas L, Battersby M. Single-Session exposure therapy for problem gambling: a single-case experimental design. Behav Change. 2006;23:148–55. [Google Scholar]
  • 12.Petry NM. Pathological gambling: etiology, comorbidity, and treatment. 1st edn. Washington DC: American Psychological Association; 2005. [Google Scholar]
  • 13.Albanese MJ, Shaffer HJ. Treatment considerations in patients with addictions. Primary Psych. 2003;10:55–60. [Google Scholar]
  • 14.LaBrie RA, Kaplan SA, LaPlante DA, et al. Inside the virtual casino: a prospective longitudinal study of actual Internet casino gambling. Eur J Public Health. 2008;18:410–6. doi: 10.1093/eurpub/ckn021. [DOI] [PubMed] [Google Scholar]
  • 15.LaBrie RA, LaPlante DA, Nelson SE, et al. Assessing the playing field: a prospective longitudinal study of Internet sports gambling behavior. J Gambling Stud. 2007;23:347–62. doi: 10.1007/s10899-007-9067-3. [DOI] [PubMed] [Google Scholar]
  • 16.LaPlante DA, Nelson SE, LaBrie RA, et al. Stability and progression of disordered gambling: lessons from longitudinal studies. Can J Psych. 2008;53:52–60. doi: 10.1177/070674370805300108. [DOI] [PubMed] [Google Scholar]
  • 17.Laplante DA, Schumann A, Labrie RA, et al. Population trends in Internet sports gambling. Comp Human Behav. 2008;24:2399–414. [Google Scholar]
  • 18.LaPlante DA, Shaffer HJ. Understanding the influence of gambling opportunities: expanding exposure models to include adaptation. Am J Orthopsych. 2007;77:616–23. doi: 10.1037/0002-9432.77.4.616. [DOI] [PubMed] [Google Scholar]
  • 19.Nelson SE, LaPlante DA, Peller AJ, et al. Real limits in the virtual world: self-limiting behavior of Internet gamblers. J Gambling Stud. 2008;24:463–77. doi: 10.1007/s10899-008-9106-8. [DOI] [PubMed] [Google Scholar]
  • 20.Peller AJ, LaPlante DA, Shaffer HJ. Parameters for safer gambling behavior: examining the empirical research. J Gambling Stud. 2008;24:519–34. doi: 10.1007/s10899-008-9097-5. [DOI] [PubMed] [Google Scholar]
  • 21.Shaffer HJ, Peller AJ, LaPlante DA, et al. Toward a paradigm shift in Internet gambling research: from opinion and self-report to actual behavior. Addict Res Theory. (in press) [Google Scholar]
  • 22.LaPlante DA, Kleschinsky JH, LaBrie RA, et al. Sitting at the virtual poker table: a prospective epidemiological study of actual Internet poker gambling behavior. Comp Human Behav. 2009;25:711–7. [Google Scholar]
  • 23.Xuan Z, Shaffer HJ. How do gamblers end gambling: Longitudinal analysis of Internet gambling behaviors prior to account closure due to gambling related problems. J Gambling Stud. 2009;25:239–52. doi: 10.1007/s10899-009-9118-z. [DOI] [PubMed] [Google Scholar]
  • 24.Shaffer HJ, LaPlante DA, LaBrie RA, et al. Toward a syndrome model of addiction: multiple expressions, common etiology. Harvard Rev Psych. 2004;12:367–74. doi: 10.1080/10673220490905705. [DOI] [PubMed] [Google Scholar]
  • 25.Bierut LJ, Dinwiddie SH, Begleiter H, et al. Familial transmission of substance dependence: alcohol, marijuana, cocaine, and habitual smoking. A report from the Collaborative Study on the Genetics of Alcoholism. Arch Gen Psych. 1998;55:982–8. doi: 10.1001/archpsyc.55.11.982. [DOI] [PubMed] [Google Scholar]
  • 26.Merikangas KR, Stolar M, Stevens DE, et al. Familial transmission of substance use disorders. Arch Gen Psych. 1998;55:973–9. doi: 10.1001/archpsyc.55.11.973. [DOI] [PubMed] [Google Scholar]
  • 27.Karkowski LM, Prescott CA, Kendler KS. Multivariate assessment of factors influencing illicit substance use in twins from female-female pairs. Amer J Med Genet. 2000;96:665–70. [PubMed] [Google Scholar]
  • 28.Shaffer HJ, Korn DA. Gambling and related mental disorders: a public health analysis. Ann Rev Public Health. 2002;23:171–212. doi: 10.1146/annurev.publhealth.23.100901.140532. [DOI] [PubMed] [Google Scholar]
  • 29.Petry NM. Pathological gambling: etiology, comorbidity, and treatment. Washington, DC: American Psychological Association; 2005. Brief and motivational interventions; pp. 257–65. [Google Scholar]
  • 30.LaBrie RA, Kaplan SA, LaPlante DA, et al. Inside the virtual casino: a prospective longitudinal study of Internet casino gambling. Eur J Public Health. 2008;18:410–6. doi: 10.1093/eurpub/ckn021. [DOI] [PubMed] [Google Scholar]
  • 31.Psychiatric Association American. Diagnostic and Statistical Manual of Mental Disorders—Text Revision. Fourth edn. Washington, DC: American Psychiatric Association; 2000. [Google Scholar]
  • 32.Kanungo T, Mount DM, Netanyahu NS, et al. An efficient k-means clustering algorithm: analysis and implementation. IEEE. 2002;24:881–92. [Google Scholar]
  • 33.MacQueen J. Fifth Berkeley Symposium on Mathematics, Statistics and Probability. Berkeley, CA: University of California Press; 1967. Some methods for classification and analysis of multivariate observations; pp. 281–97. [Google Scholar]
  • 34.Slutske WS, Jackson KM, Sher KJ. The natural history of problem gambling from age 18 to 29. J Abnorm Psychol. 2003;112:263–74. doi: 10.1037/0021-843x.112.2.263. [DOI] [PubMed] [Google Scholar]
  • 35.Shaffer HJ, Hall MN. Estimating the prevalence of adolescent gambling disorders: a quantitative synthesis and guide toward standard gambling nomenclature. J Gambling Stud. 1996;12:193–214. doi: 10.1007/BF01539174. [DOI] [PubMed] [Google Scholar]
  • 36.Blaszczynski AP. London: Robinson Publishing Ltd.; 1998. Overcoming compulsive gambling: a self-help guide using cognitive behavioral techniques. [Google Scholar]

Articles from The European Journal of Public Health are provided here courtesy of Oxford University Press

RESOURCES