This project studied a four-variable single mediator model, a single mediator model: X (independent variable) to M (mediator) to Y (dependent variable), and a confounder (U) that influences M and Y. Confounding represents a threat to the causal interpretation in mediation analysis. For instance, if X represents random assignment to control and treatment conditions, the effect of X on M and the effect of X on Y have a causal interpretation under certain reasonable assumptions. However, the randomization of X does not allow for a causal interpretation of the M to Y effect unless certain confounding assumptions are satisfied. The aim of this project was to develop a significance test and an effect size comparison for two sensitivity to confounding analyses methods: Left Out Variables Error (L.O.V.E.) and the correlated residuals method. Further, the project assessed the accuracy of the methods for identifying confounding bias by simulating data with and without confounding bias.

Scale scores play a significant role in research and practice in a wide range of areas such as education, psychology, and health sciences. Although the methods of scale scoring have advanced considerably over the last 100 years, researchers and practitioners have generally been slow to implement these advances. There are many topics that fall under this umbrella but the current study focuses on two. The first topic is that of subscores and total scores. Many of the scales in psychological and health research are designed to yield subscores, yet it is common to see total scores reported instead. Simplifying scores in this way, however, may have important implications for researchers and scale users in terms of interpretation and use. The second topic is subscore augmentation. That is, if there are subscores, how much value is there in using a subscore augmentation method? Most people using psychological assessments are unfamiliar with score augmentation techniques and the potential benefits they may have over the traditional sum score approach. The current study borrows methods from education to explore the magnitude of improvement of using augmented scores over observed scores. Data was simulated using the Graded Response Model. Factors controlled in the simulation were number of subscales, number of items per subscale, level of correlation between subscales, and sample size. Four estimates of the true subscore were considered (raw, subscore-adjusted, total score-adjusted, joint score-adjusted). Results from the simulation suggest that the score adjusted with total score information may perform poorly when the level of inter-subscore correlation is 0.3. Joint scores perform well most of the time, and the subscore-adjusted scores and joint-adjusted scores were always better performers than raw scores. Finally, general advice to applied users is provided.

Negative Urgency and Positive Urgency are important subfacets of a propensity to rash action. There is currently limited research on parental antecedents of Negative Urgency and Positive Urgency. The current study investigated whether parent personality and parenting behaviors predict adolescent Negative Urgency and Positive Urgency. Data were taken from a community sample with parent personality, positive parenting behaviors, and child Negative Urgency and Positive Urgency measured at separate timepoints. Structural equation models were used to examine whether parent personality predicted adolescent Negative Urgency and Positive Urgency and whether positive parenting mediated this relationship. There was no evidence for a relationship between parent personality and children’s Negative Urgency and Positive Urgency. In addition, there was no relationship between parenting behaviors and child Negative and Positive Urgency in cross-reporter models, but child-reported parenting predicted later adolescent-reported Negative and Positive Urgency. Greater positive parenting, as perceived by children, was related to less Negative and Positive Urgency when they were adolescents. More research is needed to understand whether the current results are due to reporter bias or whether child-perceived parenting behaviors influence the development of adolescent Negative and Positive Urgency.

Despite the established co-prevalence of substance use (SU) and disordered eating (DE), few longitudinal studies have sought to examine their shared development. Findings have been inconsistent within the extant literature. This may be attributable in part to several methodological aspects, including overlooking distinct psychopharmacological properties of common substances of abuse, examining only between-person relations, and failing to account for shared risk factors. The current study sought to address these gaps by applying latent curve models with structured residuals (LCM-SR) to a preexisting, national sample of adolescent girls followed into adulthood, Add Health. In Aim 1, between-person effects examined the simultaneous development of alcohol, tobacco, and marijuana use and DE behaviors in substance-specific models. In Aim 2, bivariate latent curve models were expanded to account for within-person effects (LCM-SR) in order to examine the potentially bidirectional, prospective relationship between use of a specific substance and DE. Lastly, models accounted for shared developmental risk factors. Findings of the current study demonstrate preliminary evidence of substance-specific effects with DE emerging in adolescence. Across model-building steps, DE engagement in early adolescence was significantly associated with growth in tobacco use and marginally associated with marijuana use. Appetitive side-effects of both substances may link use with DE behaviors and enhance instrumental use for weight control. Significant associations did not emerge between alcohol and DE, and results of the conditional model indicate this co-occurrence is best explained by third variable mechanisms. Implications for prevention are discussed.

The ability to walk while completing a secondary task, dual-task walking (DTW), poses notable challenges for individuals affected by neurological disorders, such as multiple sclerosis (MS), who experience both cognitive and motor problems secondary to their disease. However, DTW is an everyday activity that has putative importance for optimal function. Although some research in the past decade has begun to examine changes in DTW in MS, there is still limited work to understand the predictors of DTW, the factors that might moderate relationships between baseline cognitive and motor function and DTW ability, and its consequences (e.g., for quality of life [QoL] or fall risk). To contribute to the understanding of these phenomena and their intersections, three secondary data analyses of two relatively large data sets in the area were conducted to address five major aims. The first step was to identify of the most relevant of these inherently involved domains (cognitive [aim 1] and motor [aim 2] abilities). Lasso regression for inference was performed to address this question for both cognitive (South Shore Neurologic Associates, PC data) and motor (University of Kansas Medical Center [KUMC] data) domains. Next, evaluations to explore the moderating role of the psychological impacts that are common in MS (e.g., depression and falls self-efficacy) were undertaken to determine whether the relationships between cognitive and motor function and DTW ability are different for individuals with different levels of these factors using regression with factor scores performed with each data set (aim 3). As a final step, relationships between DTW and distal outcomes like QoL (cross-sectionally using both data sets and factor score regression; aim 4) and falls (cross-sectionally and longitudinally using KUMC data and negative binomial regression; aim 5). These studies contribute to the corpus of knowledge about DTW in MS in needed ways.

Examining processes that characterize the ebb and flow of emotions offers insight into how infants modulate their own emotional experience as well as how both mothers and infants jointly regulate their emotional states. Drawing from polyvagal theory, which posits that vagal tone supports the capacity to quickly, flexibly, and adaptively respond to contextual demands (Porges, 2003, 2007), I hypothesized that infants with greater vagal tone (indexed by respiratory sinus arrhythmia; RSA) would show stronger evidence of emotion regulation and coregulation processes during free play and a frustrating task at 24 weeks child age. To evaluate these hypotheses, I used dynamic structural equation modeling (DSEM; Asparouhov, Hamaker, & Muthén, 2018) to examine biologically-based differences in second-by-second infant emotion regulation (equilibria, volatility, carryover, and feedback loops in positive and negative affect engagement) and mother- and infant-driven coregulation processes, among a sample of 210 low-income, Mexican-origin mother-infant dyads. Results offered evidence of both mother-driven and infant-driven emotion coregulatory processes during free play, which did not differ based on infant RSA. Results offered limited support for RSA-based differences in infant self-regulation processes during the teaching task, such that infants with below average RSA tended to respond to increased negative affect with subsequent increases in positive affect engagement. Prenatal maternal depressive symptoms also accounted for greater infant emotional volatility and weaker mother-driven emotion coregulation. Results highlight the unique roles mothers and infants play in achieving emotion regulation, as well as between-dyad differences in these processes, suggesting multiple pathways towards resilience among low-income, Mexican-origin families.

Family disruption, or the separation of children from caregivers, has been well-established in prior literature as a risk factor for child maladjustment; however, little is known about how family disruption impacts youth into adulthood, particularly how it influences children’s later parenting of their own offspring. The present study examined whether cumulative family disruption (i.e., parental hospitalization, death, incarceration, divorce) in childhood exerts effects on children’s parenting of their own offspring in adulthood, beyond other demographic characteristics and risk factors. Further, several potential mechanisms were hypothesized to underlie the association between family disruption in the first and second generation (G1-G2) family and later parenting provided from second-generation (G2) adults to third-generation (G3) children. Mediators included conflict and disorganization in the G1-G2 family and dysregulation in the G2 child.

Participants (N = 236 in models that included multiple G2 siblings; N = 110 in models without siblings) were drawn from a larger sample of at-risk (i.e., alcoholic) and comparison families followed longitudinally for over 30 years and across three generations. Four mediation models were estimated to examine effects of two separate G1-G2 family disruption components (deviance-related and health-related disruption) on parenting of G3, mediated by family conflict, family disorganization, and G2 dysregulation. Results indicated that health-related disruption impairs consistency of parenting provided to G3 offspring through conflict in the G1-G2 family. A direct effect of health-related disruption was also seen on parental monitoring. There were no direct or mediated effects of deviance-related disruption on parenting. Implications and future directions will be discussed.

Longitudinal recursive partitioning (LRP) is a tree-based method for longitudinal data. It takes a sample of individuals that were each measured repeatedly across time, and it splits them based on a set of covariates such that individuals with similar trajectories become grouped together into nodes. LRP does this by fitting a mixed-effects model to each node every time that it becomes partitioned and extracting the deviance, which is the measure of node purity. LRP is implemented using the classification and regression tree algorithm, which suffers from a variable selection bias and does not guarantee reaching a global optimum. Additionally, fitting mixed-effects models to each potential split only to extract the deviance and discard the rest of the information is a computationally intensive procedure. Therefore, in this dissertation, I address the high computational demand, variable selection bias, and local optimum solution. I propose three approximation methods that reduce the computational demand of LRP, and at the same time, allow for a straightforward extension to recursive partitioning algorithms that do not have a variable selection bias and can reach the global optimum solution. In the three proposed approximations, a mixed-effects model is fit to the full data, and the growth curve coefficients for each individual are extracted. Then, (1) a principal component analysis is fit to the set of coefficients and the principal component score is extracted for each individual, (2) a one-factor model is fit to the coefficients and the factor score is extracted, or (3) the coefficients are summed. The three methods result in each individual having a single score that represents the growth curve trajectory. Therefore, now that the outcome is a single score for each individual, any tree-based method may be used for partitioning the data and group the individuals together. Once the individuals are assigned to their final nodes, a mixed-effects model is fit to each terminal node with the individuals belonging to it.

I conduct a simulation study, where I show that the approximation methods achieve the goals proposed while maintaining a similar level of out-of-sample prediction accuracy as LRP. I then illustrate and compare the methods using an applied data.