1	Thesis Defense A Meta-Analysis of Adventure Therapy Program Outcomes Submitted in partial fulfillment of the degree Master of Science: Kinesiology- Outdoor Education By Norman Staunton
2	Committee Members Chair: Dr. Michael A. Gass, Ph.D. Members: Dr. Deborah Sugerman, Ph.D. Dr. Jason Bocarro, Ph.D. Advisor: Mr. James Neill MANY THANKS TO THEM ALL!
3	Gillis (1992) “Someone needs to conduct a meta-analysis on [the] therapeutic aspect[s] of adventure-challenge-outdoor-wilderness that includes the criteria of clinical significance along with traditional methods of effect size”
4	Gillis (2000) “To the best of my knowledge such a study has not been done.” As of December of 2002 no such study had been completed, and it represented a significant gap in the body of AT research. This study fills that gap by conducting such a meta-analysis.
5	Reasons for conducting this study: Not a duplication of previous research- it provides new information to the field Represents a significant contribution to the field- it fills a significant gap of needed information Provides documentation of collective efficacy- which is necessary for future funding, program design, research design
6	Question I: What are the effects of Adventure Therapy? How do the effects of Adventure Therapy compare to Adventure Education? How do the effects of Adventure Therapy compare to other forms of therapy?
7	Question II: What are the major correlates of Adventure Therapy program outcomes? Such as for: Programming (e.g. length, type, status) Group (e.g. social support, cohesion) Client (e.g. gender, age, diagnosis) Outcome (e.g. clinical, affective, attitude)
8	Definitions Adventure Therapy- Is programming aimed at changing [specified] dysfunctional behavior patterns, using adventure experiences as forms of habilitation and rehabilitation." (Priest and Gass 1997). For our purposes, this includes any adventure experience with diagnosed clients or reporting a specified therapeutic outcome.
9	Definitions Meta-analysis- A statistical technique for accumulating and representing the research results reported in multiple studies. Meta-analyses pool the findings about a single research question from many different sources and analyze the overall effects.
10	Definitions Effect Size- Is a statistical standardized coefficient that allows for comparisons of changes in means over time across studies. ES is calculated by the equation: (M₁-M₂)/SD_pooled
11	Practical Basis of Methodology Based on the previous OE meta-analyses Cason (1993)- AE outcomes on adolescents Hattie, Marsh, Neill & Richards (1997)- AE/ OB outcomes for all populations Marsh (1999)- Outcomes of camp experiences Hans (2000)- AE effects on locus of control
12	Practical Basis of Methodology This study departed from earlier OE meta-analyses by: Focusing specifically on AT and therapeutic outcomes Looks outside of OE for applicable research Being up to date
13	Methodology Study Methodology is divided into four sections: Search Methods Selection Criteria Coding Methods Statistical Methods
14	Search Methods Searches were made in six categories: Electronic Databases Bibliographies Top AT/OE Web Sites Commonly Known Research Email Listserv Requests Existing Data These methods returned over 2000 hits with over 600 unique articles
15	Selection Criteria Studies were included based on the following criteria: Empirically-based and likely reporting descriptive statistics Adventure-based intervention with diagnosed populations Studies meeting these criteria were actively sought by the researcher
16	Selection Criteria Studies were further included based on: Further investigation revealed the first criteria were correct 75 articles passed the third criteria. 22 were unavailable, taking the effective number of studies to 53. Analysis during coding eliminated 36 studies. 17 studies remained.
17	Coding Methodology Primary Variable Codes: Year of Study Quality of Measure Authors Quality of Study Institution Pre M and SD Country 1^st day M & SD Publication Source Last day M & SD Type of Program Follow up M & SD Outcomes Measure (M & SD calculated as needed)
18	Secondary Variable Codes Hierarchy \PROGRAM/ Type, Modality, Delivery Format, Duration, Daily Duration \GROUP/ Staff, Retention, Length of Program, Age, Gender \INDIVIDUAL/ Age, Gender, Category, Diagnosis \OUTCOME/ Behavioral, Attitudinal, Affective, Physical, Cognitive
19	Statistical Methods Effect Sizes (ES) ESs were calculated, then corrected to account for their relative strength. Correlations Relationship between effect sizes and primary and secondary variables were analyzed using univariate analysis. 95% confidence intervals were used to compare significant differences from zero and between variables.
20	Statistical Methods Regression Analysis Regression analysis was conducted until all returned correlates of variance were significant at the p=.05 level File Drawer Effect Rosenthal’s File Drawer Effect was calculated to estimate the number of negative or non-significant effects needed to negate the findings.
21	Results 17 Studies were included 95 Effect sizes were calculated (91/4) No negative effect sizes were found Overall WES of .53 found .53 for immediate effects, .35 for follow-up
22	Results
23	Results Univariate Analysis of Possible Predictors: Found significant positive results for: Dissertations (.17); Medium Quality (.74); High Quality (.44); ABC Programs (.72); Continuous Programs (.59); Intermittent Programs (.37); Residential Programs (.55); Single Sex Groups (.47); Coed Groups (.53); Adolescents (.49); Adults (1.11); Self Constructs (.25)
24	Results Univariate Analysis of Possible Predictors: All other variables were found to be significantly different from zero, but not between variables. Only Outpatient Programs had a non-significant ES (.19)
25	Results File Drawer Effect: 19 studies/ 648 ES would be needed to negate the findings of this study Regression returned Dissertations (-.28); ABC Programs (.38); Mixed Diagnoses (.74) as the three predictors significant at p=.05 Regressed Predictors account for 35.9% of variance in ES
26	Discussion Sample Size: small and some sampling error likely to exist. However, Quality of Studies and File Drawer Effect suggest some strength to the sample. Dahlen (2002) & Russell (2000) deserve special comment. ES drops to .35 without these studies.
27	Discussion
28	In other words...
29	Discussion Why ABC Programs may be more effective: More Developed Theory; Clinical Associations; More Educated Practitioners; Flexibility Why Mixed Diagnoses may be more effective: Holistic Nature of AT Programs; Role of Processing and Behavior
30	Discussion Why Dissertations may be least effective: Rigors of Publishing; Self-designed Measures; Least Experienced Researchers
31	Conclusions AT is effective (.53) and has lasting (increasing?) affects (.37). As a whole, the body of research is overwhelmingly positive. However, the overall quality of research and reporting is low AT compares favorably to other treatment modalities.
32	Conclusions ABC Methodologies and Mixed Diagnoses appear to be correlates of large ES. Further research is needed to test the causes and correlates of AT program effects. Further Meta-analysis is recommended. AT should become a respected Tx Modality.
33	THE END Thanks For Coming. Any Questions?