Use of social ecology and mixed methods research to develop an instrument to assess determinants of HIV/STI risk among a young adult population

Globally, 340 million people are affected by HIV/AIDS. Approximately 50,000 are newly infected within the United States each year with HIV. One in five of those infected is unaware of his/her status. Young adults are particularly impacted by the disease; 39% of new HIV infections are diagnosed in individuals between the ages of 13 and 29.;To comprehensively examine the independent and interrelated determinants of HIV/STI risks among young adults, the researcher developed, validated, and psychometrically examined a self-report instrument guided by constructs from the Social Ecological Model (SEM). Development and validation of the content within the survey was conducted using a six-phase qualitative and quantitative methodology. Proportional stratified sampling was used to attain a representative sample of 2-year and 4-year academic institutions from which participants (N = 728) for Phase VI (final administration) were recruited.;An exploratory factor analysis using principal component analysis (PCA) with an orthogonal varimax rotation was conducted to explore the latent structure of the instrument. A 9-factor solution consistent with the theoretical framework was maintained. Components accounted for 52% of the total variance. Calculated estimates of internal consistency revealed reliable scales representative of the Intrapersonal (i.e. attitudes and beliefs, condom use); Interpersonal (i.e. parent communication, partner communication); Institutional (i.e. campus resources); and Societal (i.e. policy) domains. Cronbach alpha coefficients ranged from α = 0.62 to α = 0.81.;Bivariate correlations and regression analyses were performed to assess determinants of HIV/STI risk among participants and further analyze predictive utility of identified scales. Independent variables for correlation analyses were taken from each theoretical construct. The dependent variable, HIV/STI risk, was defined by a summated score using five behavioral practices. Significant risk correlations were found within each theoretical domain. Multiple regression models with factors extracted during PCA accounted for a total of 21% of variance within the dependent variable. Several recommendations for research and policy are provided.