Introduction

Studies completed to provide estimates of the extent of problem or pathological gamblers within the prison population necessarily miss a very large portion of offenders who enter the Criminal Justice System. Due to the funnel effect (Champion 2004) from initial arrestees to prison inmates, there is a loss of a large portion of the arrestee population that remains hidden from estimates of the size of the total forensic gambling population when only a prison population is studied. It is in this context of identifying a heretofore not documented population of gamblers in Florida at the earliest stage of the criminal justice process (at the intake/booking facility) that this study was undertaken. Since intake/booking facilities in cities and counties throughout the US process all individuals arrested within a given jurisdiction they are optimum places to interact with individuals who have come into contact with the Criminal Justice System but who may not remain within the system after the first court appearance. Fingerprinting, photographing and medical screening are among the procedures commonly done at intake/booking facilities.

The importance of identifying problem gamblers among arrestees takes on special significance when we consider the possibility that it could enable the means for intervention and diversion that may ultimately reduce crime recidivism related to gambling thus reducing costs to the State very much in the same manner that drug courts, for example, have done (NIJ 2006). This is likely to be a somewhat large population in most urban areas according to the extant literature of persons who have gambled and have also committed crimes. Some have studied the gambling and crime connection within the framework of violations of anti-gambling laws, e.g., numbers, racing, or sports betting through bookmakers (Abadinsky 2009; Liddick 1999), while others have conducted studies of problem gamblers who have committed property or violent crimes that may or may not have been related to gambling needs (for a review and analysis of studies see Williams, Royston and Hagen, 2005).

The research presented and analyzed in this study seeks to: (1) demonstrate if a recently developed screening instrument (Lie/Bet Questionnaire) can expeditiously identify problem gamblers during the initial Criminal Justice System intake stage, and (2) examine the relationships between gambling, problem gambling and criminality found among arrestees. In examining the relationships between gambling, problem gambling, and criminality, a number of important questions arise when we consider arrestee populations:

  1. 1.

    What types of gambling specific crimes have resulted in arrests?

  2. 2.

    What is the frequency of gambling among arrestees by the time of the intake?

  3. 3.

    What is the size of the problem gambler population among arrestees?

  4. 4.

    What types of crimes are committed by gamblers and problem gamblers compared to non-gamblers?

Literature Review

Gambling and Criminals

In a thorough literature review of national and international studies regarding problematic gambling among criminals who are in custody, Williams, et al. (2005) found 16 studies that had been conducted in the United States between 1985 and 2002. The U. S. studies of inmates in state, federal and local facilities, as well as one sample of probationers, surveyed 11,903 subjects including approximately 3,820 females. The predominant instrument used to evaluate problem gambling was the South Oaks Gambling Screen (SOGS) or some variation of it, used by 15 of the studies. One study used the National Opinion Research Center DSM IV Screening Device (NODS). According to Williams, et al. (2005) these studies indicate that about one-third (33%) of the criminal offenders studied are problematic gamblers. Although the percent of offenders attributing their incarceration to their gambling activities was relatively low, from 4 to 9%, a larger percent of offenders reported having engaged in criminal activities to pay for debts or be able to gamble: 15–24%. Only one of the studies (McCorkle, 2002) reported that 6% of offenders found to be problem or pathological gamblers indicated ever having received treatment.

The McCorkle study (2002) is particularly important for the purposes of the present study since it is the only one that surveyed arrestees at the time of intake. McCorkle conducted his study of the prevalence of problem and pathological gambling among arrestees in Las Vegas, Nevada and Des Moines, Iowa. Tapping into a quarterly sample accessed at booking facilities by the National Institute of Justice’s ADAM (Arrestee Drug Abuse Monitoring) programs in these two cities, McCorkle obtained responses from 2,307 arrestees. The screening questions to determine problematic or pathological gambling used were those developed by the National Opinion Research Center (NORC) in their national prevalence study commissioned by the National Gambling Impact Study Commission of 1999 and later constructed as the National Opinion Research Center DSM IV Screening Device (NODS).

As with most of the other studies analyzed by Williams (2005), McCorkle found a higher prevalence of problem and pathological gamblers among offenders compared with the general population. This was particularly true for Las Vegas, NV where 16.2% of the arrestees were found to be problem or pathological gamblers compared to 7.5% of the general population. When examining the differences in seriousness of crime engaged in, misdemeanor versus felony, no differences were found between gamblers, pathological gamblers, and non-gamblers. That is, pathological gamblers were no more likely to engage in felonies that non-gamblers. Regarding the type of crime charged with (i.e., violent, property, drug sale, drug possession, or other) McCorkle found that “gambling disorders and the nature of arrest charges were unrelated in both Las Vegas and Des Moines” (2002, p. 38). However, analysis of self-reported engagement in certain crimes (asked only of those reporting gambling activities during the past 12 months and having incurred losses of over $100) found that pathological gamblers were more likely to report committing assault, theft, or drug sales in the 12 months prior than other gamblers.

The Lie/Bet Instrument

The Lie/Bet Questionnaire was the result of Johnson, Hamer, Nora, Tan, Eisenstein, and Engelhart’s effort in 1997 to explore the development of a shorter screening instrument than the ones that existed at the time. Johnson, et al., considered the existing screening devices to be too long and time consuming for the situations in which they often needed to be applied. These instruments were: “the 16-item South Oaks Gambling Screen (Lesieur and Blum 1987), the Gamblers Anonymous Twenty Questions (1984), and the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria of the American Psychiatric Association (1994).” (1997, p. 83–84) Johnson et al. (1997) developed their instrument based on a study of a sample of 362 men (191 men predetermined to be pathological gamblers and 171 men predetermined to be non-problem gamblers). They were asked to complete a questionnaire consisting of the “Demographic Information Form, the Gamblers Anonymous Twenty Questions, and a 12-item Gambling Questionnaire adapted from the Diagnostic criteria for pathological gambling listed in DSM-IV” (Johnson et al. 1997, p. 84). In their analysis of the data, two questions were found to be salient as predictors for both the pathological and the non-problem gamblers. These “Yes/No” questions (both from the Gamblers Anonymous questionnaire) were: “Have you ever felt the need to bet more and more money?” and “Have you ever had to lie to people important to you about how much you gambled?”

The validity of using only these two questions as a diagnostic tool was tested through evaluation of its Sensitivity (defined as “the probability that a test result will be positive, given that the person has the disease”) and its Specificity (defined as “the probability that a test result will be negative, given that the person does not have the disease”) (Johnson et al. 1997, p. 86). The sensitivity or negative predictive values (accuracy in predicting the non-problem gamblers) and the specificity or positive predictive values (accuracy in predicting the pathological gamblers) were found to be very high, .99 and .92, respectively.

The researchers replicated this study in 1998 in order to correct the limitation of not having had any females in the 1997 sample. As in 1997 the validation of the Lie/Bet Questionnaire found the instrument to be highly accurate in its identification of pathological gamblers. The predictive patterns of the Lie/Bet Questionnaire were found to be similar for both males and females, and the prevalence of pathological gambling was found to be significantly greater among males (Johnson et al. 1998). In sum, the screen was not only highly likely to correctly identify those with a gambling problem, but perhaps more important if using a screening instrument in an environment where funding may be an issue, it was even more accurate in identifying those who did not need unnecessary further intervention.

The Lie/Bet Questionnaire has also been assessed to be useful when applied to a youth population (Rossow and Molde 2006) and senior gamblers, over 60 years of age (Zaranek and Lichtenberg 2008). Zaranek and Lichtenberg (2008) found that being identified, as a pathological gambler by the Lie/Bet Questionnaire is significantly associated with, among other variables, visiting casinos, poor mental health status, being a widower, and a low social support network. These factors have been found by other researchers as well to be related to problem and pathological gambling among different samples (e.g., National Research Council 1999; Shaffer and Korn 2002; Volberg 2003).

A review of the literature since the development and validation of the Lie/Bet Questionnaire indicates that the Lie/Bet is as good a screening device as any other screening instrument available for a variety of samples and has the important advantage of being very short and thus minimally intrusive in such instances as the booking process. However, a note of caution: The Lie/Bet was never intended as a full diagnostic instrument but merely a convenient first line screening device. Hence it may not adequately classify individuals at the early stages of the development of their gambling problems.

Methodology

Data Collection

Data was collected during the intake process at the Central Booking Facility at Orient Road, Tampa, Florida. Based upon the rate of arrests for the three-month period of March, April and May of 2009, the estimated numbers of arrestees to be interviewed for this sample was calculated to be 999 for the data collection period of March through May of 2010. In total 1,445 usable questionnaires were obtained in order to reach a sample Margin of Error of 3% and a Confidence Level of 95%. Interviews were conducted by trained Graduate Students from the University of South Florida’s Criminology Department and their supervisor who worked in shifts to maximize the kinds of arrest conditions from early morning to late at night covering a sample of each day throughout a week.

It was suggested by supervisors at the booking facility, to minimize confusion and any possible disruption to the booking process, that the interview be part of the stages the inmate would could through prior to appearing before the Judge. It was determined that the period after the medical checkup and before the inmate changed out of the jail uniform would be the best. Interviews were conducted in a private area out of the hearing of the other arrestees and staff.

Voluntary participation was sought from all arrestees by reading a statement concerning the purposes of the study and explaining that the study was: “…designed to estimate the need for additional gambling education, prevention and treatment programs.” Even though the statement contained assurances concerning confidentiality and safety of cooperation 486 of the 1,445 arrestees approached, termed “Decliners” decided not to provide the interviewers with the information requested in the questionnaire concerning their possible gambling behavior and its relationship to present and past criminal activity. However, much of their remaining data: age, ethnicity, gender and current charges, was obtained from the County Sheriff’s public access web page (www.hcso.tampa.fl.us/PublicInquiry/ArrestInquiry) for comparison. A total of 959 arrestees of the 1445 approached did agree to cooperate. These we have termed “Volunteers” for presentation in the findings below.

Survey Instrument

Because of the decision to reach the target population at the point prior to the arraignment process, the instrument that could be used had to have minimal intrusive properties as well as taking as brief a period of time as possible regarding questions asked directly from the subjects. To obtain prior criminal activity related to gambling, only two questions on self-reporting of relationship to past and present activities had to suffice (Did gambling have anything to do with the allege crime(s) for which for are being booked? and Was gambling ever involved in any other crimes you may have committed in the past?). Prior gambling was also established through only two questions related to frequency for 30 days and past year (How many times have you gambled for money in the past 30 days? and Overall, during the past year, about how often would you say that you gambled for money?). The Lie/Bet questions (presented above) were used to determine possible gambling problems or pathology. Socio-demographic information (age, education and race/ethnicity) was kept to a minimum and if omitted by the respondents was obtained through matching the booking number with the public information data available on the Sheriff’s website where the Charge Descriptions, Charge Class, and Charge Types at time of arrest was also obtained.

The questionnaire was translated into Spanish as well and both versions were pre-tested and modified prior to final usage. The time estimate for each interview was generally only about five minutes since much of the data was obtained or verified at the Sheriff’s website.

Findings

Sample Description

Eleven hundred and eighty-four (81.9%) persons in the total sample of both Volunteers and Decliners (N=1445) were male and 258 (17.9%) were female. Three persons (.2%) claimed that they were “Transgender.” Of the 959 subjects in the Volunteer group 772 arrestees were male (80.6%) and 186 (19.4%) were female. The ages of the total sample of Volunteers and Decliners ranged from 18 to 73 with a mean age of 32.6 and a median age of 30.0. The mean age for all males was 32.6 and for females 32.8. For the Volunteers only, the mean age was 32.9. For the Decliners, the mean age was 31.9 and the difference was not significant. Over forty-one percent (41.6%) of the entire sample of 1,445 was African-American or Black, 38.9% were White, 17.8% identified themselves as Hispanic or Latino and 1.7% were from other racial/ethnic designations. Among the Volunteer group there were 343 Blacks (35.8%), 405 Whites (43.3%), 185 Hispanics (19.3%), and 23 Others (2.4%). For the Decliners, 256 Blacks (52.9%), 155 Whites (32.0%), 71 Hispanics (14.7%), and 2 Others (.4%). The ethnic difference between Volunteers and Decliners was statistically significant (P < .001). Education information, available only for the Volunteers, showed that nearly one-third, 30.6% (291) did not graduate High School; 37.4% (356) had graduated High School, completed their GED or graduated from a trade school; 20.1% (191) had some College; and 12% (114) were College graduates.

Gambling Related Arrests

Given the extent of legal gambling in the area surrounding and including Hillsborough County, it is not surprising that only 1 person in the entire sample of 1,445 arrestees had an arrest for a gambling violation. However, 12 arrestees 1.3% of the 959 Volunteers claimed that the current alleged crimes for which they were being booked was related to their gambling behavior but were not gambling specific crimes. No specific pattern was reported regarding the types of non-gambling specific crimes for which they had been charged.

When the 959 Volunteers were asked about past crimes and gambling, the number of arrestees claiming that gambling was ever involved in any other crimes they committed (but not necessarily resulting in an arrest) rose to 20 (2.1%). These crimes were also distributed over a number of categories including five persons admitting to violent crimes related to gambling. When combining the data from past and present crimes committed, we find that 28 persons (2.9%) had committed some non-specific gambling related crimes for which they may or may not even have been arrested.

Frequency of Gambling

Two questions were asked of the 959 Volunteers concerning the frequency of gambling for money, one referred to the 30-day period prior to the arraignment and another to the year prior to arrest. See Table 1.

Table 1 Frequency of gambling past 30 days and past year
Full size table

As shown in Table 1, the majority of the sample (Volunteers) did not gamble or very rarely gambled in the previous 30 days or in the past year: 89 and 83%, respectively. The findings for “a couple of times a week” and “almost every day” are of particular concern and relevance to the focus of this study. While a high frequency of gambling may not at all be indicative of problem or compulsive gambling, the converse of this relationship is usually true, i.e., those persons with a problem with gambling, present or past, or who may become problem gamblers, and who are not in treatment or recovery, are almost always likely to be frequent gamblers and gamble at least a couple of times a week or more. This indicator may provide a first line suggestion of the prevalence of problem gambling within a specific population. Thus the 9% of arrestees who claimed to have gambled for money at this frequency level during the past year is troublesome.

Arrestee Problem Gambler according Lie/Bet Questionnaire

Respondent reporting they had gambled during the past 30 days or during the past year were asked the Lie/Bet questions. As mentioned, the Lie/Bet relies on just 2 questions to reach a determination: “Have you ever had to lie to people important to you about how much you gambled?” and, “Have you ever felt the need to bet more and more money?” The responses to these two questions for the Volunteers are presented in Table 2.

Table 2 The Lie/Bet questionnaire responses
Full size table

Since the interpretation of the Lie/Bet Questionnaire rests upon whether the respondent answered, “Yes” to either question the responses were combined in an Index and presented in Table 3 for all arrestees who were Volunteers.

Table 3 The Lie/Bet index
Full size table

From Table 3 we may infer a probable indication of the potential size of the population that may benefit from gambling education and/or treatment in this setting. In examining the sub-group of Volunteers who have gambled in the last year, “Gamblers Only,” we find that 32.7% of them fall into the categories suggesting that they may have a problem with gambling according to the Lie/Bet and should be referred for further assessment, diagnosis and/or treatment. The finding of 32.7% indicating probable problem or compulsive gamblers is consistent with the average percent of 33% determined by Williams, et al. (2005) in the forensic samples in the United States. However, if we include, in examination of Table 3, those who said that they did not gamble during the past 30 days or the past year then our finding drops to 17.4%. Based on the reliability and validity of the Lie/Bet this statistic is a more reasonable indicator of the presence of gambling problems in the entire population of arrestees. In the only comparable study of arrestees, McCorkle (2002) found that 16.2% of Las Vegas sample of arrestees were problem and pathological gamblers (based on his use of the NODS Screen).

In comparing the percent of problem and pathological gamblers for the entire state of Florida, our finding of 17.4% for the Volunteers is much higher than for the state as a whole, 2% (Shapira et al. 2002). Similarly, McCorkle (2002) found a large disparity between the percentage of problem and pathological gamblers among arrestees and the general population. He cites Volberg’s study (2002) as estimating the prevalence of pathological gambling in Nevada as being 3.5% compared to the 16.2% he found for arrestees.

The Lie/Bet Positives: Socio-Demographic Characteristics

In order to more easily present the characteristics of those responding positively to the Lie/Bet Questionnaire, all positive responses for the separate Lie and Bet questions were combined into a single indicator: Lie/Bet Positive. As shown in Table 4, the Lie/Bet Positives significantly differ only on Sex and Ethnicity from those that do not have a gambling problem. Males are nearly twice as likely as females to have a gambling problem and Blacks are most likely to have a gambling problem (20.2%), followed by Whites (17.5%), Hispanics (13.6%) and Others (4.3%).

Table 4 Demographics by Lie/Bet index
Full size table

Problem Gambling and Frequency of Gambling

As we indicated in the discussion following Table 1 above, the frequency of gambling in the past year is a fair but not always a reliable indication of the possibility of the presence of problem or pathological gambling. This is borne out by the finding, presented in Table 5, that 81 respondents (23.7%) of those who gambled about once a week or more frequently did not respond, “Yes” to either Lie/Bet questions. Conversely 81 (49%) of the Lie/Bet Positives never or rarely gambled during the past year but did respond positively to one or both of these questions. This is not to be viewed as an inconsistency but that the latter group’s responses to these questions may reflect early negative experiences with, or consequences of, gambling and resulted in a curtailment of their gambling activities. While it is clear that the increase in percents (from 10.3% for “Seldom Gambled” to 61.3% for the “Almost Everyday” gambler) reflects the strong relationship between frequency and problems, it would be necessary to have further screening for a more precise diagnosis or categorization.

Table 5 Lie/Bet by frequency of gambling past year**
Full size table

Crime and Lie/Bet Positives

What kinds of crimes are Lie/Bet Positives more likely to be charged with? Crimes charged at arraignment were classified into the general categories: violent (homicide, robbery, assault, domestic violence, etc.), property (burglary, theft, stolen property, etc.), drugs (sale and possession), status (concealed weapon, prostitution, open container, non-reporting of status as sex offender, etc.), and other (vehicular, warrants, in transit, court issues, etc.). In addition, since the focus of the study is gamblers the crime categories of gambling and fraud were analyzed. Table 6 shows that drugs and status crimes are those that are significantly correlated with the problem gamblers.

Table 6 Crime charge type by Lie/Bet index
Full size table

That there may have been different motivations and influencing factors involved in the commission of crimes is reinforced in Table 7 where we find that the Lie/Bet Positive gamblers are more likely than those without a gambling problem to be charged with a felony. That problem gamblers who commit crimes may be different in their “criminal profile” from non-gamblers or recreational gamblers is suggested by this finding. If when gamblers violate the law, their motivation for commission of a crime may be based more on desperation due to their needs related to gambling (whether the relationship is perceived or not) rather than other issues, and the impact that this may have on the type and severity of the crime they are engaging in needs to be further studied.

Table 7 Crime type/severity by Lie/Bet index
Full size table

Discussion

Studies of prevalence of problem and pathological gamblers within the Criminal Justice System (CJS) have been mainly conducted among prisoners (post-sentence stage). These studies have consistently found that a disproportionate number of prisoners (when compared to general population rates) can be classified as problem or pathological gamblers. Given these findings and the plausibility that problematic gambling may be a factor in the individual’s criminal behavior, it appears logical that the CJS should consider identifying individuals with gambling problems at an earlier stage of the process. This study has demonstrated the feasibility of screening inmates in a booking facility using the Lie/Bet Questionnaire with minimum, if any, disruption to the intake process. Furthermore, we found that there is a need for such screening considering the percentage of problematic gamblers that was found, as well as the greater likelihood of problem gamblers being charged with more severe type crimes (felonies).

Feasibility of Lie/Bet Screen Adoption at Booking/Intake Stage

Two important features of the Lie/Bet Screen may explain the ease with which it was used during the study and could be adapted for use at booking/intake sites: (1) the brevity of the screen (only two questions), and (2) the fact that no special training is required to administer the screen. In addition, considering that the intake process is very much a series of data gathering steps about the arrestee (photographs, fingerprinting, classification, medical data, etc.), it is not necessarily unexpected that adding the Lie/Bet Screen to the process occurred seamlessly. While establishing how effective the Lie/Bet is in correctly identifying problem gamblers in this setting is outside the scope of this study, it is certainly a study that should be undertaken in order to further validate its use at the intake stage.

Need for Problem Gambling Screen at Initial Stages of Criminal Justice System Contact

Even in conditions that some may consider less than ideal to conduct a mental health screen we found a considerable percentage of persons who were deemed to be problem gamblers as per the Lie/Bet Screen (17.4%). When we extrapolate to the total number of bookings during 2009 at the Hillsborough County Jail (60,183) we estimate that 10,472 persons at arraignment during one full year may have been found to be problem gamblers upon screening. The size of this estimated problem population is consistent with previous research findings on criminality and gambling, and is understandable from a theoretical perspective in Criminology. Many theorists have long maintained that risk takers, among them penal code violators and gamblers, are often found to engage in different kinds of risk taking behaviors during the same time periods in their lives (Gottfredson and Hirschi 1990).

We do not meant to suggest that all extrapolated 10,472 Lie/Bet Positives would necessarily be compulsive or even problem gamblers, but that further assessment using full diagnostic tools developed for such purpose or early intervention in the form of preventive education (e.g. informational brochure availability, placement of posters with 800 help line information, Gamblers Anonymous literature, etc.) is definitely warranted.

Understanding of the size of the arrestee population who may need intervention can also result in more intensive intervention during the detention stage by having gambling professionals or Gamblers Anonymous members available during this time for consultation. Institutionalization of gambling prevention and treatment programs also seems warranted to help prevent further decline for problem gamblers.

Possible Impact from Adopting a Problem Gambling Screen at Booking/Intake Stage

While the actual impact of adopting a gambling screening program can only be determined after data has been collected for such purposes we can speculate that as with existing Criminal Justice System diversion and intervention mechanisms some benefits may be obtained. The Lie/Bet Screen can easily be adopted in order to begin the process that can lead to intervention regarding problem gambling in particular for those who will move further into the Criminal Justice System in very much the same way that knowing that an arrestee has a drug problem impacts Criminal Justice System decisions regarding intervention and processing (e.g., referral to drug courts, mandatory treatment, assessment by health professionals, etc.).

We found that drugs and status crimes are those that are significantly correlated with the problem gamblers. This is not only an interesting finding but also one that may suggest policy and programmatic implications. Drugs and status crimes are for the most part crimes that, albeit arguably, do not have obvious and direct other-person victims. Whether this suggests different personality types of offenders, with problem gamblers being less of the “criminal types” or at an earlier stages of criminal careers, or they have some other distinguishing criminal characteristics is beyond the scope of this research but certainly it raises the question again of whether those with gambling problems, just as those with drug problems, should be assigned to a separate court that specializes in these problems. The objective would be to intervene at the earliest possible stage to determine appropriate societal intervention to help ameliorate anti-social behavior into the direction of society’s goals as well as provide for individual needs and protection. This practical approach has grown in the past two decades and in part provides a model for intervention with pathological gamblers. Certainly the initiation of Gambling Courts (or the expansion of drug courts to include other addictions) seems reasonably warranted given the benefits that have been found to be created by such Courts (NIJ 2006).

Early intervention for this population via screening may also prove of great value to some arrestees who may not be aware of any serious gambling problems but who may have had some thoughts of the possibility. Screening for gambling problems may lead to providing the spark to ask for help which can be provided by, at a minimum, supplying information such as brochures with help-line phone numbers.