A Study on Prosocial Behaviors of Our Minds Matter Participants
Shane McCarty, PhD & Alyssa Gatto, ABD
Abstract
Youth-led change occurs when student leaders engage in prosocial behaviors that benefit their peers. Yet, few studies have explored the factors associated with youth leadership and diverse prosocial behaviors (dyadic and school-wide). The Our Minds Matter program aims to increase the frequency of prosocial behaviors by its club members and leaders. The aimed result is socialization of peers through dyadic prosocial behaviors and affecting the broader mental health culture at school through school-wide campaigns. The current study examined how peer influence and leader role (as a member or club leader) among 220 high school students in the Our Minds Matter program related to the frequency of prosocial behaviors directed toward peers at school and the support of school-wide campaigns to promote wellness and prevent stigma. Results showed peer prosocial influence is related to the amount of support for school-wide campaigns irrespective of leader role (member or leader). The amount of prosocial behaviors performed by club members did not differ between members and leaders. This study affirms the multi-dimensionality of prosocial behaviors among OMM club members and leaders, highlighting the need for program goals to specify the type of prosocial behavior (direct helping: promotive, preventive, responsive, and recovery-focused; indirect school-wide campaigns: wellness promotive, stigma preventive). Additionally, peer prosocial influence relates to support for school-wide campaigns. As a result, the OMM program could harness influencers to focus on school-wide campaign development while targeting other prosocial types for less influential students.
Introduction
Classification of prosocial behaviors varies across the research literature: as individual prosocial tendencies across time (Carlo & Randall, 2002), as prosocial responses to negative states of others (Dunfield, 2014), and as a multi-dimensional construct (e.g., planned vs. spontaneous, serious vs. nonserious, indirect vs. direct) (Pearce & Amato, 1980). For this project, prosocial actions were classified based on recipient: direct helping to an individual (dyadic prosocial behavior) vs. indirect helping (school-wide campaigns). The dyadic prosocial behaviors were categorized based on Promote Care & Prevent Harm’s prosocial continuum (PCPH, 2020), which was adapted from the Institute of Medicine’s Continuum of Care framework (Begun, 2019; IOM, 1994). Dyadic prosocial behaviors to benefit the mental health of peers consists of promotive prosocial behavior (to promote mental wellness for promotion), preventive prosocial behavior (to identify signs of mental health challenges for prevention), responsive prosocial behavior (to respond to a crisis for indicated prevention), and recovery prosocial behavior (to support a student during recovery after a crisis). Based on the care continuum, promotive and preventive prosocial behaviors are more proactive and similar than responsive and recovery-focused prosocial behaviors.
A host of factors are related to prosocial behaviors. During adolescence, peer influence and related constructs (e.g., peer status, well-liked, perceived popularity) are particularly important, spanning across the research literature on relationships (e.g., Prinstein, 2018), physical and relational aggression (e.g., Cillessen & Mayeux, 2004), prosocial goals and behavior (e.g., Rodkin, Ryan, Jamison, & Wilson, 2013). In fact, leveraging peer status to influence others is part of the theory of change for youth-led change programs to address mental health (e.g., Sources of Strength), dating violence (e.g., Dating Matters; Briggs et al., 2012), and others. The framework for youth-led social change (Ho, Clarke, Dougherty, 2015) identifies influence as one of the four primary strategies for change along with socialization, power, and partnership. Prior research suggests influence is related to support for youth-led prosocial campaigns. OMM youth leaders are critical to creating a positive group climate and influencing their members to develop school-wide campaigns that benefit the entire school.
OMM Program
The Our Minds Matter (OMM) program aims to empower high school students to change the mental health culture at school by promoting mental wellness and reducing stigma associated with mental health. OMM Leaders select a theme and activities to benefit the learning, engagement and mental health needs of OMM members during after-school club meetings (see OMM Activities). OMM leaders are supported by club sponsors, who provide structure and support to student leaders as they build a team culture and lead OMM members in strategies to benefit the school. To create school-wide impact, OMM leaders and members engage in direct dyadic prosocial behaviors to impact students at school and indirect school-wide prosocial behaviors to improve the school culture.
The Current Study
This research study explores whether peer prosocial influence and leadership role relate to prosocial behaviors for mental health and support of school-wide campaigns for mental health among OMM students. More specifically, these potential explanatory variables may relate to the type of prosocial behavior (promotive, preventive, responsive, and recovery) and the frequency of engaging in these behavioral types.
The explanatory variable of leadership role was measured based on a student’s self-reported role as an OMM leader or OMM member. The explanatory variable of peer influence was measured by students’ self-reported ability to influence their peers to participate in school-wide campaigns. The purpose of this study is to better understand how leader role and influence relate to dyadic prosocial behaviors and school-wide campaigns to promote mental wellness and reduce stigma.
Research Questions
- Do OMM leaders score higher on the frequency of prosocial behaviors compared to OMM members?
- Do OMM leaders support school-wide campaigns more than OMM members?
- Does peer prosocial influence relate to prosocial behaviors performed? Does the relationship between these two variables change based on leader role (OMM leader vs. OMM member)?
- Do students prefer their OMM club to focus on mental wellness promotion or stigma prevention?
Methods
Research Design
This cross-sectional research study is part of a larger predictive research study to evaluate the impact of the OMM program. These data are from the pre-test assessment associated with the OMM program evaluation for the 2019-2020 school year.
Participants
OMM club members were invited to participate in a survey on demographics and mental health factors from August 12th, 2019 to September 30th, 2019. Students (n = 220) from 28 high schools completed the survey and provided permission for the results to be used to improve the program and evaluate program effects.
Measures
Explanatory Variables
OMM Leader
Leader Role was assessed by asking: Are you a club leader? The response options were: yes or no. Respondents were labeled as “OMM Member” or “OMM Leader”.
Peer Prosocial Influence
Peer Prosocial Influence was assessed by adapting the MacArthur Scale of Subjective Social Status for peer prosocial influence at school. The prompt and question were: “Think of this ladder as representing where students stand in their school. At the top of the ladder (10) are the people who have the highest ability to influence their peers to participate in a school-wide campaign, event, or activity. At the bottom (1) are the people who have the lowest ability to influence their peers to participate in school-wide campaign, event, or activity.”
Outcome Variables
Support for School-Wide Prosocial Campaigns
Prosocial school-wide campaigns was measured on a five-point scale (never, rarely, occasionally, frequently, and always) using two items. Two items were aggregated to form a composite.
- Promotive Campaigns: I support school-wide campaigns that promote mental wellness.
- Preventive Campaigns: I support school-wide campaigns that reduce stigma associated with mental health and increase help-seeking.
Prosocial Behaviors
Prosocial behaviors were measured on a five-point scale (never, rarely, occasionally, frequently, and always). Four items were aggregated to form the prosocial composite.
- Promotive prosocial behavior: When I see friends who do something that helps their mental health, I ______ recognize the benefits and say something positive to my friend, such as “good job,” “I am glad you are doing that”.
- Preventive prosocial behavior: When I see friends who are struggling with mental health issues, I ______ notice the signs and say something before it is a bigger problem or crisis.
- Responsive prosocial behavior: When I know friends who are receiving counseling for mental health issues, I ______ support them as a friend.
- Recovery-focused prosocial behavior: When I know friends who are recovering from a serious mental health crisis, I _______ support them as a friend.
Exploratory Variables
Club Goal
Club Goal was assessed based on the following prompt and question: “Club members probably view mental health challenges and the associated problems and goals of the club differently. In your view, what primary goal should the OMM club focus on this year?” The response options were: (0) = Reduce stigma associated with mental health, (50) Both are equally important, and (100) Promote mental wellness.
Statistical Analyses
The data was collected using Survey Monkey, cleaned on Microsoft Excel for Mac v16.40, and analyzed using 4.0.2 and RStudio v1.3.1056. The HTML output was created using “rmarkdown”, “markdown”, and “knitr” packages in RStudio along with two applications: LaTeXiT for Mac and XQuartz. The purpose of using the aforementioned software was to create a fully transparent and reproducible research study. The data file (.xlsx), R script (.R), R notebook (.Rmd), and related documents are available at the OMM project GitHub.
## Error in tinytex::install_tinytex(): lazy-load database '/Library/Frameworks/R.framework/Versions/4.0/Resources/library/tinytex/R/tinytex.rdb' is corrupt
## Error: package or namespace load failed for 'sjPlot':
## object 'standardize_names' is not exported by 'namespace:insight'
Results
Participant Characteristics
The bar plot shows the counts for gender identity of the participants. Nearly 40% of the respondents did not choose to select a gender identity, which was marked as “NA” (not available). Of respondents selecting a gender identity, Cis Female participants accounted for 93% of the responses.
Explanatory Variables and Outcomes
| Mean | Std.Dev | Min | Median | Max | N.Valid | Pct.Valid | |
|---|---|---|---|---|---|---|---|
| Campaigns | 4.25 | 0.915 | 1 | 4.5 | 5 | 199 | 90.5 |
| Promotive.Campaigns | 4.24 | 0.940 | 1 | 4.5 | 5 | 200 | 90.9 |
| Preventive.Campaigns | 4.27 | 0.957 | 1 | 5.0 | 5 | 199 | 90.5 |
| Prosocial | 4.40 | 0.457 | 3 | 4.5 | 5 | 200 | 90.9 |
| Promotive.Prosocial | 4.03 | 0.910 | 1 | 4.0 | 5 | 200 | 90.9 |
| Preventive.Prosocial | 3.92 | 0.819 | 1 | 4.0 | 5 | 200 | 90.9 |
| Responsive.Prosocial | 4.84 | 0.406 | 3 | 5.0 | 5 | 200 | 90.9 |
| Recovery.Prosocial | 4.81 | 0.441 | 3 | 5.0 | 5 | 200 | 90.9 |
Correlations Among Prosocial Items
As shown in the correlation matrix below, promotive prosocial behavior and preventive prosocial behavior items were moderately correlated, r(198) = 0.394. The responsive prosocial behavior and recovery prosocial behavior are strongly correlated, r(198) = 0.782. However, the proactive prosocial items (i.e., promotive and preventive prosocial behavior) were least correlated with the reactive prosocial items (i.e., responsive and recovery-focused prosocial behavior), ranging from 0.162 to 0.276.
Cronbach’s alpha for Prosocial Items
As shown in the table below, the reliability of the prosocial composite is low (see raw_alpha = .59). Therefore, a single prosocial composite was not used for future analyses. Each prosocial indicator was used as an independent outcome.
## Error in psych::alpha(Prosocialdf): lazy-load database '/Library/Frameworks/R.framework/Versions/4.0/Resources/library/psych/R/psych.rdb' is corrupt
Correlation Between Campaign Items and Cronbach’s alpha
The two campaign items are highly correlated, r(197) = 0.858.
Correlations Between Peer Prosocial Influence and Outcomes
The correlation matrix below shows the correlation among prosocial items, campaign items, and influence. Influence has a weak correlation (r = 0.211 and 0.197) with the proactive prosocial items and no relationship with the reactive prosocial items (r = 0.057 and -0.021). Influence is moderately correlated with promotive campaigns (r = 0.308) and preventive campaigns (r = 0.269). The outcomes are regressed on influence in the forthcoming regression analyses to analyze these relationships with additional categorical variables.
Primary OMM Club Goal
Students identified the primary OMM club goal by selecting their preference for a promotion goal, prevention goal or both. Student responses range across the entire dimension from 0 to 100 with “0” denoting a full endorsement of the prevention goal (to reduce mental health stigma) and a score of “100” referring to a full endorsement of the promotion goal (to promote mental wellness).
The average student score for the Ideal OMM Club Goal was 54.31 and the median was 50(equal). Simply put, the average OMM club respondent identified the promotion goal of furthering mental wellness and the prevention goal of reducing stigma associated with mental health as equally important.
While the median and mean provide clarity on where the average student scores, the variation in responses provides insight on the range of responses from students to capture all of the individual’s preferences for the Primary OMM Club Goal. The bar plot below depicts the count across the prevention-promotion goal dimension. To better understand the spread, the scores were z-scored to obtain a normal distribution for the data. These responses were split into three categories: .67 SD below the mean, .67 SD above the mean, and the middle range between the two SD cut offs. Color was used to denote the percentage of respondents from the total sample in each the three groups: pink was used for the < -.67 SD group (25% of respondents with the lowest scores toward the prevention side of the dimension, the middle 50% of respondents are denoted in gray, and blue was used to denote the 25% of respondents with largest scores toward the promotion side of the dimension. While the median response was 50, the responses ranged from 0 (prevention goal) to 100 (promotion goal).
The plot above categorizes respondent scores based on arbitrary cut off scores of 33 and 66. Responses of 33 and below were color coded purple for prevention goal. Responses between 34 and 66 were marked in gray for balanced goals. Responses of 67 and above were marked in light blue to represent promotion goal.
Regression Analysis
Regression analyses were performed to answer the research questions related to prosocial behaviors: How does influence and leader role relate to prosocial behavior? Is the relationship different based on role as member or leader? To answer these questions, the type of prosocial behavior (e.g., promotive, preventive, responsive, recovery-focused) was regressed on influence, leader role, and the interaction term of influence by leader role.
Promotive Prosocial Behavior Regressed on Influence and Leader Role
Results of the multiple linear regression indicate there was no effect for the model, F(3,129) = 2.171, p = .09, R2 = .05). Simply, the amount of influence and role as member or leader does not relate to the amount of promotive prosocial behavior (see estimates, t-value and p value in the regression summary table). As shown in the second plot, OMM members and OMM leaders have similar flat slopes.
##
## Call:
## lm(formula = Promotive.Prosocial ~ Influence + Leader + Influence:Leader,
## data = preJAFselect)
##
## Residuals:
## LABEL: Promotive Prosocial Behavior
## VALUES:
## -3.1, -0.3567, 0, 0.7014, 1.4
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 3.5000 0.2643
## Influence 0.1000 0.0439
## LeaderOMM Leader 0.3336 0.5256
## Influence:LeaderOMM Leader -0.0419 0.0823
## t value
## (Intercept) 13.24
## Influence 2.28
## LeaderOMM Leader 0.63
## Influence:LeaderOMM Leader -0.51
## Pr(>|t|)
## (Intercept) <0.0000000000000002 ***
## Influence 0.024 *
## LeaderOMM Leader 0.527
## Influence:LeaderOMM Leader 0.612
## ---
## Signif. codes:
## 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.878 on 129 degrees of freedom
## (87 observations deleted due to missingness)
## Multiple R-squared: 0.0481, Adjusted R-squared: 0.0259
## F-statistic: 2.17 on 3 and 129 DF, p-value: 0.0945
Preventive Prosocial Behavior Regressed on Influence and Leader Role
Results of the multiple linear regression indicate there was no effect for the model, F(3,129) = 1.98, p = .12, R2 = .04). Simply, the amount of influence and role as member or leader does not relate to the amount of preventive prosocial behavior. As shown in the second plot, OMM members and OMM leaders have similar flat slopes.
##
## Call:
## lm(formula = Preventive.Prosocial ~ Influence + Leader + Influence:Leader,
## data = preJAFselect)
##
## Residuals:
## LABEL: Preventive Prosocial Behavior
## VALUES:
## -2.6679, -0.7291, 0.0874, 0.6678, 1.3321
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 3.5456 0.2502
## Influence 0.0612 0.0415
## LeaderOMM Leader -0.2887 0.4975
## Influence:LeaderOMM Leader 0.0583 0.0779
## t value
## (Intercept) 14.17
## Influence 1.47
## LeaderOMM Leader -0.58
## Influence:LeaderOMM Leader 0.75
## Pr(>|t|)
## (Intercept) <0.0000000000000002 ***
## Influence 0.14
## LeaderOMM Leader 0.56
## Influence:LeaderOMM Leader 0.46
## ---
## Signif. codes:
## 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.831 on 129 degrees of freedom
## (87 observations deleted due to missingness)
## Multiple R-squared: 0.044, Adjusted R-squared: 0.0218
## F-statistic: 1.98 on 3 and 129 DF, p-value: 0.12
Responsive Prosocial Behavior Regressed on Influence and Leader Role
Results of the multiple linear regression indicate there was no effect for the model, F(3,129) = .37, p = .77, R2 = .01). Simply, the amount of influence and role as member or leader does not relate to the amount of responsive prosocial behavior.
##
## Call:
## lm(formula = Responsive.Prosocial ~ Influence + Leader + Influence:Leader,
## data = preJAFselect)
##
## Residuals:
## LABEL: Responsive Prosocial Behavior
## VALUES:
## -1.8156, 0.115, 0.1377, 0.1844, 0.2486
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 4.7354 0.1229
## Influence 0.0161 0.0204
## LeaderOMM Leader 0.1780 0.2443
## Influence:LeaderOMM Leader -0.0217 0.0383
## t value
## (Intercept) 38.54
## Influence 0.79
## LeaderOMM Leader 0.73
## Influence:LeaderOMM Leader -0.57
## Pr(>|t|)
## (Intercept) <0.0000000000000002 ***
## Influence 0.43
## LeaderOMM Leader 0.47
## Influence:LeaderOMM Leader 0.57
## ---
## Signif. codes:
## 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.408 on 129 degrees of freedom
## (87 observations deleted due to missingness)
## Multiple R-squared: 0.00853, Adjusted R-squared: -0.0145
## F-statistic: 0.37 on 3 and 129 DF, p-value: 0.775
Recovery Prosocial Behavior Regressed on Influence and Leader Role
Results of the multiple linear regression indicate there was no effect for the model, F(3,129) = .72, p = .54, R2 = .02). Simply, the amount of influence and role as member or leader does not relate to the amount of recovery-focused prosocial behavior.
##
## Call:
## lm(formula = Recovery.Prosocial ~ Influence + Leader + Influence:Leader,
## data = preJAFselect)
##
## Residuals:
## LABEL: Recovery-Focused Prosocial Behavior
## VALUES:
## -1.7962, 0.1135, 0.2038, 0.224, 0.3012
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 4.7256 0.1378
## Influence 0.0101 0.0229
## LeaderOMM Leader 0.3956 0.2741
## Influence:LeaderOMM Leader -0.0570 0.0429
## t value
## (Intercept) 34.28
## Influence 0.44
## LeaderOMM Leader 1.44
## Influence:LeaderOMM Leader -1.33
## Pr(>|t|)
## (Intercept) <0.0000000000000002 ***
## Influence 0.66
## LeaderOMM Leader 0.15
## Influence:LeaderOMM Leader 0.19
## ---
## Signif. codes:
## 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.458 on 129 degrees of freedom
## (87 observations deleted due to missingness)
## Multiple R-squared: 0.0165, Adjusted R-squared: -0.00642
## F-statistic: 0.719 on 3 and 129 DF, p-value: 0.542
Promotive Campaigns Regressed on Influence and Leader Role
Results of the multiple linear regression indicate there is a significant effect for the model, F(3,129) = 7.17, p = .001, R2 = .143). The amount of influence (t = 3.413, p = .001) was a significant predictor, whereas leader role (t = 1.70, p = .09) and the interaction were not (t = -.96, p = .34). Simply put, as the amount of peer prosocial influence increases, the amount of support for wellness promotive school-wide campaigns also increases.
##
## Call:
## lm(formula = Promotive.Campaigns ~ Influence + Leader + Influence:Leader,
## data = preJAFselect)
##
## Residuals:
## LABEL: Promotive Campaigns
## VALUES:
## -3.0985, -0.4693, 0.2301, 0.6058, 1.6922
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 3.1497 0.2792
## Influence 0.1581 0.0463
## LeaderOMM Leader 0.9439 0.5552
## Influence:LeaderOMM Leader -0.0830 0.0869
## t value
## (Intercept) 11.28
## Influence 3.41
## LeaderOMM Leader 1.70
## Influence:LeaderOMM Leader -0.95
## Pr(>|t|)
## (Intercept) < 0.0000000000000002 ***
## Influence 0.00086 ***
## LeaderOMM Leader 0.09151 .
## Influence:LeaderOMM Leader 0.34160
## ---
## Signif. codes:
## 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.928 on 129 degrees of freedom
## (87 observations deleted due to missingness)
## Multiple R-squared: 0.143, Adjusted R-squared: 0.123
## F-statistic: 7.17 on 3 and 129 DF, p-value: 0.000171
Preventive Campaigns Regressed on Influence and Leader Role
Results of the multiple linear regression indicate there is a significant effect for the model, F(3,128) = 6.16, p = .001, R2 = .126). The amount of influence (t = 3.026, p = .003) was a significant predictor, whereas leader role (t = 1.843, p = .07) and the interaction were not (t = -1.08, p = .28). Simply put, as the amount of peer prosocial influence increases, the amount of support for preventive school-wide campaigns also increases.
##
## Call:
## lm(formula = Preventive.Campaigns ~ Influence + Leader + Influence:Leader,
## data = preJAFselect)
##
## Residuals:
## LABEL: Preventive Campaigns
## VALUES:
## -3.239, -0.526, 0.328, 0.615, 1.635
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 3.2191 0.2907
## Influence 0.1457 0.0482
## LeaderOMM Leader 1.0638 0.5771
## Influence:LeaderOMM Leader -0.0971 0.0903
## t value
## (Intercept) 11.07
## Influence 3.03
## LeaderOMM Leader 1.84
## Influence:LeaderOMM Leader -1.08
## Pr(>|t|)
## (Intercept) <0.0000000000000002 ***
## Influence 0.003 **
## LeaderOMM Leader 0.068 .
## Influence:LeaderOMM Leader 0.284
## ---
## Signif. codes:
## 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.964 on 128 degrees of freedom
## (88 observations deleted due to missingness)
## Multiple R-squared: 0.126, Adjusted R-squared: 0.106
## F-statistic: 6.16 on 3 and 128 DF, p-value: 0.000604
Summary
We hypothesized that leaders (who often have more participation and tenure in OMM at their school) would perform more prosocial behaviors and be more supportive of school-wide prosocial campaigns for change. These hypotheses were not supported by the data. OMM leaders and members score at similar rates across prosocial behavior types and school-wide campaign types.
We hypothesized peer prosocial influence, as measured by a respondent’s self-report, would relate to prosocial behaviors and school-wide campaigns. The results supported the hypotheses associated with school-wide campaigns: as peer prosocial influence increases, the amount of support for school-wide campaigns increases too. The hypotheses associated with prosocial behaviors were not supported: peer prosocial influence is not related to prosocial behaviors.
Limitations
This cross-sectional study was limited by various factors. The reliability and validity of the single-item measures is a serious concern. Future studies should use validated scales with additional items when possible. These data were collected at a single time point, which eliminated the possibility of prediction with casual inference. The primary research questions would be better answered with a longitudinal study using multiple time points with consideration for youth developmental change and the school calendar year.
OMM Programming
This study reaffirmed the multi-dimensionality of prosocial behavior, which should influence the program development of Our Minds Matter. The correlation among the four dyadic helping indices suggests a proactive helping (promotive and preventive prosocial indices) differs from a reactive helping (responsive and recovery-focused prosocial behaviors), which is consistent with past research. OMM program could focus on teaching students how to be more effective at proactive and reactive helping to benefit the mental health of their peers.
The results also demonstrated the lack of relationship between the four prosocial indicators and the two school-wide campaign items, suggesting prosocial behavior toward an individual is different from prosocial behavior toward a group/school. These results are also consistent with research on the relational context of helping. OMM could equip youth with the direct/spontaneous dyadic helping skills to benefit the mental health of one person and the indirect/school-wide helping skills necessary to improve the mental health of all peers at school.
The exploratory analyses on students’ perceptions of the ideal OMM club goal provide insight for future OMM program revisions. Intentionally building a wellness promotion and stigma prevention goal into themes and activities aligns with student preferences and positions OMM as a wellness promotion and stigma prevention program.
Future Research
Research related to the Our Minds Matter program should further explore the pre-disposing characteristics of youth leaders and level-two variables related to the OMM clubs/schools (e.g., fidelity measures, characteristics of the school). Themes and activities could be evaluated by students after each session based on whether they have a promotion goal, prevention goal or both.