The Diagnostic and Statistical Manual of Mental Disorders (1) describes pathological gambling as a disorder that involves preoccupation with, tolerance of, and loss of control relating to gambling. Prevalence surveys suggest that about 1.9 percent of adults in North America suffer from pathological gambling and that another 3.9 percent are problem gamblers—that is, they experience some gambling-related difficulties, but not to the extent of pathological gamblers (2,3).
Studies in the United States and elsewhere have shown that members of ethnic minority groups experience gambling problems at higher rates than members of the majority ethnic group in a given country. For example, the National Gambling Impact Study Commission (4) and another U.S. survey by Welte and colleagues (3) found that minority status was associated with an increased risk of gambling problems. In New Zealand, non-Caucasians accounted for only 15 percent of the 3,712 respondents to a national survey but accounted for 41 percent of problem gamblers (5). Similarly, an Australian survey (6) found that being of non-Caucasian ancestry was associated with an increased risk of gambling problems. In Sweden, non-native-born residents had twice the prevalence of gambling problems as respondents who were born in Sweden (7).
Although belonging to a nonmajority ethnic group appears to be associated with gambling disorders in countries throughout the world, little research has focused specifically on prevalence rates of these disorders in Asian populations. Surveys have been conducted in three Asian countries: Taiwan, Hong Kong, and Korea. These studies found prevalence rates consistent with those in European and New World countries—about 5 percent in Taiwan (8), 1.5 percent in Hong Kong (9), and 1 percent in Korea (10). In all three studies, men were much more likely than women to experience gambling problems.
In studies from the United States, Canada, Australia, and New Zealand, Asians were included in the surveys only if they spoke English. Among English-speaking Asians surveyed by Welte and colleagues (3), 6.5 percent met the criteria for current problem gambling or pathological gambling, compared with 1.8 percent of Caucasians, but these differences were not statistically significant. In addition, only 45 Asian persons were interviewed in that survey, and the respondents could not be further subdivided by country of origin. Attitudes toward gambling, the cultural context of gambling, and other life events may differ across Asian cultures.
The only known prevalence surveys of recent Asian immigrants to Western countries are limited to Chinese immigrants. In Montreal, a Chinese version of the South Oaks Gambling Screen (SOGS) (11) was administered to a nonrandom sample of 229 persons. On the basis of scores on this scale, 5 percent of the study participants were problem gamblers and an additional 2 percent were pathological gamblers (12). Even higher rates were noted in a sample recruited from Calgary's Chinatown, in which 8 percent were classified as problem gamblers (13). In another study of Chinese-speaking persons in Sydney, Australia, 10 percent of respondents were pathological gamblers (14).
Anecdotal and observational reports suggest that high rates of problem gambling and its consequences in Asian communities are not limited to Chinese populations. In Australia, Hallebone (15) reported that only two family breakups occurred among Vietnamese families in the year before the opening of a casino in Melbourne in 1993. In the years after the casino opened, more than 200 family breakups were noted in the Vietnamese community and were attributed to problem gambling. In New Zealand, Asian clients accounted for more than 20 percent of gambling losses (16). Likewise, in the United States, gambling is a popular activity in Asian communities, who organize bus trips to gambling facilities; in addition, casinos offer translation services and information materials in a variety of Asian languages.
The study reported here was designed to assess, for the first known time, rates of gambling participation and gambling problems among South East Asian refugees to the United States. Three Asian groups participated: Laotians, Cambodians, and Vietnamese. We expected that rates of recent gambling activities, as well as problem and pathological gambling behaviors, would be higher than those reported in national population surveys. We also examined whether rates of problem gambling differed by ethnic group and whether demographic factors were associated with pathological gambling.
The participants were individuals attending community service organizations for Laotian, Vietnamese, and Cambodian people in Connecticut. The organizations provide basic needs assessment and assistance as well as educational, recreational, and cultural activities. About 4,000 people are members of the Laotian, 3,500 of the Cambodian, and 10,000 of the Vietnamese organizations. Throughout a three-week period in the summer of 2002, social workers at each organization asked individuals attending the programs whether they would be willing to complete a survey about their gambling. Systematic data were not collected for persons who refused to participate, but refusal rates were estimated to be about 20 percent in each organization. Of the 96 participants included in this study, 30 were recruited from the Laotian, 30 from the Cambodian, and 36 from the Vietnamese centers.
The study was approved by the institutional review board of the University of Connecticut Health Center. After being read a description of the study, the participants provided verbal assent to participate. Written informed consent procedures were waived because of the low risks associated with participating in the study.
The questionnaire consisted of the SOGS (11), a reliable and valid instrument for assessing pathological gambling. In original studies using this instrument, internal consistency was .97, and test-retest reliability was .71. The SOGS has been translated into more than 20 languages and has generally demonstrated adequate psychometric properties when used in other cultures (17,18). For the survey reported here, persons whose native language was Laotian, Cambodian, or Vietnamese and who were fluent in English translated the questionnaire into their native languages. Reverse translations were conducted by another native speaker. All questions were read aloud to participants, because a substantial number were unable to read their native languages. Cronbach's alpha of the 96 respondents was adequate at .90.
Possible scores on the SOGS range from 0 to 20, with a score of 0 to 2 indicating nonproblem gambling, 3 or 4 indicating problem gambling, and 5 or more indicating pathological gambling (2,19). We also asked our study participants questions about demographic characteristics and recent gambling activities.
Differences among participants from the three ethnic groups were first evaluated by using chi square tests for categorical data, analysis of variance for continuous data, and Mann-Whitney U tests for nonnormally distributed continuous data. Differences were then evaluated among nonproblem gamblers, problem gamblers, and pathological gamblers by using these same statistical techniques. Logistic regression was used to identify variables that were associated with pathological gambling status. SPSS was used to conduct the analyses.
Study participants from the three ethnic groups differed on some demographic variables, as can be seen in t1. Both sexes were equally represented across the groups. Vietnamese participants were younger than those in the other two ethnic groups. Cambodian participants were more likely to be separated or divorced than Laotian or Vietnamese participants. Cambodians had the fewest years of formal education, and Laotians had the most. Income also differed across the groups; larger proportions of Vietnamese participants were in lower income brackets compared with the other ethnic groups.
In terms of gambling variables, Cambodian persons were more likely than the other two ethnic groups to wager frequently on a number of games: slots, animal races, dice, sporting events, and bingo. Data about recent gambling activities are summarized in t2. Specifically, participants were asked how often they had gambled in the previous year and how much money they typically spent when they gambled. Responses to both these questions differed among the groups (U=15.34 and U=24.13, respectively; p<.001). Cambodian respondents were more likely to report that they wagered frequently and spent more on typical gambling days in the previous year than Laotian or Vietnamese participants. Gambling frequencies and intensities within the previous two months also differed among the three ethnic groups (U=7.97 and U=12.06, respectively; p<.05).
In response to a question about the date on which respondents last gambled, the median and interquartile range of days that had elapsed between the most recent gambling day and the interview was 12 for Laotians (interquartile range, 9), 12 for Cambodians (interquartile range, 14), and 15 for Vietnamese (interquartile range, 39). The recency of the last gambling activity did not differ among the groups. However, the amount wagered on the most recent gambling day did: Laotians gambled a median amount of $50 (interquartile range, 160), compared with $350 (interquartile range, 1,900) for Cambodians and $150 (interquartile range, 420) for Vietnamese (U=19.02, p<.001).
Although overall SOGS scores did not differ significantly among the ethnic groups, responses to some of the individual items did differ. The 20 items of the SOGS are listed in the bottom half of t1 in order of frequency of affirmative responses. Because of the number of tests conducted, the p value was corrected by using the Bonferroni procedure; a p value below .002 was considered significant. Responses of the ethnic groups differed on five items: hiding one's gambling from others, wanting to stop gambling, feeling that one has or had a problem with gambling, going back to win lost money, and borrowing money from banks to gamble.
Overall, 27 of the study participants (28 percent) were nonproblem gamblers, with scores of 0 to 2 on the SOGS. An additional 11 participants (12 percent) were problem gamblers, with scores of 3 or 4, and 56 (58 percent) were classified as pathological gamblers, with scores of 5 or higher. Of all respondents, 29 (30 percent) scored extremely high on the SOGS (a score of at least 10).
As can be seen from t3, the nonproblem gamblers, problem gamblers, and pathological gamblers did not differ significantly in demographic characteristics. As expected, gambling variables differed among the three types of gamblers, with problem and pathological gamblers being more likely to gamble often and spend more money gambling on all indexes measured. Although the SOGS measures lifetime gambling problems, virtually all the problem and pathological gambling appeared to be recent—almost all respondents with scores in the problem or pathological ranges reported frequent and heavy recent gambling activity.
Logistic regression was used to identify predictors of pathological gambling. The model, which included age, education, gender, marital status, and ethnicity as independent variables, was significant (χ2=17.37, df=7, p=.01), and 66.7 percent of the cases were correctly identified. Age was a significant predictor (odds ratio of .93, suggesting that for each additional year of age there is a 7 percent reduction in risk; 95 percent confidence interval [CI] of .88 to .98), gender was significant (odds ratio of 3.49 for men compared with women; CI of 1.28 to 9.54), and marital status was significant (odds ratio 25.74 for divorced or separated compared with married; CI of 3.20 to 206.77).
The study participants were also asked whether they would be interested in participating in an evaluation of their gambling behaviors and problems. None of the nonproblem gamblers were interested, one (9 percent) of the problem gamblers were interested, and 20 (35 percent) of the pathological gamblers were interested. When queried about their interest in learning about ways to reduce or stop their gambling, four (15 percent) of the nonproblem gamblers expressed interest, as did three (27 percent) of the problem gamblers and 24 (42 percent) of the pathological gamblers.
This study evaluated gambling participation and problems among 96 South East Asian refugees to the United States. As far as we are aware, this is the first known study of gambling conducted in these particular ethnic groups. Extraordinarily high rates of gambling participation and problems were noted, with all but three respondents reporting gambling in their lifetimes, 95 percent having gambled in the previous year, and 93 percent having gambled in the previous two months. More than 60 percent of the respondents reported that they wagered more than $100 on a typical gambling day, and 42 percent reported wagering more than $500 in the two months before the interview. Given that the median annual income in this sample was about $25,000, very large proportions of income were expended on gambling.
The SOGS was translated into participants' native languages, and the instrument classified about 59 percent of respondents as pathological gamblers. This rate of pathological gambling is ten to 25 times as high as that noted in general population surveys using this same instrument (2,3). It is even higher than rates among high-risk groups. For example, although persons with substance abuse experience high rates of pathological gambling, even in treatment-seeking samples only about 5 to 20 percent of substance abusers are pathological gamblers (20,21).
One possible explanation for the high rates of pathological gambling among these refugees may be related to lack of reliability or validity of the translated versions of the SOGS. However, internal consistency remained high (alpha=.90). The frequency and intensity of gambling in the previous year were correlated with SOGS scores in this sample (r=.39 and r=.63, respectively; p<.001). Problem and pathological gamblers differed from nonproblem gamblers on all measures of gambling, providing another index of convergent validity.
Finally, the translated version of the SOGS was administered to another group of 60 Cambodian persons, half of whom were thought by the interviewer to not gamble often or problematically and the other half of whom were known within their community to gamble problematically. Among the individuals who were thought to be nongamblers, 93 percent obtained a score of 0 on the SOGS, and the remainder had scores of 1 or 2. Among those who were known to gamble often, 77 percent obtained scores in the pathological range, with a mean score of 9.1. These data suggest external validity of the translated version, at least in the Cambodian subsample.
The types of gambling problems endorsed by this sample differ from those typically reported in Caucasian samples. On the SOGS, the refugees were more likely than Caucasian samples (22) to report hiding gambling from others but were less likely than Caucasian samples to claim to be winning when actually losing, to gamble to win back losses, or to feel guilty about gambling. These latter two items, along with gambling more than intended, are the most frequently endorsed items in Caucasian samples in the same geographic area (22).
Not only do the problems associated with gambling appear to differ across ethnic groups, but so do the types of gambling in which individuals participate. Cambodian respondents were more likely than Laotian or Vietnamese respondents to gamble frequently on multiple activities, including slot machines, animal races, dice, sporting events, and bingo. In other samples from this same geographic area, lotteries represent the most common gambling activity, followed by slots, scratch tickets, cards, and sports (22,23). Of course, the popularity of gambling activities may depend not only on the cultural context but also on the availability in the community. South East Asian refugees living in other areas of the United States may not exhibit such high rates of pathological gambling as those observed in our sample, or they may prefer other gambling activities in areas in which casinos are not readily accessible.
Risk factors associated with pathological gambling in this sample were similar to those reported in other adult samples. The National Research Council (24) reviewed the literature related to risk factors for gambling disorders and noted that age was inversely associated with prevalence rates of pathological gambling in almost all studies. Younger age was likewise associated with an increased risk of pathological gambling in our study, with an odds ratio of .93.
The NRC (24) also found that men's risk of pathological gambling is two to four times that of women. In our study, the odds ratio for male gender was 3.49. The NRC (24) reported that being married may be a protective factor against pathological gambling, with increased rates of disordered gambling noted among divorced or single persons. In our sample of refugees, divorced or separated marital status had a very high odds ratio (25.74). Although these three demographic variables were associated with pathological gambling status, ethnic group—Laotian, Cambodian, or Vietnamese—was not associated with pathological gambling when other factors were held constant.
The NRC (24) reported somewhat mixed results in studies examining associations between socioeconomic status and pathological gambling, but in general inverse relationships were noted. In the study reported here we did not find an association between educational achievement and pathological gambling, but—not surprisingly—the level of formal education was substantially lower among these refugees than in the general population of the United States. Similar results were obtained when income, which was correlated with education, was included in the logistic regression (data not shown).
More than a third of the pathological gamblers identified in this study were interested in learning about ways to reduce or stop their gambling. These rates of interest in assistance are consistent with those noted in English-speaking samples. For example, we asked the same questions of 904 employees (22) and 398 general medical or dental patients (23) who also completed the SOGS, and about half of those who were identified as pathological gamblers reported interest in learning about ways to reduce gambling. Thus, despite the low rates of mental health treatment seeking in many Asian cultures (25), a substantial number in this sample reported interest in obtaining information about gambling-specific services.
Strengths of this study include the fact that the SOGS was administered to groups of individuals who are rarely captured by general population surveys and for whom gambling may serve an important cultural role. Translation of the instrument and administration to three distinct ethnic groups is also a unique feature of this study, and the response rate appeared to be equal to or higher than that found in many surveys (3,26). Initial attempts at validation of the SOGS suggest reliability and validity of the translated versions.
However, some weaknesses of the study are also apparent. Further psychometric testing of the SOGS in these languages is needed. In addition, the individual ethnic groups were small. Furthermore, we did not use random sampling procedures to recruit all members of these ethnic groups, and there may have been response bias. Refugees who gamble frequently or who have a gambling problem may be more likely to attend community centers than those who do not attend these centers and who may be more integrated into American culture.
Nevertheless, the results of this study call for further investigation of gambling and pathological gambling among South East Asian refugees, especially those who obtain services at community centers. Larger-scale, random-sampling surveys in the United States and other countries where South East Asians immigrate are needed to further investigate prevalence rates and correlates of pathological gambling in these populations. Many members of South East Asian community centers (27) experienced physical torture in their home countries. Gambling may have a unique draw for persons who have experienced severe and persistent abuse. Given the high rates of pathological gambling in this sample, development of culturally sensitive services to help these refugees with their gambling and other psychosocial problems is imperative.
This research was supported by the State of Connecticut Department of Mental Health and Addiction Services; Problem Gambling Services; grants R01-MH-60417, R01-MH-60417-suppl, and M01-RR-06192 from the National Institutes of Health; and the Patrick and Catherine Weldon Donaghue Medical Research Foundation. The authors thank Yola Wray and George Ladd, Ph.D., for their assistance with data management.
Dr. Petry is affiliated with the department of psychiatry at the University of Connecticut Health Center, 263 Farmington Avenue, Farmington, Connecticut 06030-3944 (e-mail, firstname.lastname@example.org). Mr. Armentano is with Problem Gambling Services in Middletown, Connecticut. Ms. Kuoch is with Khmer Health Advocates in West Hartford, Connecticut. Dr. Norinth is with the Lao Association of Connecticut in Bridgeport. Ms. Smith is with the Coalition of Mutual Assistance Associations in Hartford, Connecticut.
Demographic and gambling variables for participants in a survey on gambling among South East Asian refugees, by ethnic group
Gambling variables for participants in a survey on gambling among South East Asian refugees, by ethnic groupa
a Not all study participants responded to all questions
Demographic and gambling variables for Laotian, Cambodian, and Vietnamese refugees, by type of gambler