Stacked Deck: An Effective, School-based Program for the Prevention of Problem Gambling

Robert J. Williams; Robert T. Wood; Shawn R. Currie

J Prim Prev. 2010;31(3):109-125.

Discussion

As expected, and consistent with previous research, the Stacked Deck program produced significant and sustained changes in gambling attitudes, knowledge, and fallacies. At follow-up, students in the intervention schools demonstrated significantly more negative attitudes toward gambling, greater knowledge of both gambling and problem gambling, and greater resistance to gambling fallacies. This is an important result, indicating that the content of the Stacked Deck program was appropriate and delivered in a fashion that allowed for retention of this material. It is also likely the case that changes in these attributes are preconditions for actual changes in gambling behavior.

In addition, the present program also produced significant improvements in applied decision making and problem solving, a decrease in the percentage of gamblers, decreased overall gambling frequency, and, most importantly, some evidence of decreased rates of problem gambling (i.e., significantly lower rates of problem gambling at follow-up compared to the control group).³ To our knowledge, this is the first time that a school-based problem gambling prevention program has produced actual behavioral changes.

An argument can be made that decreased gambling participation may not be an appropriate goal of problem gambling prevention, when gambling is a normative activity in western society as well as a non-problematic activity for the large majority of people who engage in it (including high school students). However, as seen in the present study, gambling is not a normative activity among adolescents (only 41% of students gambled at baseline), and some types of gambling they report engaging in are illegal for their age group. Second, a decrease in gambling may well be appropriate outcome when considering that our primary goal was for students to decrease any behavior that can be construed as a bad gamble (i.e., when the odds are not in your favor and when the advantages of engaging in the behavior are less than the disadvantages). Finally, a decrease in the rate of problem gambling is definitely an appropriate measure of effectiveness, and it is hard to imagine this occurring without a concomitant decrease in overall gambling involvement.

The program was also found to be equally effective for all ages, genders, grades, schools, and communities. Furthermore, it also appears to be useful for students with high baseline levels of gambling frequency and/or symptoms of problem gambling. This is partly due to regression to the mean as well as these individuals having more "room for improvement." However, what this also speaks to is the instability of gambling and problem gambling in young people. As seen in Table 3, only a minority of problem gamblers at baseline were still problem gamblers at follow-up, including problem gamblers in the control group who received no intervention at all.

Other variables related to an enhanced outcome were more positive attitudes toward gambling at baseline, later date of program administration, higher baseline involvement in high-risk activities, and which trainer delivered the program. A more positive attitude is related to better outcome likely because the program had an impact on changing these attitudes, which presumably led to decreases in gambling behavior. Receiving the Stacked Deck program later in the study probably reflects the trainers' greater fluency in administering the program with greater practice. A lower rate of involvement in high-risk activities is related to better outcome perhaps because it reflects a less entrenched pattern of behavior more amenable to change. Finally, it comes as no surprise that there should be individual variability in the effectiveness of different trainers.

There are several important differences with previous programs that may explain why behavioral change was obtained in this study and not in others.

The focus on improving decision making and problem solving was somewhat different, as was the reported improvement in these skills. However, the failure of this improvement in decision making and problem solving to decrease other high-risk behaviors (e.g., substance use) makes it uncertain about their role in decreasing gambling behavior.
The orientation of the Stacked Deck program in advocating smart gambling/risk taking is different.
Targeting entire cohorts of students so as to include most of their peers was unique to the present study.
The average age of the students (16) is older than in other studies. Other problem gambling prevention programs have modeled themselves after tobacco and illicit drug use programs where the intervention is delivered to elementary school students who have not yet engaged in the behavior that is the target of prevention. This makes sense for prevention of tobacco and illicit drug use where total noninvolvement is the goal. However, noninvolvement in gambling is not a realistic (or appropriate) goal. Furthermore, some of the important concepts (e.g., odds, probabilities, independence of random events, law of large numbers) require a degree of mathematical and intellectual sophistication that may be beyond the grasp of many elementary school students. A stronger case can be made for intervening prior to the typical onset of problem gambling, which, admittedly, does appear to be present in high school students. However, as seen earlier, problem gambling does not appear to be a very stable or well-formed entity in this age group. Furthermore, the ability of the Stacked Deck program to change the behavior of heavily involved gamblers is evidence of its utility in this age group.⁴
Perhaps the most important difference from previous studies is the much heavier emphasis on the development and retention of skills, accomplished by making the program much longer (up to 600 min over 6 sessions), by spaced administration of lessons, and by its interactive and skill-oriented content. The superiority of the Booster Program over the Standard Program in some areas supports the notion that length and spacing is a contributing factor to the program's effectiveness. However, the comprehensiveness of the program is also likely important, as the authors have implemented other, even more substantive prevention initiatives in focused areas (i.e., mathematics of gambling) to university students that have failed to produce behavioral change (Williams and Connolly 2006). The authors appreciate the difficulty in incorporating multisession programs into already tight high school curriculums. However, it is important to recognize that the limited effectiveness of most current problem gambling educational and policy initiatives have to do with the fact that the ones that are implemented tend to be those that cause the least inconvenience, and consequently, have the least actual impact (Williams et al. 2007, 2008). We believe that problem gambling prevention needs to aspire to avoid the situation found in the substance abuse area, where the most commonly used (and entrenched) prevention and treatment interventions tend to be the less effective ones (e.g., Miller et al. 2003). Efficiencies may be obtained if the Stacked Deck program were administered in conjunction with prevention modules for tobacco, illicit drugs, and other substances because of potentially common content with respect to improving decision making, social problem solving, and coping skills.

1 2 3 4 5

Next Section

J Prim Prev. 2010;31(3):109-125. © 2010 Springer
Springer Science+Business Media

Cite this: Stacked Deck: An Effective, School-based Program for the Prevention of Problem Gambling - Medscape - May 01, 2010.

³ The failure to find unambiguous decreases in problem gambling from baseline to follow-up is likely due to low statistical power. The failure to find significant decreases in gambling monetary losses is partly due to the low average amounts being wagered plus the high variability in these amounts.
⁴ The ascertainment of problem gambling in adolescents is an under researched and controversial area (Derevensky et al. 2003). There is good evidence that certain instruments have produced elevated prevalence rates due to measurement error (i.e., South Oaks Gambling Scale and South Oaks Gambling Scale Revised for Adolescents; Ladouceur et al. 2000). The possibility also exists that the instability in problem gambling seen in the present study reflects a measurement problem. A team of researchers under the aegis of the Canadian Centre on Substance Abuse has been working on the development of a new, well validated instrument for the past several years.

Table 1. Sample sizes and follow-up rates for each group
Group	Baseline	Follow-up
Standard program	911	682 (74.9%)
Booster program	342	267 (78.1%)
Control group	433	291 (67.2%)
Total	1686	1240 (73.5%)

Table 2. Changes in dependent variables from baseline to follow-up in the three groups
Dependent variable (Scale)	Groups	Baseline M (SD)	Follow-up M (SD)
Gambling attitudes (−4 to +4)	Control group	−.33 (1.8)	−.52 (1.7)
	Standard program	−.24 (1.7)	−1.06 (1.7),*
	Booster program	−.38 (1.7)	−1.37 (1.7),**
Gambling knowledge (0–10)	Control group	4.66 (1.4)	4.84 (1.6)
	Standard program	4.72 (1.6)	5.73 (1.6),*
	Booster program	4.94 (1.4)	6.39 (1.7),**
Resistance to gambling fallacies (0–10)	Control group	5.01 (1.7)	4.89 (1.8)
	Standard program	4.95 (1.8)	5.94 (2.1),*
	Booster program	4.86 (1.7)	6.19 (2.0),*
Decision making & problem solving (past 3 months; 0–8)	Control group	5.83 (1.6)	5.71 (1.8)
	Standard program	5.74 (1.8)	6.07 (1.7),*
	Booster program	5.77 (1.7)	6.06 (1.8),*
High-risk activities (past 3 months; 0–9)	Control group	1.68 (1.7)	1.64 (1.6)
	Standard program	1.59 (1.5)	1.63 (1.7)
	Booster program	1.79 (1.4)	1.83 (1.6)
Gamblers (past 3 months)	Control group	46.4%	43.0%
	Standard program	39.7%	27.1%,*
	Booster program	41.9%	23.2%,*
Gambling frequency (all participants; # days gambling in past 90)	Control group	15.87 (44.3)	16.78 (35.5)
	Standard program	13.88 (41.0)	6.91 (23.9),*
	Booster program	13.27 (39.4)	6.12 (21.1),*
Money lost gambling (all participants; past 3 months)	Control group	$5.88 (21.2)	$5.18 (19.9)
	Standard program	$5.66 (22.3)	$3.52 (18.9)
	Booster program	$7.77 (28.8)	$3.01 (17.7)
Problem gamblers DSM-IV-MR-J (past 12 months)	Control group	3.0% (6/197)	5.1% (10/197)
	Standard program	3.5% (16/462)	3.0% (14/462)
	Booster program	3.1% (6/193)	1.6% (3/193)**
Problem gamblers self-reported (past 12 months)	Control group	4.8% (14/291)	7.2% (21/291)
	Standard program	5.1% (35/682)	4.0% (27/682)**
	Booster program	5.2% (14/267)	2.6% (7/267)**

Note: In the majority of cases the actual p-value is considerably less than .05
* significant change (p < .05) from baseline
** significantly different (p < .05) from Control Group at follow-up
*** significantly different (p < .05) from Control Group and Standard Group at follow-up

Table 3. Changes in problem gambling (PG) and non-problem gambling (NPG) status from baseline to follow-up in the three groups
Group	DSM-IV-MR-J criteria			Self-reported problem gambling criteria
	Baseline	Follow-up		Baseline	Follow-up
	Baseline	NPG	PG	Baseline	NPG	PG
Control group	NPG = 191	183	8	NPG = 277	259	18
	PG =6	4	2	PG = 14	11	3
Standard program	NPG = 446	435	11	NPG = 647	627	20
	PG = 16	13	3	PG = 35	28	7
Booster program	NPG = 187	186	1	NPG = 253	248	5
	PG =6	4	2	PG = 14	12	2

Table 4. Stepwise multiple regressions of variables predicting decreases in gambling behavior in the intervention groups from baseline to follow-up
Independent variable	Regression coefficients (B)	Standardized regression coefficients (β)	Adjusted R ²	Step
Change in status from gambler to non-gambler
Baseline frequency of gambling	.003	.235	.061	1
Baseline gambling attitudes	.026	.077	.066	2
Decreased gambling frequency
Baseline frequency of gambling	.833	.839	.686	1
Baseline gambling attitudes	1.314	.057	.689	2
Decreased DSM-IV-MR-J scores
Baseline DSM-IV-MR-J score	.992	.678	.186	1
Baseline frequency of gambling	.011	.252	.218	2
Date of program administration	.000001	.133	.244	3
Baseline involvement in high-risk activities	−.166	−.152	.264	4
Trainer A	−.767	−.117	.274	5

Comments

Commenting is limited to medical professionals. To comment please Log-in.

Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.