Journal of Addictive Behaviors,Therapy & RehabilitationISSN: 2324-9005

All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.
holiganbetperabetimajbettipobetVdcasino

Research Article, J Addict Behav Ther Rehabil Vol: 2 Issue: 2

Measuring the Contingencies Maintaining Gambling Behavior in a Sample of Non, Light, and Heavy Smokers

Jeffrey N. Weatherly* and Derek Bogenreif
University of North Dakota, USA
Corresponding author : Jeffrey N. Weatherly, PhD
Department of Psychology, University of North Dakota, Grand Forks, ND 58202-8380, USA
Tel: (701) 777-3470; Fax: (701) 777-3454
E-mail: [email protected]
Received: April 26, 2013 Accepted: June 27, 2013 Published: June 29, 2013
Citation: Weatherly JN, Bogenreif D (2013) Measuring the Contingencies Maintaining Gambling Behavior in a Sample of Non, Light, and Heavy Smokers. J Addict Behav Ther Rehabil 2:2. doi:10.4172/2324-9005.1000108

Abstract

Measuring the Contingencies Maintaining Gambling Behavior in a Sample of Non, Light, and Heavy Smokers

Previous research suggests that there is a relationship between smoking and problem/pathological gambling. Research has also linked gambling problems to the contingency of escape. The present study had 45 nonsmokers, 49 light smokers, and 29 heavy smokers from the United States complete the South Oaks Gambling Screen, Problem Gambling Severity Index, and Gambling Functional Assessment – Revised. Results showed that level of smoking was related to the display of gambling problems, but the relationship was not linear. Endorsing gambling as an escape varied significantly as a function of smoking level, but gambling for positive reinforcement did not. Lastly, when only data from the smokers were analyzed, endorsing gambling as an escape, but not gambling for positive reinforcement, was a significant predictor of gambling problems. These results suggest that smoking level may not always be predictive of potential gambling problems, that smoking is related to the same factor (endorsing gambling as an escape) that is strongly related to pathological gambling, and that the same predictors of problem/pathological gambling found in samples of predominantly nonsmokers also predict problem/ pathological gambling in smokers.

Keywords: Smokers, Pathological Gambling, Escape, Positive Reinforcement

Keywords

Smokers; Pathological gambling; Escape; Positive Reinforcement

Introduction

Drug use and abuse has long been known to be associated with gambling behavior [1]. Petry [2] listed drug use as the strongest of the six risk factors for pathological gambling. In fact, she argued that substance abuse and pathological gambling were so strongly related that professionals screening for one disorder would be wise to also screen for the other. That relationship does not appear to be drug specific. For example, Griffiths et al. [3] reported significant correlations between Internet gambling and both smoking (i.e., nicotine) and drinking alcohol. In terms of smoking, Grant et al. [4] found significant relationships between level of smoking and severity of gambling problems as measured by the South Oaks Gambling Screen [5] in a sample of treatment-seeking pathological gamblers.
More recently, McGrath et al. [6] sampled a large number of Canadian smokers (n = 622) and nonsmokers (n = 997). McGrath et al. [6] reported finding significant associations between smoking and several different aspects of gambling, such as severity scores and amount of money spent while gambling. One interesting aspect of their results is that McGrath et al. [6] also attempted to measure the contingences maintaining the respondent’s gambling behavior. In an open-ended response, participants provided their main reasons for gambling. The researchers then categorized those responses into gambling for positive reinforcement, negative reinforcement, or charity. Gambling for positive reinforcement would indicate that the person’s gambling was maintained by the attainment of something, such as money or entertainment. Gambling for negative reinforcement (i.e., as an escape) would indicate that he person’s gambling was maintained by getting away from something, such as a negative emotional state or a demanding issue in one’s life. McGrath et al. [6] results showed that gambling for positive or negative reinforcement significantly delineated smokers from nonsmokers, but that endorsing gambling for charity did not.
These finding are of potential interest because numerous researchers have linked gambling as an escape, rather than for positive reinforcement, to the presence of gambling problems [7-11]. If the reasons for substance use and abuse are similar to the reasons for pathological gambling, or if substance use and abuse contributes to pathological gambling, then one would expect to find similar relationships with substance use and abuse as with gambling problems. Phrased differently, if endorsing gambling as an escape is strongly related to gambling problems, then one would expect that substance use and abuse would be more closely associated with gambling as an escape than it would be with gambling for positive reinforcement. However, that result was not observed by McGrath et al. [6].
Several aspects of the procedure used by McGrath et al. [6] might explain why this predicted result was not observed. For one, their procedure did not use a psychometrically validated measure of gambling for positive reinforcement or negative reinforcement/ escape. Secondly, their analyses used the contingencies maintaining gambling behavior to predict smoking behavior, not vice versa. Thirdly, their procedure classified respondents into two categories: smokers and nonsmokers. The outcome predicted above might have been observed had the respondents been separated along a continuum of smoking (nonsmokers, light smokers, & heavy smokers), as these groups have been demonstrated to differ from one another on a variety of measures [12].
The goals of the present study were threefold. The first was to replicate previous findings that level of smoking was related to level of gambling problems. The second was to determine whether level of smoking would be more strongly related to gambling as an escape than gambling for positive reinforcement using a psychometrically validated measure of the contingencies that control gambling behavior. The third was to determine, in a sample of smokers, whether severity of gambling problems would be more strongly related to gambling as an escape than gambling for positive reinforcement as has been found with nonsmokers.
Making these determinations would seem useful at both a theoretical and applied level. On a theoretical level, such results would help advance our understanding of the relationship between smoking and gambling by potentially identifying the contingencies that are maintaining the person’s behavior. At the applied level, finding that numerous behavioral problems are associated with the contingency of escape would be potentially useful for practitioners because screening for behavior controlled by such contingencies might be broad useful instead of having to screen for multiple potential individual disorders.
The hypotheses were as follows. First, level of smoking would be positively and linearly related to level of gambling problems. Specifically, participants who qualified as heavy smokers were predicted to display the greatest levels of gambling problems. Second, level of smoking would be more strongly associated with gambling as an escape than for positive reinforcement. This prediction was based both on the first hypothesis and previous research that has linked gambling problems with gambling as an escape. The final hypothesis was that, for smokers, gambling problems would be more strongly related to gambling as an escape than for positive reinforcement as has been found in samples of predominantly nonsmokers. Addressing this last hypothesis will be helpful in determining whether smokers differ functionally in their gambling behavior relative to nonsmokers.

Method

Participants
The participants were 45 nonsmokers, 49 light smokers, and 29 heavy smokers1. All participants were citizens of the United States and were 21 years of age or older. The median age of participants in each of the three groups was 25 - 34 years of age. Other demographic information pertaining to all three groups can be found in table 1. Participants received monetary compensation on Amazon.com in return for their participation. Participants were recruited through Amazon’s Mechanical Turk (MTurk; http://www.mturk.com) by posting the study’s title and description.

Materials and Procedure

Five documents/materials were used. The first was an informed consent form that outlined the benefits and risks of the study as approved by the Institutional Review Board at the University of North Dakota. The second was a demographic questionnaire that asked participants about the information presented in the participants section and table 1.
Table 1: Demographic information for participants in each smoking group.
The third material was the SOGS [5]. The SOGS consists of 20 self-report items pertaining to the individual’s gambling history. Researchers have interpreted scores of 0 to 2 on the SOGS to indicate no problem/pathological gambling, scores of 3 or 4 as suggesting possible problem gambling [13], and scores of 5 or more as suggesting the probable presence of pathology [5]. Lesieur et al. [5] reported that the internal consistency of the SOGS was high (α = 0.97). Other researchers have reported fair (α = 0.69) [14] to good (α = 0.81) [15] internal consistency. The internal consistency of the SOGS in the present study was α = 0.88. The temporal reliability of the SOGS has also been shown to be good (r = 0.89 at four weeks and r = 0.67 at 12 weeks) [16].
The fourth material was the Problem Gambling Severity Index (PGSI) [17], which was designed to measure gambling problems in the general population and the consequences associated with that gambling. The PGSI consists of 12 self-report items, with only the first nine used to calculate the individual’s score. Research indicates that the PGSI is psychometrically sound [18]. All items are answered on a four-point scale ranging from 0 (Never) to 3 (Almost always), which are then summed to provide a total score. Rockloff and Dyer [17] suggested that scores of 0 indicated no gambling problems, from 1-2 indicated a low level of gambling problems and few negative consequences, from 3-7 indicated a moderate level of gambling problems and some negative consequences, and of 8 or more indicated problem gambling that included negative consequences. Rockloff and Dyer [17] demonstrated that the PGSI had good internal consistency (α = 0.84), which has been replicated by subsequent research [20]. The internal consistency of the PGSI in the present study was α = 0.96. Ferris and Wynne also reported the measure has good temporal reliability (r = 0.78).
The final material was the Gambling Functional Assessment - Revised (GFA-R) [21], which was designed to determine whether the individual’s gambling behavior is maintained by positive reinforcement or escape. The GFA-R consists of 16 self-report items that are answered on a scale ranging from 0 (Never) to 6 (Always). Eight items each are associated with gambling for positive reinforcement and as an escape, with the scores for each subscales being calculated by summing the scores across those eight questions. The GFA-R has been shown to have high internal consistency (α = 0.91) [16] and good temporal reliability (r = 0.80 at four weeks and r = 0.81 at 12 weeks). The internal consistency of the GFA-R in the present study was α = 0.90. The psychometric properties of the GFA-R have also been shown to be retained when tested in different cultural settings [22,23].
Participants completed the study on Amazon’s MTurk. The study title appeared on the MTurk website and participants were informed that their participation time would be 15 min or less. Research on using MTurk has demonstrated that it provides a good way to reach a broad population [24] and to address issues relevant to clinical populations [25]. Participants completed the materials in the order described above. They received compensation for their participation within three days of completing the study.

Results

To test the first hypothesis, two one-way analyses of variance (ANOVAs) were conducted. The first was on participants’ SOGS scores with smoking group used as the grouping variable. Results from this analysis indicated that a significant difference in SOGS scores was observed across groups, F(2, 120) = 5.73, p = 0.004, η2 = 0.087. Tukey HSD post hoc tests indicated that the light smokers had significantly higher SOGS scores (Mean = 3.57, SD = 4.22) than the nonsmokers (Mean = 1.29, SD = 2.23; p = 0.003) but not the heavy smokers (Mean = 2.07, SD = 2.95; p = 0.134). The difference in SOGS scores between the nonsmokers and heavy smokers was not significant (p = 0.586). Results from these, and all following, analyses were considered significant at p < 0.05.
The second analysis was an identical ANOVA using participants’ PGSI scores as the dependent measure. Results of this analysis also showed that PGSI scores varied significantly as a function of smoking group, F(2, 120) = 4.70, p = 0.011, η2= 0.073. Tukey HSD post hoc tests again indicated that the light smokers had significantly higher PGSI scores (Mean = 5.10, SD = 8.14) than the nonsmokers (Mean = 1.73, SD = 3.73; p = 0.017), but not the heavy smokers (Mean = 1.93, SD = 3.26; p = 0.134). The difference in PGSI scores between the nonsmokers and heavy smokers was again not significant (p = 0.989).
To test the second hypothesis, two one-way ANOVA were conducted with smoking group serving as the grouping variable. The first analysis was on the GFA-R positive reinforcement subscale scores. Results showed that positive reinforcement subscale scores did not vary significantly across the nonsmokers (Mean = 16.82, SD = 9.43), light smokers (Mean = 21.16, SD = 11.52), and heavy smokers (Mean = 21.24, SD = 10.34), F(2, 120) = 2.46, p = 0.089, η2 = 0.039. The second ANOVA was on the GFA-R escape subscale scores. The result of this analysis was significant, F(2, 120) = 4.77, p = 0.010, η2 = 0.074. Tukey HSD post hoc tests indicated that the light smokers had significantly higher GFA-R escape scores (Mean = 8.04, SD = 10.35) than the nonsmokers (Mean = 3.27, SD = 6.53; p = 0.017), but that the difference with the heavy smokers did not reach significance (Mean = 3.86, SD = 5.01, p = 0.071). The difference in escape subscale scores between the nonsmokers and heavy smokers was not significant (p = 0.948). Together, results of these ANOVAs indicated that smoking level was differentially related to the contingencies maintaining gambling behavior2. To test the third hypothesis, two multiple linear regressions were conducted with the GFA-R positive reinforcement and escape subscale scores used as the predictor variables. Only data from the light and heavy smokers were included in these analyses; data from the nonsmokers were excluded because we were specifically interested in whether the same relationship between either the SOGS or PGSI and the GFA-R subscales that has been found in samples of predominantly nonsmokers would be replicated when only smokers were tested.
SOGS scores were used as the dependent measure in the first analysis. Results indicated that the regression model was significant, F(2, 75) = 57.46, p < 0.001, R2 = 0.605. However, only the escape subscale scores, β = 0.81, p < 0.001, significantly predicted SOGS scores. Positive reinforcement scores did not, β = -0.08, p = 0.369. first analysis, only the escape subscale scores, β = 0.88, p < 0.001, significantly predicted PGSI scores. Positive reinforcement scores were not a significant predictor, β = -0.07, p = 0.326.
PGSI scores were used as the dependent measure in the second analysis. Again, results indicated that the regression model was significant, F(2, 75) = 99.76, p < 0.001, R2 = 0.727. As with the first analysis, only the escape subscale scores, β = 0.88, p < 0.001, significantly predicted PGSI scores. Positive reinforcement scores were not a significant predictor, β = -0.07, p = 0.326.

Discussion

The first hypothesis was that there would be a linear relationship between smoking level and severity of gambling problems as measured by the SOGS and PGSI. That prediction was not supported. In terms of both the SOGS and PGSI scores, a bitonic relationship was observed, with light smokers displaying higher scores than either non- or heavy smokers. The second hypothesis was that level of smoking would be more strongly associated with gambling as an escape than for positive reinforcement. That hypothesis was partially supported. Gambling for positive reinforcement, as measured by GFA-R positive reinforcement subscale scores, did not differ significantly as a function of smoking level, although the results did approach statistical significance. However, light smokers scored higher than nonsmokers on endorsing gambling as an escape. The third hypothesis was that gambling problems would be more strongly related to gambling as an escape than for positive reinforcement when only data from the smokers were tested. This hypothesis was supported, with smokers’ GFA-R escape subscale scores, but not their GFA-R positive reinforcement subscale scores, being significant predictors of their SOGS and PGSI scores.
Finding that level of smoking was not linearly related to gambling problem severity is not entirely consistent with the findings in the research literature, but it is intriguing. This result may have been the outcome of the present sample and would need to be replicated before drawing broad generalizations from it. However, results would suggest that light smokers were at greater risk for gambling problems than heavy smokers. Overall, the present results would seem to promote caution when using smoking level as a potential predictor of gambling problems. It is possible that, if a predictive relationship does exist, it not be a linear relationship.
McGrath et al. [6] reported that gambling for positive or negative reinforcement were both predictive of whether participants were smokers. The present study found that level of smoking was not associated with differential endorsement of gambling for positive reinforcement. Smoking was, however, associated with differential endorsement of gambling as an escape. The difference in results between the two studies could potentially be due to a number of different things. It is possible that smoking would have been associated with gambling for positive reinforcement in the present study had the sample size been increased. McGrath et al. [6] did not use a psychometrically validated measure for gambling for positive or negative reinforcement, whereas the present study did, which may also account for the difference in results.
Regardless of the reason for the different results between studies, both studies reported that smoking is related to endorsing gambling as an escape. This similarity is important because numerous empirical and theoretical reports on gambling behavior have shown that endorsing gambling as an escape is strongly linked to pathological gambling [7-9,11]. Given that substance use and abuse is a strong risk factor for pathological gambling [2], one might predict that smoking level would be associated with endorsing gambling as an escape. The present results, and those of McGrath et al. [6], would appear to support that prediction. Coupled with the results regarding the first hypothesis, the present results would suggest that smoking level is not necessarily (linearly) related to gambling problems, but it is related to why people gamble.
The truly novel contribution of the present study is the finding that the endorsement of gambling as an escape, but not gambling for positive reinforcement, is a significant predictor of gambling problems in smokers. With both the SOGS, which is intended to measure the potential presence of pathology, and the PGSI, which is intended to measure the negative consequences of one’s gambling, endorsing gambling as an escape on the GFA-R was a significant predictor of those scores. Endorsement of gambling for positive reinforcement predicted neither SOGS nor PGSI scores. These results replicate those found with university samples populated predominantly by nonsmokers [9]. Thus, the present results strengthen the ties between endorsing gambling as an escape and problem/pathological gambling. For both researchers and practitioners, the present results suggest that it might be a useful pursuit to measure why people engage in behavior rather than focusing solely on whether their behavior is disordered in some way. It is possible that behaviors primarily maintained by negative reinforcement contingencies may be a good behavioral marker for a number of different disorders.
The present results should be considered with the procedural limitations in mind. The data are self-reported data, so one cannot verify the veracity of the responses. However, this limitation is somewhat counteracted by the fact that several aspects of the present results replicate those previous published in the research literature. The present procedure also did not take any measure of level of nicotine dependence in the respondents. If it had, different relationships between smoking and gambling may have emerged. The present data were also collected from participants within the United States. There is no guarantee that the same relationships would be found in smokers in different countries.

References


























Track Your Manuscript

Recommended Journals