UC Santa Barbara

This applied study presented method illustrations of the regression analytic models for moderated mediation hypotheses using data from a survey program. In spite of the theoretical and empirical importance of employing appropriate models for analyzing moderated mediation hypotheses, or conditional indirect effects in social and behavioral science research, there have been limited examples of formally testing those effects in the literature of applied psychology or education mainly due to the lack of clear methods for examining these effects.

Therefore formal methods that were initially introduced and organized by Preacher, Rucker, & Hayes (2007), Hayes (2011), and Hayes (2013) were illustrated with the use of commercially available statistical software packages of IBM SPSS Statistics Version 23 (IBM, 1989-2015) and Mplus Version 6.11 (Muthen and Muthen, 1998-2011) using data from The Positive Body Image Survey (Romo, Mireles-Rios, & Nylund-Gibson, 2012). The focus population of the survey was a group of adolescent girls with the intentions to identify behavioral factors that contribute to their resilience in spite of sociocultural pressures that tend to produce negative feeling about their bodies, and to examine whether positive body perceptions can lead to the practice of health-promoting lifestyle.

Results from this study will help applied researchers using models for moderated mediation hypotheses, or conditional indirect effects as to how social or behavioral phenomena in substantive areas of the study can be better understood. Results from this dissertation highlight the importance of framing research questions about positive body image development among Latino adolescent girls in terms of path analytic models with protective factors specified as a predictor variable, positive body image as a mediator variable, and engagement in health-promoting behaviors as an outcome variable. Findings of this study provide evidence that this mediation models can be affected by moderator variables of interest.