Many factors have been identified as being influential in the development of and participation in risk-taking behaviors during adolescence. The majority of these are considered psychosocial in nature. Results from these investigations have yielded valuable information that has been used in the design of prevention and intervention programs for youth. However, most of these studies have tended to be categorical in nature, rather than comprehensive, and thus do not account for the interaction between psychosocial influences on behavior. Therefore, due to the paucity of comprehensive research on the relationship between psychosocial factors and risk behaviors in adolescence, the overall purpose of this study was to determine the percent of variance a multi-dimensional model of psychosocial factors, jointly labeled "wellness", would account for in selected risk behaviors. To accomplish this, wellness themes relevant to the adolescent population were determined, a wellness framework developed based on these wellness themes, and an Adolescent Wellness Survey (AWS) created and validated.
Results from the pilot study indicated that six psychosocial domains constituted the wellness framework. These six domains were physical wellness, social wellness, emotional wellness, sense of purpose, locus of control, and cognitive wellness. The AWS was then developed with six subscales based on these domains. However, it was determined, based upon a principal axis factor analysis, that the AWS was composed of only five factors. Based upon the items that loaded on each of the five factors, the domains included in the final wellness framework were physical wellness, social wellness--family, social wellness--emotional wellness, and self-esteem. Once the factors were identified and subscales finalized, internal consistency reliability was determined.
While modified slightly based upon statistical results, the proposed wellness framework was supported by the factor analysis in that 95% of all items loaded at .40 or above. The AWS also showed evidence of internal consistency with all but one subscale having alpha coefficients at .70 or higher. The one subscale whose alpha was low (α = .59) was modified in Study Three by adding additional items to the survey. This increased the alpha level to .72.
During the final phase of this dissertation (Study Three), the AWS was used to measure the independent variable (wellness domains) and then a multiple regression analysis performed to determine the models ability to predict selected risk behaviors. The risk behaviors were measured using a modified version of the Youth Risk Behavior Survey. Risk behaviors measured included alcohol use, marijuana use, other drug use, and violence-related behaviors. One health-promoting behavior, physical activity level, was also measured. Results indicate that the model was only partially successful in predicting these behaviors as a whole. With the exception of physical activity level (which was predicted by only two wellness domains) each behavior was predicted by at least three wellness domains. Recommendations for improving the predictive ability of the entire model are discussed in Chapters Five and Six.