Alternative Effect Size Estimation in Randomized Controlled Trials with Heterogeneous Treatment Effects

190785-Thumbnail Image.png
Description
Psychologists report effect sizes in randomized controlled trials to facilitate interpretation and inform clinical or policy guidance. Since commonly used effect size measures (e.g., standardized mean difference) are not sensitive to heterogeneous treatment effects, methodologists have suggested the use of

Psychologists report effect sizes in randomized controlled trials to facilitate interpretation and inform clinical or policy guidance. Since commonly used effect size measures (e.g., standardized mean difference) are not sensitive to heterogeneous treatment effects, methodologists have suggested the use of an alternative effect size δ, a between-subjects causal parameter describing the probability that the outcome of a random participant in the treatment group is better than the outcome of another random participant in the control group. Although this effect size is useful, researchers could mistakenly use δ to describe its within-subject analogue, ψ, the probability that an individual will do better under the treatment than the control. Hand’s paradox describes the situation where ψ and δ are on opposing sides of 0.5: δ may imply most are helped whereas the (unknown) underlying ψ indicates that most are harmed by the treatment. The current study used Monte Carlo simulations to investigate plausible situations under which Hand’s paradox does and does not occur, tracked the magnitude of the discrepancy between ψ and δ, and explored whether the size of the discrepancy could be reduced with a relevant covariate. The findings suggested that although the paradox should not occur under bivariate normal data conditions in the population, there could be sample cases with the paradox. The magnitude of the discrepancy between ψ and δ depended on both the size of the average treatment effect and the underlying correlation between the potential outcomes, ρ. Smaller effects led to larger discrepancies when ρ < 0 and ρ = 1, whereas larger effects led to larger discrepancies when 0 < ρ < 1. It was useful to consider a relevant covariate when calculating ψ and δ. Although ψ and δ were still discrepant within covariate levels, results indicated that conditioning upon relevant covariates is still useful in describing heterogeneous treatment effects.
Date Created
2023
Agent

FINDING-Food: Frequent Informational Nudges Directing Individuals to Needed Goods. A Food Insecurity Intervention.

187681-Thumbnail Image.png
Description
Food insecurity is an economic and social condition involving limited or uncertain access to food. The problem of food insecurity in communities is influenced by economic conditions, food deserts, and barriers to accessing healthy food. Individuals experiencing food insecurity often

Food insecurity is an economic and social condition involving limited or uncertain access to food. The problem of food insecurity in communities is influenced by economic conditions, food deserts, and barriers to accessing healthy food. Individuals experiencing food insecurity often endure concurrent problems of financial instability, hunger, and poor mental and physical health. Public and non-profit services in the U.S., such as the federally supported Supplemental Nutrition Assistance Program (SNAP) and community food banks, provide food-related assistance to individuals who are at a high risk of experiencing food insecurity. Unfortunately, many individuals who qualify for these services still experience food insecurity due to barriers preventing them from accessing food, which may include inadequate finances, transportation, skills, and information. Effective approaches for removing barriers that prevent individuals from accessing food are needed to mitigate the increased risk of hunger, nutritional deficiencies, and chronic disease among vulnerable populations. This dissertation tested a novel food insecurity intervention using informational nudges to promote food security through the elimination of information barriers to accessing food. The intervention used in this mixed-methods feasibility study consisted of informational nudges in the form of weekly text messages that were sent to food pantry clients experiencing food insecurity. The study aims were to test the efficacy and acceptability of the intervention by examining whether the informational nudges could enhance food pantry utilization, increase SNAP registration, and promote food security. Quantitative study results showed a lower prevalence of food insecurity in the intervention group than the control group. Qualitative findings revealed how the intervention group found the text messages to be helpful and informative. These study findings can enhance future food insecurity interventions aiming to eliminate barriers that prevent individuals who are food insecure from accessing healthy food.
Date Created
2023
Agent

Beyond Moderation: Exploring Person-Level Mediation with Residuals and Individual Model Fit

171394-Thumbnail Image.png
Description
Mediation analysis is integral to psychology, investigating human behavior’s causal mechanisms. The diversity of explanations for human behavior has implications for the estimation and interpretation of statistical mediation models. Individuals can have similar observed outcomes while undergoing different causal processes

Mediation analysis is integral to psychology, investigating human behavior’s causal mechanisms. The diversity of explanations for human behavior has implications for the estimation and interpretation of statistical mediation models. Individuals can have similar observed outcomes while undergoing different causal processes or different observed outcomes while receiving the same treatment. Researchers can employ diverse strategies when studying individual differences in multiple mediation pathways, including individual fit measures and analysis of residuals. This dissertation investigates the use of individual residuals and fit measures to identify individual differences in multiple mediation pathways. More specifically, this study focuses on mediation model residuals in a heterogeneous population in which some people experience indirect effects through one mediator and others experience indirect effects through a different mediator. A simulation study investigates 162 conditions defined by effect size and sample size for three proposed methods: residual differences, delta z, and generalized Cook’s distance. Results indicate that analogs of Type 1 error rates are generally acceptable for the method of residual differences, but statistical power is limited. Likewise, neither delta z nor gCd could reliably distinguish between contrasts that had true effects and those that did not. The outcomes of this study reveal the potential for statistical measures of individual mediation. However, limitations related to unequal subpopulation variances, multiple dependent variables, the inherent relationship between direct effects and unestimated indirect effects, and minimal contrast effects require more research to develop a simple method that researchers can use on single data sets.
Date Created
2022
Agent

Brief, Non-Pharmacological, Interventions for Pediatric Anxiety: Meta-Analysis and Evidence Base Status

157646-Thumbnail Image.png
Description
There is a need to reinvent evidence-based interventions (EBIs) for pediatric anxiety problems to better address the demands of real-word service delivery settings and achieve public health impact. The time- and resource-intensive nature of most EBIs for youth anxiety has

There is a need to reinvent evidence-based interventions (EBIs) for pediatric anxiety problems to better address the demands of real-word service delivery settings and achieve public health impact. The time- and resource-intensive nature of most EBIs for youth anxiety has frequently been noted as a barrier to the utilization of EBIs in community settings, leading to increased attention towards exploring the viability of briefer, more accessible protocols. Principally, this research reports between-group effect sizes from brief-interventions targeting pediatric anxiety and classifies each as well-established, probably efficacious, possibly efficacious, experimental, or questionable. brief interventions yielded an overall mean effect size of 0.19 on pediatric anxiety outcomes from pre to post. Effect sizes varied significantly by level of intervention: Pre to post-intervention effects were strongest for brief-treatments (0.35), followed by brief-targeted prevention (0.22), and weakest for brief-universal prevention (0.09). No participant or other intervention characteristic emerged as significant moderators of effect sizes. In terms of standard of evidence, one brief intervention is well-established, and five are probably efficacious, with most drawing on cognitive and behavioral change procedures and/or family systems models. At this juncture, the minimal intervention needed for clinical change in pediatric anxiety points to in-vivo exposures for specific phobias (~3 hours), cognitive-behavioral therapy (CBT) with social skills training (~3 hours), and CBT based parent training (~6 hours, eight digital modules with clinician support). This research concludes with a discussion on limitations to available brief EBIs, practice guidelines, and future research needed to capitalize on the viability of briefer protocols in enhancing access to, and impact of, evidence-based care in the real-world.
Date Created
2019
Agent

Addressing the Variable Selection Bias and Local Optimum Limitations of Longitudinal Recursive Partitioning with Time-Efficient Approximations

157544-Thumbnail Image.png
Description
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

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.
Date Created
2019
Agent

The Safety and Health Improvement: Enhancing Law Enforcement Departments Study: Feasibility and Findings

128195-Thumbnail Image.png
Description

This randomized prospective trial aimed to assess the feasibility and efficacy of a team-based worksite health and safety intervention for law enforcement personnel. Four-hundred and eight subjects were enrolled and half were randomized to meet for weekly, peer-led sessions delivered

This randomized prospective trial aimed to assess the feasibility and efficacy of a team-based worksite health and safety intervention for law enforcement personnel. Four-hundred and eight subjects were enrolled and half were randomized to meet for weekly, peer-led sessions delivered from a scripted team-based health and safety curriculum. Curriculum addressed: exercise, nutrition, stress, sleep, body weight, injury, and other unhealthy lifestyle behaviors such as smoking and heavy alcohol use. Health and safety questionnaires administered before and after the intervention found significant improvements for increased fruit and vegetable consumption, overall healthy eating, increased sleep quantity and sleep quality, and reduced personal stress.

Date Created
2014-05-08
Agent

Mediation Analysis With Survival Outcomes: Accelerated Failure Time VS. Proportional Hazards Models

128235-Thumbnail Image.png
Description

Objective: Survival time is an important type of outcome variable in treatment research. Currently, limited guidance is available regarding performing mediation analyses with survival outcomes, which generally do not have normally distributed errors, and contain unobserved (censored) events. We present

Objective: Survival time is an important type of outcome variable in treatment research. Currently, limited guidance is available regarding performing mediation analyses with survival outcomes, which generally do not have normally distributed errors, and contain unobserved (censored) events. We present considerations for choosing an approach, using a comparison of semi-parametric proportional hazards (PH) and fully parametric accelerated failure time (AFT) approaches for illustration.

Method: We compare PH and AFT models and procedures in their integration into mediation models and review their ability to produce coefficients that estimate causal effects. Using simulation studies modeling Weibull-distributed survival times, we compare statistical properties of mediation analyses incorporating PH and AFT approaches (employing SAS procedures PHREG and LIFEREG, respectively) under varied data conditions, some including censoring. A simulated data set illustrates the findings.

Results: AFT models integrate more easily than PH models into mediation models. Furthermore, mediation analyses incorporating LIFEREG produce coefficients that can estimate causal effects, and demonstrate superior statistical properties. Censoring introduces bias in the coefficient estimate representing the treatment effect on outcome—underestimation in LIFEREG, and overestimation in PHREG. With LIFEREG, this bias can be addressed using an alternative estimate obtained from combining other coefficients, whereas this is not possible with PHREG.

Conclusions: When Weibull assumptions are not violated, there are compelling advantages to using LIFEREG over PHREG for mediation analyses involving survival-time outcomes. Irrespective of the procedures used, the interpretation of coefficients, effects of censoring on coefficient estimates, and statistical properties should be taken into account when reporting results.

Date Created
2016-03-30
Agent

Serotonin functioning and adolescents' alcohol use: a genetically informed study examining mechanisms of risk

155680-Thumbnail Image.png
Description
The current study utilized data from two longitudinal samples to test mechanisms in the relation between a polygenic risk score indexing serotonin functioning and alcohol use in adolescence. Specifically, this study tested whether individuals with lower levels of serotonin functioning

The current study utilized data from two longitudinal samples to test mechanisms in the relation between a polygenic risk score indexing serotonin functioning and alcohol use in adolescence. Specifically, this study tested whether individuals with lower levels of serotonin functioning as indexed by a polygenic risk score were vulnerable to poorer self-regulation, and whether poorer self-regulation subsequently predicted the divergent outcomes of depressive symptoms and aggressive/antisocial behaviors. This study then examined whether depressive symptoms and aggressive/antisocial behaviors conferred risk for later alcohol use in adolescence, and whether polygenic risk and effortful control had direct effects on alcohol use that were not mediated through problem behaviors. Finally, the study examined the potential moderating role of gender in these pathways to alcohol use.

Structural equation modeling was used to test hypotheses. Results from an independent genome-wide association study of 5-hydroxyindoleacetic acid in the cerebrospinal fluid were used to create serotonin (5-HT) polygenic risk scores, wherein higher scores reflected lower levels of 5-HT functioning. Data from three time points were drawn from each sample, and all paths were prospective. Findings suggested that 5-HT polygenic risk did not predict self-regulatory constructs. However, 5-HT polygenic risk did predict the divergent outcomes of depression and aggression/antisociality, such that higher levels of 5-HT polygenic risk predicted greater levels of depression and aggression/antisociality. Results most clearly supported adolescents’ aggression/antisociality as a mechanism in the relation between 5-HT polygenic risk and later alcohol use. Deficits in self-regulation also predicted depression and aggression/antisociality, and indirectly predicted alcohol use through aggression/antisociality. These pathways to alcohol use might be the most salient for boys with low levels of socioeconomic status.

Results are novel contributions to the literature. The previously observed association between serotonin functioning and alcohol use might be due, in part, to the fact that individuals with lower levels of serotonin functioning are predisposed towards developing earlier aggression/antisociality. Results did not support the hypothesis that serotonin functioning predisposes individuals to deficits in self-regulatory abilities. Findings extend previous research by suggesting that serotonin functioning and self-regulation might be transdiagnostic risk factors for many types of psychopathology.
Date Created
2017
Agent

A Bayesian Synthesis approach to data fusion using augmented data-dependent priors

155625-Thumbnail Image.png
Description
The process of combining data is one in which information from disjoint datasets sharing at least a number of common variables is merged. This process is commonly referred to as data fusion, with the main objective of creating a new

The process of combining data is one in which information from disjoint datasets sharing at least a number of common variables is merged. This process is commonly referred to as data fusion, with the main objective of creating a new dataset permitting more flexible analyses than the separate analysis of each individual dataset. Many data fusion methods have been proposed in the literature, although most utilize the frequentist framework. This dissertation investigates a new approach called Bayesian Synthesis in which information obtained from one dataset acts as priors for the next analysis. This process continues sequentially until a single posterior distribution is created using all available data. These informative augmented data-dependent priors provide an extra source of information that may aid in the accuracy of estimation. To examine the performance of the proposed Bayesian Synthesis approach, first, results of simulated data with known population values under a variety of conditions were examined. Next, these results were compared to those from the traditional maximum likelihood approach to data fusion, as well as the data fusion approach analyzed via Bayes. The assessment of parameter recovery based on the proposed Bayesian Synthesis approach was evaluated using four criteria to reflect measures of raw bias, relative bias, accuracy, and efficiency. Subsequently, empirical analyses with real data were conducted. For this purpose, the fusion of real data from five longitudinal studies of mathematics ability varying in their assessment of ability and in the timing of measurement occasions was used. Results from the Bayesian Synthesis and data fusion approaches with combined data using Bayesian and maximum likelihood estimation methods were reported. The results illustrate that Bayesian Synthesis with data driven priors is a highly effective approach, provided that the sample sizes for the fused data are large enough to provide unbiased estimates. Bayesian Synthesis provides another beneficial approach to data fusion that can effectively be used to enhance the validity of conclusions obtained from the merging of data from different studies.
Date Created
2017
Agent

Mechanisms of Motivational Interviewing in Health Promotion: A Bayesian Mediation Analysis

129005-Thumbnail Image.png
Description

Background: Counselor behaviors that mediate the efficacy of motivational interviewing (MI) are not well understood, especially when applied to health behavior promotion. We hypothesized that client change talk mediates the relationship between counselor variables and subsequent client behavior change.

Methods: Purposeful sampling identified

Background: Counselor behaviors that mediate the efficacy of motivational interviewing (MI) are not well understood, especially when applied to health behavior promotion. We hypothesized that client change talk mediates the relationship between counselor variables and subsequent client behavior change.

Methods: Purposeful sampling identified individuals from a prospective randomized worksite trial using an MI intervention to promote firefighters’ healthy diet and regular exercise that increased dietary intake of fruits and vegetables (n = 21) or did not increase intake of fruits and vegetables (n = 22). MI interactions were coded using the Motivational Interviewing Skill Code (MISC 2.1) to categorize counselor and firefighter verbal utterances. Both Bayesian and frequentist mediation analyses were used to investigate whether client change talk mediated the relationship between counselor skills and behavior change.

Results: Counselors’ global spirit, empathy, and direction and MI-consistent behavioral counts (e.g., reflections, open questions, affirmations, emphasize control) significantly correlated with firefighters’ total client change talk utterances (rs = 0.42, 0.40, 0.30, and 0.61, respectively), which correlated significantly with their fruit and vegetable intake increase (r = 0.33). Both Bayesian and frequentist mediation analyses demonstrated that findings were consistent with hypotheses, such that total client change talk mediated the relationship between counselor’s skills—MI-consistent behaviors [Bayesian mediated effect: αβ = .06 (.03), 95% CI = .02, .12] and MI spirit [Bayesian mediated effect: αβ = .06 (.03), 95% CI = .01, .13]—and increased fruit and vegetable consumption.

Conclusion: Motivational interviewing is a resource- and time-intensive intervention, and is currently being applied in many arenas. Previous research has identified the importance of counselor behaviors and client change talk in the treatment of substance use disorders. Our results indicate that similar mechanisms may underlie the effects of MI for dietary change. These results inform MI training and application by identifying those processes critical for MI success in health promotion domains.

Date Created
2012-06-08
Agent