Search

Researchers interested in understanding causal relationships must not only test if X causes Y, but how and/or when X causes Y. Mediation analysis is a tool that allows researchers to identify the mechanism(s) by which one variable causes another, whereas moderation analysis allows researchers to detect when one variable’s effect is heterogenous across levels of another variable (or multiple variables). Although these analyses lead to a deeper understanding of an observed relationship, they are still often too simplistic in isolation to properly model real-world effects. Combining mediation and moderation into a single analysis allows one to study conditional indirect effects—that is, when an indirect effect of X on Y is variable across the levels of a moderator. Methodological researchers have paid much attention on how to test for conditional indirect effects. However, considerably less work has been devoted to evaluating the performance of these proposed methods or interpreting the results of these tests. A review of the simulation studies that have been done reveals that current testing methods have relatively poor performance except for the most optimistic combinations of effect and sample size. Despite this, many substantive researchers continue to use these methods and rely on them for dichotomous decisions about and interpretations of such effects. In this dissertation, I aim to clarify the best way(s) for researchers to specify, test, and interpret conditional indirect effects. In Chapter 1, I introduce the concepts of mediation, moderation, and conditional indirect effects and conduct a literature review to learn how methodological and substantive researchers think about and apply conditional indirect effects. In Chapter 2, I introduce the math underlying mediation, moderation, and conditional indirect effects and step through substantive examples of each. I also introduce a graphical presentation of effect size to aid in the interpretation of conditional indirect effects. I conduct a simulation study in Chapter 3 to evaluate three inferential procedures for conditional indirect effects by their Type I error and power and examine the relative effects of sample size, effect size, and the distribution of X (whether it is dichotomous as in an experiment with two conditions or a measured, normally distributed variable) on these procedures. All three procedures had low Type I error and power in many of the study conditions, but test performance improved as sample and effect size increased. The distribution of X had little effect on the Type I error or power of the procedures. Chapter 4 steps through a complex substantive example and provides guidelines for researchers on handling models with conditional indirect effects. Chapter 5 summarizes the findings of the dissertation and limitations of the investigation. Implications for practice and possible avenues for future research are also considered.

The number of adolescents who engage in suicidality is staggering. Adolescence is a unique developmental period where engagement in health-risk behaviors is prominent. Factors influencing engagement in suicidality are vast and occur both internally and externally in the adolescent. Substance use during adolescence has shown an association with increased suicidality. Similarly, bullying, both cyberbullying and in-person bullying, have contributed to adolescent suicidality. Alternatively, recreation engagement has yielded benefits in both psychosocial and physical health for adolescents. However, limited research has identified that engaging in recreation can minimize suicidality. Therefore, the purpose of this study was to explore how substance use, age, race, recreation engagement, and bullying can influence suicidality in high school students. The study employed quantitative methods for each aim. Conditional process analyses were used to identify the mechanisms which operate within and outside of the adolescent and their environments that affect suicidality in high school students. Results revealed that suicidality is associated with alcohol and opioid use, and adolescent sex and sexual identity. Cyberbullying is more prevalent than in-person bullying but both are significant for high school students. Conversely, neither form of bullying influenced suicidality. Additionally, female and male students engaged on more sports teams with a reduction in suicidality. Implications for practice include the use of recreation as a program to support adolescents engaged in health-risk behaviors (e.g., substance use and suicidality). Another implication for practice is the need to identify barriers to engagement in recreation programs, and more diverse recreation programs for the unique needs of all adolescents. Opportunities for future research include identifying evidenced-based recreation that can promote a reduction to adolescent health-risk behavior engagement, intervention-based research on adolescents who are engaging in health-risk behaviors, and recreation opportunities and barriers to engagement for diverse adolescent communities.

Conditional process models are commonly used in many areas of psychology research as well as research in other academic fields (e.g., marketing, communication, and education). Conditional process models combine mediation analysis and moderation analysis. Mediation analysis, sometimes called process analysis, investigates if an independent variable influences an outcome variable through a specific intermediary variable, sometimes called a mediator. Moderation analysis investigates if the relationship between two variables depends on another. Conditional process models are very popular because they allow us to better understand how the processes we are interested in might vary depending on characteristics of different individuals, situations, and other moderating variables. Methodological developments in conditional process analysis have primarily focused on the analysis of data collected using between-subjects experimental designs orcross-sectional designs. However, another very common design is the two-instance repeated-measures design. A two-instance repeated-measures design is one where each subject is measured twice; once in each of two instances. In the analysis discussed in this dissertation, the factor that differentiates the two repeated measurements is the independent variable of interest. Research on how to statistically test mediation, moderation, and conditional process models in these designs has been minimal. Judd, Kenny, and McClelland (2001) introduced a piecewise method for testing for mediation, reminiscent of the Baron and Kenny causal steps approach for between-participant designs. Montoya and Hayes (2017) took this piecewise approach and translated it to a path-analytic approach, allowing for a quantification of the indirect effect, more sophisticated methods of inference, and the extension to multiple mediator models. Moderation analysis in these designs has been described by Judd, McClelland, and Smith (1996), Judd et al. (2001), and Montoya (in press). However, the generalization to conditional process analysis, or moderated mediation, remains unknown. Describing this approach is the purpose of this dissertation. In this dissertation I propose a path analytic approach to assessing moderated mediation models for two-instance repeated-measures designs. InChapter 1, I review the development of conditional process analysis in between-subjects designs. In Chapter 2, I review existing methods for mediation and moderation in two-instance repeated-measures designs. In Chapter 3, I provide a general framework for estimating conditional process models for two-instance repeated-measures designs with one moderator and one mediator. I describe how simplifications of this general model correspond to more commonly used conditional process models, such as first-stage conditional process models and second-stage conditional process models. In Chapter 4, I provide examples of the models described in Chapter 3 using data sets from a variety of areas of psychology. Theseexamples show how to implement conditional process analysis and how to interpret the results of such an analysis. In Chapter 5, I discuss alternative methods for evaluating moderated mediation in two-instance repeated measures designs. I describe how the first-stage conditional process model in Chapter 3 is related to previous methods for testing mediation using a 2(within) X 2(between) design as described by MacKinnon (2008) and Valente and MacKinnon (2017). Two particularly popular alternative methodological approaches include multilevel models and structural equation modeling. I connect the regression based methods proposed in Chapter 3 to those described for 1-1-1 moderated mediation models. There is very little existing literature on assessing moderated mediation in repeated measures designs with structural equation models. Thus, I connect the proposed methods in this dissertation to existing approaches to mediation in repeated-measures designs with structural equation models: correlated residuals, latent-difference score models, latent-growth curve models, and cross-lag panel models. The final chapter provides a discussion of the general framework for approaching conditional process analysis in two-instance repeated-measures designs, including future directions as well as limitations of the design and analytic framework presented.

Researchers in the social sciences are often interested in explaining causal processes in which the effect of one variable is transmitted to another through one or more intervening (or mediator) variables. Statistical mediation analysis can be used by researchers to gain a better understanding of these causal processes. The methods traditionally used to quantify and test a mediation process, or indirect effect, are not directly applicable when the data are nested hierarchically, which is common in educational, organizational, and behavioral sciences where students may be nested within classrooms, employees may be nested within companies, or repeated measurements may be nested within individuals. Multilevel modeling (MLM) is a method for analyzing hierarchical data that allows for the simultaneous estimation of the effects of variables at multiple levels of the hierarchy. As such, several researchers have proposed methods for testing mediation using MLM. A majority of these methods have focused on models in which the independent variable is a level-2 variable, the mediator is either a level-2 or a level-1 variable, and the dependent variable is a level-1 variable. Although some (Kenny, Korchmaros, & Bolger, 2003; Bauer, Preacher, & Gil, 2006) have focused on models in which all three variables are measured at the lowest level, the methods proposed are not without their shortcomings. The aim of this thesis is to address and mitigate some of these shortcomings, as well as provide advancements in the formulation and testing of a number of interesting effects that can be modeled using MLM. The major contributions include demonstrating how to simultaneously estimate between-group and within-group indirect effects using traditional MLM software. A test for the difference between within-group and between-group indirect effects is also conceptualized and explained. Further, the model is expanded to include multiple mediators with either fixed or random slopes. In the case where one or more paths for one indirect effect covary with one or more paths from another, the covariance between the random indirect effects is derived. Finally, the index of moderated mediation (Hayes, 2015) is applied to the multilevel context to test the moderating effect of a level-2 variable on the within-group and/or between-group indirect effect. To demonstrate the methods discussed throughout this thesis, two real-world datasets are analyzed using MLmed, a free SPSS macro developed in conjunction with this thesis and designed specifically to fit multilevel mediation and moderated mediation models. The parameter estimates obtained using MLmed are comparable to those obtained using Mplus (Muthen & Muthen, 2015) and the lme4 package (Bates, Machler, Bolker, & Walker, 2015) in R (R Core Team, 2016), but with considerably less programming effort required.

Passing the National Council Licensure Examination (NCLEX) is required for a graduate of an accredited nursing program to practice nursing. The current study investigated first time student success on the NCLEX-RN, which is the examination used to grant licensure to Registered Nurses (RNs) after graduation from an accredited institution such as a school or a college of nursing. Universities, colleges, faculty, students, parents, healthcare employers, the National Council of States Boards of Nursing (NCSBN), and society in general have a vested interest in the success of nursing students on the NCLEX-RN. It is imperative that nurses are properly prepared and competent in providing safe and reliable healthcare services. Failure to pass the NCLEX-RN prevents practice as an RN, resulting in potential financial hardship, professional embarrassment, and a continued shortage of qualified RNs.Existing data from nursing students at a large, Midwestern university were analyzed. The main purpose of this study was to examine the relationship between Socioeconomic Status (SES), the ACT, and the NCLEX-RN. Through moderation and mediation logistic regression models, the above relationships were investigated. Additionally, Conditional Process Analysis was used to gain a more in-depth understanding of the complexity of the relationship between SES, ACT, and the NCLEX-RN. Proxies of SES were explored, which included: (1) Pell Grant Eligibility, (2) Student Race, (3) College Generation (i.e., first-generation or non first-generation), and (4) Zip Code.

This study used mediation analysis to evaluate Beck’s model and Abramson’s model in a sample of 198 girls, in the 4th through 7th grades. Data from diagnostic interviews were used to create a continuous measure of depressive symptoms. Self-report measures, including the Life Events Checklist, the Cognitive Triad Inventory for Children, and the Children’s Cognitive Style Questionnaire, were used to assess perceptions of negative life events, Beck’s cognitive triad, and Abramson’s cognitive inferential style, respectively. Results of separate mediation analyses supported both Beck’s and Abramson’s cognitive theories of depression in children. There was a significant indirect effect of life events through Beck’s cognitive triad on symptoms of depression. Similarly, there was a significant indirect effect of life events through Abramson’s negative cognitive inferential style on depressive symptoms. Higher depressive symptoms were associated with a more depressotypic cognitive triad, a more depressotypic cognitive inferential style, and a greater number and greater magnitude of negative life events. Developmentally-sensitive analyses using the weakest link approach and moderated mediation (conditional process) analysis found support for both Beck’s and Abramson’s models. However, there was also some support for the developmental hypothesis that in younger children negative life events can have a direct effect on symptoms of depression; for the 4th graders in this study, direct effects and indirect effects were significant. However, for the 5th, 6th, and 7th graders indirect effects were significant, but direct effects ceased to be significant. Implications and suggestions for future research are provided.

Despite the federal government’s $1.5 billion investment between 1993 and 2010 to fund 164 separate community-based systems of care, there has been an extremely limited attempt to measure the impact of system of care. The impetus for this research is the struggle for how the value based concept of system of care is communicated within a community. While child mental health services researchers have published a number of randomized control trials to explore individual level supports for youth served in a system of care community, researchers have struggled to devise a way to measure system of care philosophy diffusion. While system of care is a system level intervention, this study explored the role of the system of care value: family voice as it pertains to direct practice for children and families. The goal was to assess whether specific direct practices regularly associated with system of care (i.e., wraparound or home-based services) lead to greater family voice or if the mere presence of a high-functioning system of care community leads to equal family voice for all receiving community-based services. The primary finding was a relationship between the perception of family functioning and perceived empowerment/self-efficacy. This finding suggests that as functioning improves, so does a caregiver’s perception of their personal empowerment/ self-efficacy. While the framing of this study was to “unpack” the system of care value of family voice, the findings do not support any clear cut explanation for how family voice is promoted or communicated to families. Based on the findings, it appears as if families feel more empowered as their child improves. Additional research needs to be done on the application of family voice within the practice setting to better understand how to best instruct staff to infuse family voice in their daily practice.