In investigating mediating processes, researchers usually use randomized experiments and linear regression or structural equation modeling to determine if the treatment affects the hypothesized mediator and if the mediator affects the targeted outcome. However, randomizing the treatment will not yield accurate causal path estimates unless certain assumptions are satisfied. Since randomization of the mediator may not be plausible for most studies (i.e., the mediator status is not randomly assigned, but self-selected by participants), both the direct and indirect effects may be biased by confounding variables. The purpose of this dissertation is (1) to investigate the extent to which traditional mediation methods are affected by confounding variables and (2) to assess the statistical performance of several modern methods to address confounding variable effects in mediation analysis. This dissertation first reviewed the theoretical foundations of causal inference in statistical mediation analysis, modern statistical analysis for causal inference, and then described different methods to estimate causal direct and indirect effects in the presence of two post-treatment confounders. A large simulation study was designed to evaluate the extent to which ordinary regression and modern causal inference methods are able to obtain correct estimates of the direct and indirect effects when confounding variables that are present in the population are not included in the analysis. Five methods were compared in terms of bias, relative bias, mean square error, statistical power, Type I error rates, and confidence interval coverage to test how robust the methods are to the violation of the no unmeasured confounders assumption and confounder effect sizes. The methods explored were linear regression with adjustment, inverse propensity weighting, inverse propensity weighting with truncated weights, sequential g-estimation, and a doubly robust sequential g-estimation. Results showed that in estimating the direct and indirect effects, in general, sequential g-estimation performed the best in terms of bias, Type I error rates, power, and coverage across different confounder effect, direct effect, and sample sizes when all confounders were included in the estimation. When one of the two confounders were omitted from the estimation process, in general, none of the methods had acceptable relative bias in the simulation study. Omitting one of the confounders from estimation corresponded to the common case in mediation studies where no measure of a confounder is available but a confounder may affect the analysis. Failing to measure potential post-treatment confounder variables in a mediation model leads to biased estimates regardless of the analysis method used and emphasizes the importance of sensitivity analysis for causal mediation analysis.

Type 1 Diabetes Mellitus (T1DM) is a chronic disease that requires maintaining tight metabolic control through complex behavioral and pharmaceutical regimens. Subtle cognitive impairments and stress response dysregulation may partially account for problems negotiating life changes and maintaining treatment adherence among emerging adults. The current study examined whether young adults with T1DM physiologically respond to psychological stress in a dysregulated manner compared to non-diabetic peers, and if such individuals also demonstrated greater cognitive declines following psychological stress. Participants included 23 young adults with T1DM and 52 non-diabetic controls yoked to T1DM participants based on age, gender, ethnicity, participant education, and maternal education. Participants completed a laboratory-based social stressor, pre- and post-stressor neurocognitive testing, provided fingerstick blood spots (for glucose levels) and salivary samples (for cortisol levels) at five points across the protocol, and completed psychosocial questionnaires. Related measures ANOVAs were conducted to assess differences between T1DM participants and the average of yoked controls on cortisol and cognitive outcomes. Results demonstrated that differences in cortisol reactivity were dependent on T1DM participants' use of insulin pump therapy (IPT). T1DM participants not using IPT demonstrated elevated cortisol reactivity compared to matched controls. There was no difference in cortisol reactivity between the T1DM participants on IPT and matched controls. On the Stroop task, performance patterns did not differ between participants with T1DM not on IPT and matched controls. The performance of participants with T1DM on IPT slightly improved following the stressor and matched controls slightly worsened. On the Trail Making Test, the performance of participants with T1DM was not different following the stressor whereas participants without T1DM demonstrated a decline following the stressor. Participants with and without T1DM did not differ in patterns of performance on the Rey Verbal Learning Task, Sustained Attention Allocation Task, Controlled Oral Word Association Task, or overall cortisol output across participation. The results of this study are suggestive of an exaggerated cortisol response to psychological stress in T1DM and indicate potential direct and indirect protective influences of IPT.

The purpose of this study was to examine under which conditions "good" data characteristics can compensate for "poor" characteristics in Latent Class Analysis (LCA), as well as to set forth guidelines regarding the minimum sample size and ideal number and quality of indicators. In particular, we studied to which extent including a larger number of high quality indicators can compensate for a small sample size in LCA. The results suggest that in general, larger sample size, more indicators, higher quality of indicators, and a larger covariate effect correspond to more converged and proper replications, as well as fewer boundary estimates and less parameter bias. Based on the results, it is not recommended to use LCA with sample sizes lower than N = 100, and to use many high quality indicators and at least one strong covariate when using sample sizes less than N = 500.

Family plays an important yet understudied role in the development of psychopathology during childhood, particularly for children at developmental risk. Indeed, much of the research on families has actually concentrated more on risk processes in individual family members or within-family subsystems. In general, important and complex associations have been found among family-related constructs such as marital conflict, parent-child relationships, parental depression, and parenting stress, which have in turn been found to contribute to the emergence of children's behavioral problems. Research has begun to emerge that certain family system constructs, such as cohesion, organization, and control may influence children's development, but this research has been limited by a focus on parent-reports of family functioning, rather than utilizing observational methods. With notable exceptions, there is almost no observational research examining families of children at developmental risk. This study examined the longitudinal relations among family risk and family system constructs, as well as how family systems constructs mediated the relations between family risk and child outcome. Further, the study examined how developmental risk moderated these relations. The sample followed 242 families of children with and without developmental risk across the transition-to-school period. Family risk factors were assessed at 5 years, using parental reports of symptomatology, parenting stress, and marital adjustment, and observational assessments of the parent-child relationship. Family system constructs (cohesion, warmth, conflict, organization, control) were measured at age 6 using structured observations of the entire family playing a board game. Child behavior problems and social competence were assessed at age 7. Results indicated that families of children with developmental delays did not differ from families of typically developing children on the majority of family system attributes. Cohesion and organization mediated the relations between specific family risk factors and social competence for all families. For families of typically developing children only, higher levels of control were associated with more behavior problems and less social competence. These findings underscore the importance of family-level assessment in understanding the development of psychopathology. Important family effects on children's social competence were found, although the pathways among family risk and family systems attributes are complex.

Specific cultural variables have been found to protect against the onset of alcohol, tobacco and drug use among Latino adolescents. It has been suggested that targeting similar cultural components during the treatment of drug dependence and abuse for Latino adults may also enhance the effectiveness of the intervention, although few studies have explored this hypothesis. The current study attempted to remedy this disparity by exploring the potentially protective influence of two cultural variables, ethnic pride and family traditionalism, on self-efficacy to avoid drug use following residential substance abuse treatment among 99 Hispanic and 85 non-Hispanic White males. Results of the study indicate that higher levels of ethnic pride predict greater confidence to remain abstinent from drugs following substance abuse treatment, and that this relationship is stronger among Hispanic participants than non-Hispanic White participants. Family traditionalism was not a significant predictor of drug avoidance self-efficacy for either group, suggesting that some specific cultural variables may be better targets for substance abuse treatment than others. Study limitations and future directions for research and clinical practice are discussed.