During 1989, the federal government initiated basic alcohol and substance abuse education requirements. Previously, there was not a regulatory mandate obligating institutions of higher learning to set alcohol or drug use policy or bring students' attention to these rules if they existed. Schools also were not required to disseminate substance use policy information to parents or other interested parties. This situation changed with the passage of the federal Drug-Free Schools and Communities Act (DFSCA) of 1989.

The DFSCA applies to all U.S. colleges. The act specifies that "as a condition of receiving funds or any other form of financial assistance under any Federal program, an institution of higher education (IHE) must certify that it has adopted and implemented a drug [and alcohol] prevention program..."8. Thus, any U.S. college that does not maintain a drug and alcohol education program risks losing all of its federal funding. In addition, to fulfil DFSCA requirements and retain funding, schools must provide students with institutional standards of conduct that explicitly prohibit illicit drugs and illegal alcohol use, a description of potential legal and institutional sanctions for substance use violations, a description of health risks posed by drugs and alcohol, and a listing of available treatment options.

The overall impact of mandated drug and alcohol programs is still unknown; as we noted before, there is some evidence that risky and addictive behaviors on college campuses are still prevalent despite targeted efforts by administrators to reduce student substance abuse 96. Several studies have suggested that, despite prevention efforts, established norms of excessive drinking behavior and positive student attitudes regarding the effects of alcohol consumption continue to encourage alcohol consumption on college campuses 1011. The absence of universal standards governing the content of school policies on addiction might contribute to this problem. Although the DFSCA mandates that schools must make written drug and alcohol policy available to students on an annual basis, administrators at each institution still determine the content of such policy. Thus, the DFSCA mandates policy without establishing standards for content; as a result, administrative tolerance toward alcohol, drugs, and gambling can vary significantly from institution to institution.

Inconsistent policy content among institutions can create a problematic state of affairs. Although DFSCA directives aim to increase awareness of the potential dangers of alcohol and drug use among students, numerous studies continue to identify high levels of alcohol abuse on U.S. college campuses in recent years 91213146. Heavy episodic drinking adversely affects not only those students who actively participate, but also those who do not: one study identified non-heavy drinkers on heavy drinking campuses as 3.6 times as likely to experience at least one problem from another student's drinking as non-heavy drinkers on non-heavy drinking campuses 15.

Even though individual colleges have adopted different strategies for reducing the problems associated with excessive alcohol consumption, the extent and effect of these efforts are largely unknown. One approach, perhaps in response to DFSCA, has been to develop and enforce policies on student substance abuse and recovery. Although recent psychosocial programs attempting to reduce student drinking behaviors have failed to reduce binge drinking 6, official school policies on substance abuse and recovery hold the potential to reduce students' alcohol use and the multitude of consequential problems associated with drinking excessively. This potential, however, is likely contingent upon policy content: because there are few federal regulations governing the content of alcohol policies at institutions of higher learning, every college develops unique strategies of combating potentially addictive behaviors. To date, no studies have examined the policy content of a representative sample of colleges in the attempt to identify the effects of these policies on levels of alcohol and gambling involvement among students.

Students who seek help for alcohol or other substance use problems are faced with a multitude of school-provided and external treatment options. Addiction recovery programs are diverse, ranging from formal treatment programs (e.g., inpatient medical treatment and outpatient psychotherapy) to less formal self-help options, (e.g., 12-step fellowships) 16. Regardless of the selected type of treatment, attention to recovery from addiction requires significant time and determination, which can disrupt a schedule of college studies. Twelve-step programs, for example, usually involve attending regular, perhaps even daily meetings. Formal treatment programs frequently demand an even greater level of time commitment: in-patient detoxification or other residential care can remove students from the academic environment altogether. Mandatory abstinence, required by most treatment programs, poses an additional hurdle to treatment-seekers. Students, with their busy and often stressful schedules, undoubtedly face additional challenges in participating in recovery activities; academic and administrative policies that accommodate flexible scheduling will likely assist students seeking recovery, and policies that do not might complicate or inhibit students' recovery efforts.

Binge drinking, the consumption of five or more alcoholic drinks (four or more for women) on at least one occasion at one to two week intervals 12, has been unaffected by prohibitive and punitive college policies. To illustrate, on one college campus that prohibited all alcohol use in its residence halls, there was virtually no difference in the binge drinking rate among students living within areas regulated by the alcohol policy (35%) compared to those living outside the jurisdiction of the alcohol policy (34%) 17. Although school policy (or the lack thereof) is not the only factor that affects binge drinking rates – promotions aimed at students, cheap alcohol prices at surrounding establishments and high numbers of on- and off-campus drinking venues have been found to significantly increase student binge drinking 18 – placing special emphasis on the enforcement of substance abuse policies can garner positive results. For example, Knight (2003)19 found that, although the effect of policy was diluted by considerable variation in policy content among public colleges in a state-wide system, increased enforcement (i.e., application of policy consequences) of alcohol policies aimed at combating underage drinking did result in decreased alcohol consumption among students. Beneficial effects resulting from the enforcement of existing rules, however, can be difficult to interpret. For example, in that study, it is unclear whether the enforcement of rules encouraged lower levels of drinking or entry to treatment for intemperate drinking or, alternatively, simply forced problematic drinkers to withdraw from school.

Some research suggests that gambling on college campuses is commonplace. A study of student gambling at six colleges in five different states (i.e., New York, New Jersey, Nevada, Oklahoma, and Texas) showed that of 1,771 surveyed students, 23% reported that they gambled at least weekly (ranging from 11% in Texas to 39% in Nevada) 20. In that study, students reported whether they had ever experienced gambling-related problems as identified by the South Oaks Gambling Screen (SOGS) 21. Of the total student sample, 5.5% were classified as lifetime probable pathological gamblers. The prevalence of lifetime pathological gamblers among these students ranged from 4% in Nevada to 8% in New York. A recent report 22 of a four-campus Connecticut college system reported a similar SOGS-based prevalence estimate of probable pathological gamblers (i.e., 5.2%).

For comparison, the National Gambling Impact Study Commission (NGISC) considered the adult rates of lifetime pathological gambling from four sources 3. The lowest rates were 0.8% for both the University of Michigan 23 and National Opinion Research Center 24 studies; the largest (i.e., 1.5% – 1.6%) were from aggregated statistics of previously published research conducted by the National Research Council 25 and the original analysis of the same studies by the Harvard Medical School 26. This meta-analysis included 14 SOGS-based studies of disordered gambling among college students and indicated that the lifetime prevalence of pathological gambling among college students was 5.1% 26. An update of this meta-analysis expanded the number of student studies to 19 and increased the prevalence estimate to 5.6% with a 95% confidence interval of 3.5% to 7.6% 1. Based on 66 studies of the general household population in various areas (i.e., states), this estimate of the proportion of college students with gambling disorders was three times the adult rate (1.9%).

Other research contradicts the findings that college students are at elevated risk for problem gambling compared to the general adult population. For example, a recently published longitudinal study of students at the University of Missouri-Columbia showed markedly lower prevalence rates than the studies summarized above 27. In this longitudinal study, no student met the traditional criteria for problem or pathological gambling. Further, the authors note that, "there were too few participants endorsing multiple gambling problems at a single time point to obtain an adequate sample size of affected individuals for most analyses" (Slutske et al. 200327 p. 265). Overall, 3% of these students endorsed a single problem at any point during their lifetime due to gambling; one student endorsed two problems and all of the others reported never having had a problem due to gambling. At the next interview three years later, when most subjects were seniors, the subjects reported more symptoms; but only one subject (i.e., 0.2% of the sample) endorsed enough symptoms to meet the diagnostic criteria of the American Psychiatric Association 28 for lifetime pathological gambling. This evidence indicates that gambling behavior among students and its adverse consequences fluctuates with time and other factors and that the development of symptoms is not always progressive. Further, the Slutske results show that most adverse effects of student gambling remain sub-clinical, making this pattern more responsive to interventions than longer standing, more entrenched clinical disorders. Taken together, this evidence suggests that comprehensive college gambling policies might have the capacity to reduce the adverse consequences that can be associated with student gambling.

Despite the frequency with which college students engage in gambling activities, some evidence suggests that administrators are unaware of the dangers associated with excessive gambling among students; in addition, colleges do not have adequate policies addressing gambling 29. This situation prompted Shaffer to suggest that the government convene "a consortium of college presidents to review their existing gambling related policies and problems so that we can take a systematic approach to the education, prevention and treatment of America's young people, who are at higher risk for gambling related disorders than their adult counterparts"30. Although this consortium has not yet been assembled, research confirms that college students continue to view gambling as a legitimate form of entertainment; for example, 42% of a scientifically selected sample having gambled at least once in the last year 31. Unlike drug and alcohol education (i.e., DFSCA), there is no federal mandate requiring schools to educate students or parents about the dangers of excessive gambling; combined with the lack of a policy response by administrators, this situation leaves an open door for student-related gambling disorders to emerge unchecked.

This study is the first to identify patterns of drinking and gambling-related rulemaking on college campuses (e.g., punitive versus recovery oriented). By relating school policies to known school rates of drinking or gambling 316 we can identify potentially influential policies. These analyses can encourage and inform the development of guidelines, or "best practices," upon which schools can base future policy.

Given the paucity of empirical college-based policy research, this study will fill an important gap in knowledge. To fill this void, this research will test a variety of addiction-related hypotheses that have not yet been examined empirically. Based upon the extant literature, this study will test the following four primary hypotheses:

• Because there are few requirements guiding the creation of school substance use and gambling policies, the content and clarity of these policies will be heterogeneous across schools and modes of policy distribution (e.g., handbooks vs. school Web sites);

• College alcohol policies currently devote relatively little attention to student recovery;

• Due to differences in enforcement, awareness of the dangers of excessive alcohol consumption, educational programs and types of students, schools with either no or only restrictive alcohol use policies will experience higher levels of binge drinking among students than schools with prohibitive and recovery-oriented alcohol policies;

• Absent a federal mandate that requires gambling-related regulations or education on college campuses, gambling policies will be less prevalent than alcohol use policies.

The purpose of this study is to identify and assess alcohol and gambling policies among U.S. colleges. To ensure a representative national sample of colleges, we examined the scientifically selected sample of public and private American colleges that was used in a recent series of Harvard studies (e.g., Wechsler 2002 6). The detailed methods by which the previous study identified the sample are available elsewhere 612131431. The potential sample consisted of 120 scientifically selected schools located throughout the nation; one school ceased operation before the start of the study, so 119 schools were eligible to be included in the final sample. We received human subjects approval for this study through the Harvard Medical School Office for Research Subject Protection. On February 14, 2003, the Human Subjects Committee at Harvard Medical School granted an exemption for the study entitled: United States College and University Addiction and Recovery Policies. The study qualified for exemption under 46 CFR §102(f) and the assurance identification number is M1240-01.

At the beginning of the project, we submitted an e-mail request for a hard copy of their student handbook to each school's admissions office. Each e-mail specified that we were interested in collecting school alcohol and gambling policies and requested that our inquiry be forwarded to the most appropriate school official. We gathered e-mail addresses for admissions offices from each school's official Web site. Using each school's main telephone number to initiate contact, investigators contacted schools that did not respond within thirty days to our e-mail request and verbally requested a handbook and any other existing alcohol and gambling policy materials. Typically, the person answering the call referred us to admissions offices, deans' offices, or student services offices for further assistance; we identified ourselves as calling from Harvard Medical School only when asked.

Investigators developed a coding instrument by studying alcohol and gambling policies from a variety of U.S. schools outside the current sample and identifying the underlying characteristics of the policies. These characteristics were reduced to 40 items that reflected the scope and focus of school alcohol and gambling policies. The items were converted into a coding instrument that included 25 variables for alcohol policy and 15 variables for gambling policy. This instrument included items to measure the presence of specific policies and establish whether the policies were punitive or rehabilitative. All variables used a nominal scale that included common characteristics of each school policy; response choices varied slightly with the focus of each variable. All of the variables were arranged on a six-page coding form.

To simplify coding and allow for within-school comparisons between different formats of policy dissemination (e.g. school handbook vs. school Web site), we separated each school's policy materials into three categories: (1) student handbooks (electronic or paper); (2) Web-based materials; and (3) supplementary materials (paper); a potential 357 documents required coding (three coding categories for each of 119 schools in sample = 357 potential documents). However, because not every school had documents available in all three coding categories, the final document count was 164. Specifically, at the end of our data collection process, we had collected 73 student handbooks, 70 Web-based policies, and 21 supplementary documents.

We assigned 11 coders the job of evaluating each school's alcohol and gambling policies. Each coder read a selection of policies and extracted relevant information in accordance with the coding form. The coding process proceeded as follows:

1. Each policy document was assigned to two of eleven eligible DOA coders randomly. Each assigned coder independently abstracted information from each assigned policy document and recorded this information on separate coding forms.

2. For each document, one member of the research team, designated as the "arbiter," compared the two coding forms and marked discrepant items.

3. The arbiter returned the marked coding forms to their respective coders and requested that coders reconsider their answers to the items in question. Upon reconsideration, coders were free to change their answers or keep their original answers.

4. Coders resubmitted their recoded documents to the arbiter who compared the discrepant items again. Discrepancies that remained were noted and resolved by the arbiter.

5. Once all discrepancies had been resolved, the policy assessments on the coding forms were entered into an SPSS database using a procedure that screened entries for out-of-range values and discrepancies in branching among items.

6. We assessed data entry reliability by selecting 10% of the cases in our database and rechecking each data entry point. Of the 680 items entered in these 17 randomly selected cases, there were no observed data entry errors.

Shaffer and his associates have used a similar process of information extraction, coding and arbitration successfully in other published studies 132.

We assessed inter-coder reliability by comparing the total number of discrepant coded items to the total number of coded items. As described previously, each policy was assigned to two of eleven eligible DOA coders randomly. The participation of eleven coders yielded 55 possible coding-pair combinations; each of these pairings coded at least one policy. Specifically, the number of policies coded by each coder-pair ranged from a minimum of one (n = 6) to a maximum of six (n = 3). Coders had up to two opportunities to code each document: (a) an initial round of coding; and (b) a second round of coding to reconsider any discrepant items identified by the arbiter after the initial round of coding. The arbiter made the final coding decision on 345 out of a total of 4,100 possible items. The coding process yielded a study-wide inter-coder reliability rate of 91.6%.

After thirty days had passed from our initial e-mail request, 46 of 119 schools had responded by sending hard copy materials. Eighteen of these 46 colleges sent materials completely unrelated to our request for school alcohol policies (e.g., applications for admission, school newsletters). Fourteen schools sent student handbooks, and another 14 schools sent other alcohol and/or gambling related (i.e., non-handbook) materials. Seventy-three schools did not respond to our request within thirty days. Subsequent to our follow-up telephone requests, we received student handbooks and supplemental materials from an additional 22 schools. This recruitment procedure resulted in 50 schools actively providing policy information for this study; for the remainder, policy information was obtained through other investigative procedures as described earlier (e.g., Web sites).

This study sought information on alcohol and gambling policy from a representative sample of 119 colleges across the U.S. We utilized three distinct common sources of information on school alcohol policy: student handbooks, school Web sites (non-handbook related) and supplementary materials (e.g., policy manuals, pamphlets). We collected a total of 164 policy-related documents from three sources: 73 policy documents from handbooks, 70 documents from school Web sites, and 21 from supplementary materials. Table 1 presents the sources of alcohol policy information for the schools in our sample. Forty schools presented their full alcohol policy in their handbook, 31 on their Web site, 2 in supplementary materials, and 44 through a combination of handbook, Web site and supplements. We were unable to locate any policy information for two schools in our sample; these schools did not respond to our requests for information.

We aggregated and analyzed policy information across sources because a preliminary examination revealed content differences among handbooks, Web sites 33 and supplementary materials. Aggregating information across sources provides the most extensive view of each college's policy strategy because it considers all modes of policy distribution. This strategy yields the most comprehensive policy search and identifies more policy mentions than is possible by examining only one policy source. To implement this strategy, we first constructed a new database that included data for schools with a handbook, a Web site or both (n = 115). Next we created a single record of policy mentions for the 28 schools with both handbooks and Web materials by aggregating policies across sources. This database assimilated the unique handbook and Web variables into a single set of "recompiled" variables, reflecting the total number of policies attributable to either the handbook or the Web. To compare the "added value" of school Web sites (i.e., policy information presented on the Web that was not presented in the handbook), we summed the policies reflected by the recompiled variables and then subtracted the policies contained in the handbook-only variables. Of 263 total policy items present, we collected 198 (75%) policy items from student handbooks and 65 (25%) additional policy items from school Web sites that were not available in handbooks.

To determine the added contribution of supplemental materials (i.e., policy information presented in the supplements that was unavailable elsewhere), we created another set of recompiled variables following the previously outlined procedure. These variables reflected the total number of policies identified for the three schools with all three types of sources (i.e., handbooks, Web materials and supplements). We summed the policies reflected by the recompiled variables and then subtracted the policies contributed by handbooks and Web sources; this procedure revealed that supplemental materials contributed 4 of 30 (13.3%) policy items.

Although school Web sites provide a substantial amount of alcohol policy information that is not contained in the primary document customarily provided to students (i.e., handbook), the overall added contribution of the school Web site in presenting policy information varied among schools. For example, one school's Web site contained an additional eight alcohol regulations that were not included in the handbook; however, several schools' sites contained no additional information. In addition, the type of information that was presented only on Web sites also varied: while most information pertained mainly to secondary alcohol policies (e.g., school-sponsored events and drinking regulations for drinking-aged students), some schools chose to present vital alcohol policy information (e.g., stating that all drinking is prohibited for students <21) on their Web site only (n = 2). Thus, although handbooks and Web sites are both important sources of alcohol policy information and supplements contribute little additional information, consistency across sources varies. The following analyses assess the agreement of information found in multiple sources.

As mentioned earlier, of the 117 colleges for which we had information, all 117 (100%) had a written policy on student alcohol consumption and 26 schools (22%) had a student gambling policy. Because all schools had a written alcohol policy (and relatively few schools had a gambling policy), the following analyses focus on alcohol policies.

Determining concordance between handbook and Web sources is important because administrators might be unaware of inconsistencies between official school documents. In addition, contradictory information can mislead students and potential applicants. We assessed the concordance between sources of college alcohol policy materials by determining the level of agreement (i.e., presence or absence of policy information) between handbooks and Web materials; that is, we compared the content of each type of document to identify differences in the presentation of each school's policy information between sources. We did not extend this particular analysis to include supplemental materials because, as we noted before, only a small number of schools (n = 3) had all three types of sources.

Twenty-eight schools had both a handbook and Web materials; for each of these 28 schools we determined the absence or presence of the 25 alcohol policy variables in each source. We predetermined that a concordance rate of 85% would indicate a high level of agreement between documents. To be considered in agreement, complementary information had to be found in (or absent from) both sources; in cases where this requirement was not satisfied, the policies were considered in disagreement. Using these criteria, we determined that three policy variables (i.e., 12% of the policy variables) were mentioned often and were present in both handbooks and on Web sites, and consequently, showed high agreement. Either type of information resource seldom mentioned ten policy variables (i.e., 40% of the policy variables), therefore, also exhibiting high agreement. The remaining 12 policy variables (i.e., 48% of the policy variables) were often mentioned, but not consistently by both sources, indicating low agreement.

Table 2 presents the three "high agreement" alcohol policies that were mentioned consistently in both handbooks and Web materials. Variables that fell into this category generally measured broad school policies (i.e., the existence of an alcohol policy). As Table 2 illustrates, schools consistently made these types of alcohol policies available to the public in both print and electronic form, making this information highly accessible.

Policy variables that were rarely mentioned in handbooks and Web materials appear in Table 3. These variables primarily measured on- and off- campus alcohol consumption restrictions and school recovery polices regarding student alcohol use. These policies are in "high agreement," because they were seldom mentioned: as Table 3 demonstrates, this information was missing from handbooks and Web sites in nearly all cases.

Table 4 presents variables that were mentioned occasionally (i.e., concordance <85%) in handbooks or Web materials. The policies in this category primarily address consumption and event restrictions and student recovery. Table 4 illustrates that we observed considerable inconsistencies in schools' methods of distribution of these types of policies.

As noted earlier, the coding process revealed that all 117 colleges (i.e., 100% of the schools for which information was available) in this sample had a written policy on student alcohol consumption, but only 26 (22%) had a published policy that addressed gambling. The small number of schools with gambling policies precludes confident analysis of the dimensional composition of our gambling variables; therefore, we applied the factor analysis that follows only to alcohol policy variables.

Three policy variables represented a multi-dimensional measurement strategy to yield detailed policy information. Consequently, we collapsed these three redundant policy items into the primary or gate items from which they originated (e.g., "alcohol is prohibited on-campus for students ≥ 21" and "on-campus alcohol restrictions in place for students ≥ 21" became "policy on alcohol use on-campus for students ≥ 21). This resulted in 22 alcohol policy variables in all remaining analyses. These dependent variables all measured different aspects of school alcohol policies (e.g., policy presence, content, and target). To empirically examine the underlying dimensions reflected by our variables, we conducted an exploratory factor analysis. This procedure employed an initial factor extraction (i.e., component matrix) and then an orthogonal rotation to simple structure. We selected the Varimax rotation to maximize the variance of loadings within factors and minimize the covariance across factors. The orthogonal solution identified eight policy clusters with Eigenvalues greater than 1.0 that explained 72.36% of the total variation. This explained variance lies within the 50–75% useful range suggested by Overall and Klett (1972)34. Consequently, we concluded that our factor analysis provided a valid identification of the policy clusters that underlie college alcohol and gambling regulations.

Table 5 presents the structure of the interrelationships among policies. To facilitate interpretation, the table reports only factor loadings ≥ 0.50 (i.e., policies with loadings in this range correlate .50 or greater with a composite measure of the overall dimension).

College alcohol policies varied widely. Table 6 summarizes the prevalence of alcohol-related policy and the mean prevalence of alcohol policies within each factor. The prevalence of alcohol policies ranged from 100% (i.e., the presence of an alcohol use policy) to 1.7% (i.e., policy that permits a leave of absence to participate in a recovery program). The mean prevalence for the eight policy factors ranged from 92.3% (i.e., Prohibition policies) to 5.4% (i.e., Limits and restrictions-off-campus).

We conducted several analyses to determine the nature of relationships between student alcohol consumption, gambling behavior and policy content. As the factor analysis above illustrates and the relative prevalence of policies confirms, college alcohol-related policies are primarily intended to prevent, reduce or restrict alcohol use among students on college campuses. To test the relationships between alcohol policies and student drinking behaviour, we compared the mean binge drinking rates of students at schools with and without each policy variable. We obtained the mean binge drinking rates of the schools in our sample from the dataset used in Wechsler et al.'s Harvard School of Public Health College Alcohol Study (CAS)6. Because this is one of the first studies of college policies, we sought to identify as many potential relationships between policy, drinking and gambling as possible; therefore we set a liberal alpha level (α = .1) for this analysis. A one-way analysis of variance (ANOVA) revealed that four of the 22 policy variables had significant relationships with binge drinking rates at the colleges in our sample (see Table 7).

Schools that had either no policy restrictions or a prohibition policy for on-campus alcohol use by students ≥ 21 had lower mean past-month student binge drinking rates (39% and 36%, respectively) compared to schools that employed an intermediate level of restrictive policies (47%) (F = 7.07, df = 2,113, p < .001). Schools that allowed or did not mention alcohol use at on-campus sanctioned events for students ≥ 21 had a higher mean binge drinking rate (46% and 44%, respectively) than schools that prohibited legal drinking at events (33%) (F = 5.25, df = 2,113, p < .001). Schools that offered alcohol-free housing had a higher mean student binge drinking rate of 49% compared to schools that did not mention alcohol-free housing which had a binge rate of 42% (F = 4.01, df = 1,114, p < .01). Two other alcohol policy variables evidenced significant relationships with student binge drinking behaviors, but these policies lacked widespread implementation at a large number of schools. Schools that specifically allowed a leave of absence for a student to participate in recovery activities (n = 2) evidenced a higher mean binge drinking rate (69%) than schools that did not mention such a policy (43%, n = 114). Keeping the small number of schools in mind, it is worth noting that schools that prohibited off-campus alcohol consumption for students ≥ 21 (n = 4) had a lower mean binge rate (10%) than schools without this provision (45%, n = 112).

There were not enough schools with gambling policies to permit a detailed analysis of the relationship between policies and student gambling behavior; therefore instead of conducting an analysis of the relationship between gambling behavior and individual policy variables, we only were able to assess gambling behavior based on whether schools had a gambling policy. Using prevalence data from LaBrie et al.'s (2003) 31 recent study of student gambling behavior, we determined that no significant difference in mean past-year student gambling behavior existed between schools with a written policy on gambling (i.e., prohibitive or restrictive) and schools with no mention of gambling policy (i.e., approximately 40% regardless of policy presence).

Long ago, Pigou 35 noted that public policies can have unanticipated effects; policy intended to regulate one set of behaviors can influence other patterns of behavior. To test the relationships between alcohol policies and student gambling behavior, we compared the mean past-year gambling rate at schools with each alcohol policy to schools without the policy. As before, we used a liberal alpha (α = .1) to identify all potential relationships. A one-way analysis of variance (ANOVA) revealed that four of the 22 alcohol policy variables had significant relationships with the proportion of students who gambled in the past-year school year (see Table 8).

Schools that restricted or prohibited on-campus alcohol use for students over 21 evidenced similar mean past-year student gambling participation rates (i.e., 40% and 41%, respectively), and schools with no restrictive policy evidenced a higher student gambling participation rate (46%) (F = 2.64, df = 2,113, p < .08). Schools that did not mention or allowed alcohol at on-campus events for legal drinkers exhibited a higher mean gambling participation rate (i.e., 44% and 40%, respectively) than schools that prohibited alcohol at on-campus events (35%) (F = 5.17, df = 2,113, p < .001). Similarly, schools that did not limit the quantity of alcohol available at events showed higher past-year gambling participation among students (42%) compared to schools with no such provision (37%) (F = 5.39, df = 1,114, p < .001). Schools that had off-campus alcohol restrictions for legal-aged drinkers had a gambling rate of 38% while schools that did not had a mean gambling rate of 42% (F = 2.94, df = 1,114, p < .1).

Two other alcohol policy variables evidenced significant relationships with student gambling behaviors, but these policies were not widely implemented throughout the sample. Schools that did not expressly prohibit alcohol consumption for underage drinkers (n = 3) evidenced a mean past-year student gambling rate of 51%, while schools that prohibited underage drinking (n = 116) had a gambling rate of 41% (F = 3.19, df = 1,114, p < .1). Schools that banned all alcohol consumption, whether on- or off-campus (n = 4) evidenced a lower gambling rate (30%) than schools that allowed at least some drinking (41%, n = 112) (F = 5.80, df = 1,114, p < .05).

In addition to the direct relationships between policy variables and binge drinking and gambling, three alcohol policy variables evidenced unexpected interaction or intensification effects when gambling policies also were present. Schools that had both a policy prohibiting or restricting gambling activity among students and a policy prohibiting on-campus legal-aged drinking (n = 9) had a mean binge drinking rate of 29% – much lower than schools with just an alcohol policy (40%, n = 15), a gambling policy (47%, n = 17) or neither (45%, n = 75) (F = 3.15, df = 1,112, p < .1). Schools with both a gambling policy and a policy prohibiting alcohol at on-campus events (n = 6) had a significantly lower mean binge rate (22%) than schools with just an alcohol policy (40%, n = 9), a gambling policy (46%, n = 20) or neither (45%, n = 81) (F = 5.88, df = 1,112, p < .05). Schools that prohibited alcohol at on-campus events also evidenced significantly lower past-year student gambling rates than schools without such prohibitions: 30% versus 38% or higher (F = 8.63, df = 1,112, p < .05).