Accurate data analysis and interpretation of results may be influenced by many potential factors. The factors of interest in the current work are the chosen analysis model(s), the presence of missing data, and the type(s) of data collected. If analysis models are used which a) do not accurately capture the structure of relationships in the data such as clustered/hierarchical data, b) do not allow or control for missing values present in the data, or c) do not accurately compensate for different data types such as categorical data, then the assumptions associated with the model have not been met and the results of the analysis may be inaccurate. In the presence of clustered
ested data, hierarchical linear modeling or multilevel modeling (MLM; Raudenbush & Bryk, 2002) has the ability to predict outcomes for each level of analysis and across multiple levels (accounting for relationships between levels) providing a significant advantage over single-level analyses. When multilevel data contain missingness, multilevel multiple imputation (MLMI) techniques may be used to model both the missingness and the clustered nature of the data. With categorical multilevel data with missingness, categorical MLMI must be used. Two such routines for MLMI with continuous and categorical data were explored with missing at random (MAR) data: a formal Bayesian imputation and analysis routine in JAGS (R/JAGS) and a common MLM procedure of imputation via Bayesian estimation in BLImP with frequentist analysis of the multilevel model in Mplus (BLImP/Mplus). Manipulated variables included interclass correlations, number of clusters, and the rate of missingness. Results showed that with continuous data, R/JAGS returned more accurate parameter estimates than BLImP/Mplus for almost all parameters of interest across levels of the manipulated variables. Both R/JAGS and BLImP/Mplus encountered convergence issues and returned inaccurate parameter estimates when imputing and analyzing dichotomous data. Follow-up studies showed that JAGS and BLImP returned similar imputed datasets but the choice of analysis software for MLM impacted the recovery of accurate parameter estimates. Implications of these findings and recommendations for further research will be discussed.

Researchers are often interested in estimating interactions in multilevel models, but many researchers assume that the same procedures and interpretations for interactions in single-level models apply to multilevel models. However, estimating interactions in multilevel models is much more complex than in single-level models. Because uncentered (RAS) or grand mean centered (CGM) level-1 predictors in two-level models contain two sources of variability (i.e., within-cluster variability and between-cluster variability), interactions involving RAS or CGM level-1 predictors also contain more than one source of variability. In this Master’s thesis, I use simulations to demonstrate that ignoring the four sources of variability in a total level-1 interaction effect can lead to erroneous conclusions. I explain how to parse a total level-1 interaction effect into four specific interaction effects, derive equivalencies between CGM and centering within context (CWC) for this model, and describe how the interpretations of the fixed effects change under CGM and CWC. Finally, I provide an empirical example using diary data collected from working adults with chronic pain.

Currently, there is a clear gap in the missing data literature for three-level models.

To date, the literature has only focused on the theoretical and algorithmic work

required to implement three-level imputation using the joint model (JM) method of

imputation, leaving relatively no work done on fully conditional specication (FCS)

method. Moreover, the literature lacks any methodological evaluation of three-level

imputation. Thus, this thesis serves two purposes: (1) to develop an algorithm in

order to implement FCS in the context of a three-level model and (2) to evaluate

both imputation methods. The simulation investigated a random intercept model

under both 20% and 40% missing data rates. The ndings of this thesis suggest

that the estimates for both JM and FCS were largely unbiased, gave good coverage,

and produced similar results. The sole exception for both methods was the slope for

the level-3 variable, which was modestly biased. The bias exhibited by the methods

could be due to the small number of clusters used. This nding suggests that future

research ought to investigate and establish clear recommendations for the number of

clusters required by these imputation methods. To conclude, this thesis serves as a

preliminary start in tackling a much larger issue and gap in the current missing data

literature.

This dissertation examines a planned missing data design in the context of mediational analysis. The study considered a scenario in which the high cost of an expensive mediator limited sample size, but in which less expensive mediators could be gathered on a larger sample size. Simulated multivariate normal data were generated from a latent variable mediation model with three observed indicator variables, M1, M2, and M3. Planned missingness was implemented on M1 under the missing completely at random mechanism. Five analysis methods were employed: latent variable mediation model with all three mediators as indicators of a latent construct (Method 1), auxiliary variable model with M1 as the mediator and M2 and M3 as auxiliary variables (Method 2), auxiliary variable model with M1 as the mediator and M2 as a single auxiliary variable (Method 3), maximum likelihood estimation including all available data but incorporating only mediator M1 (Method 4), and listwise deletion (Method 5).

The main outcome of interest was empirical power to detect the mediated effect. The main effects of mediation effect size, sample size, and missing data rate performed as expected with power increasing for increasing mediation effect sizes, increasing sample sizes, and decreasing missing data rates. Consistent with expectations, power was the greatest for analysis methods that included all three mediators, and power decreased with analysis methods that included less information. Across all design cells relative to the complete data condition, Method 1 with 20% missingness on M1 produced only 2.06% loss in power for the mediated effect; with 50% missingness, 6.02% loss; and 80% missingess, only 11.86% loss. Method 2 exhibited 20.72% power loss at 80% missingness, even though the total amount of data utilized was the same as Method 1. Methods 3 – 5 exhibited greater power loss. Compared to an average power loss of 11.55% across all levels of missingness for Method 1, average power losses for Methods 3, 4, and 5 were 23.87%, 29.35%, and 32.40%, respectively. In conclusion, planned missingness in a multiple mediator design may permit higher quality characterization of the mediator construct at feasible cost.

Missing data are common in psychology research and can lead to bias and reduced power if not properly handled. Multiple imputation is a state-of-the-art missing data method recommended by methodologists. Multiple imputation methods can generally be divided into two broad categories: joint model (JM) imputation and fully conditional specification (FCS) imputation. JM draws missing values simultaneously for all incomplete variables using a multivariate distribution (e.g., multivariate normal). FCS, on the other hand, imputes variables one at a time, drawing missing values from a series of univariate distributions. In the single-level context, these two approaches have been shown to be equivalent with multivariate normal data. However, less is known about the similarities and differences of these two approaches with multilevel data, and the methodological literature provides no insight into the situations under which the approaches would produce identical results. This document examined five multilevel multiple imputation approaches (three JM methods and two FCS methods) that have been proposed in the literature. An analytic section shows that only two of the methods (one JM method and one FCS method) used imputation models equivalent to a two-level joint population model that contained random intercepts and different associations across levels. The other three methods employed imputation models that differed from the population model primarily in their ability to preserve distinct level-1 and level-2 covariances. I verified the analytic work with computer simulations, and the simulation results also showed that imputation models that failed to preserve level-specific covariances produced biased estimates. The studies also highlighted conditions that exacerbated the amount of bias produced (e.g., bias was greater for conditions with small cluster sizes). The analytic work and simulations lead to a number of practical recommendations for researchers.

Research demonstrating the importance of the paternal role has been largely conducted using samples of Caucasian men, leaving a gap in what is known about fathering in minority cultures. Family systems theories highlight the dynamic interrelations between familial roles and relationships, and suggest that comprehensive studies of fathering require attention to the broad family and cultural context. During the early infancy period, mothers' and fathers' postpartum adjustment may represent a critical source of influence on father involvement. For the current study, Mexican American (MA) women (N = 125) and a subset of their romantic partners/biological fathers (N = 57) reported on their depressive symptoms and levels of father involvement (paternal engagement, accessibility, and responsibility) during the postpartum period. Descriptive analyses suggested that fathers are involved in meaningful levels of care during infancy. Greater paternal postpartum depression (PPD) was associated with lower levels of father involvement. Maternal PPD interacted with paternal gender role attitudes to predict father involvement. At higher levels of maternal PPD, involvement increased among fathers adhering to less segregated gender role attitudes and decreased among fathers who endorsed more segregated gender role attitudes. Within select models, differences in the relations were observed between mothers' and fathers' reports of paternal involvement. Results bring attention to the importance of examining contextual influences on early fathering in MA families and highlight the unique information that may be gathered from separate maternal and paternal reports of father involvement.

The present study examined the association of pain intensity and goal progress in a community sample of 132 adults with chronic pain who participated in a 21 day diary study. Multilevel modeling was employed to investigate the effect of morning pain intensity on evening goal progress as mediated by pain's interference with afternoon goal pursuit. Moderation effects of pain acceptance and pain catastrophizing on the associations between pain and interference with both work and lifestyle goal pursuit were also tested. The results showed that the relationship between morning pain and pain's interference with work goal pursuit in the afternoon was significantly moderated by a pain acceptance. In addition, it was found that the mediated effect differed across levels of pain acceptance; that is: (1) there was a significant mediation effect when pain acceptance was at its mean and one standard deviation below the mean; but (2) there was no mediation effect when pain acceptance was one standard deviation above the mean. It appears that high pain acceptance significantly attenuates the power of nociception in disrupting one's work goal pursuit. However, in the lifestyle goal model, none of the moderators were significant nor was there a significant association between pain interference with goal pursuit and goal progress. Only morning pain intensity significantly predicted afternoon interference with lifestyle goal pursuit. Further interpretation of the present findings and potential explanations of those inconsistencies are elaborated on discussion. Limitations and the clinical implication of the current study were considered, along with suggestions for future studies.

Random Forests is a statistical learning method which has been proposed for propensity score estimation models that involve complex interactions, nonlinear relationships, or both of the covariates. In this dissertation I conducted a simulation study to examine the effects of three Random Forests model specifications in propensity score analysis. The results suggested that, depending on the nature of data, optimal specification of (1) decision rules to select the covariate and its split value in a Classification Tree, (2) the number of covariates randomly sampled for selection, and (3) methods of estimating Random Forests propensity scores could potentially produce an unbiased average treatment effect estimate after propensity scores weighting by the odds adjustment. Compared to the logistic regression estimation model using the true propensity score model, Random Forests had an additional advantage in producing unbiased estimated standard error and correct statistical inference of the average treatment effect. The relationship between the balance on the covariates' means and the bias of average treatment effect estimate was examined both within and between conditions of the simulation. Within conditions, across repeated samples there was no noticeable correlation between the covariates' mean differences and the magnitude of bias of average treatment effect estimate for the covariates that were imbalanced before adjustment. Between conditions, small mean differences of covariates after propensity score adjustment were not sensitive enough to identify the optimal Random Forests model specification for propensity score analysis.

Including a covariate can increase power to detect an effect between two variables. Although previous research has studied power in mediation models, the extent to which the inclusion of a mediator will increase the power to detect a relation between two variables has not been investigated. The first study identified situations where empirical and analytical power of two tests of significance for a single mediator model was greater than power of a bivariate significance test. Results from the first study indicated that including a mediator increased statistical power in small samples with large effects and in large samples with small effects. Next, a study was conducted to assess when power was greater for a significance test for a two mediator model as compared with power of a bivariate significance test. Results indicated that including two mediators increased power in small samples when both specific mediated effects were large and in large samples when both specific mediated effects were small. Implications of the results and directions for future research are then discussed.

Daily dairies and other intensive measurement methods are increasingly used to study the relationships between two time varying variables X and Y. These data are commonly analyzed using longitudinal multilevel or bivariate growth curve models that allow for random effects of intercept (and sometimes also slope) but which do not address the effects of weekly cycles in the data. Three Monte Carlo studies investigated the impact of omitting the weekly cycles in daily dairy data under the multilevel model framework. In cases where cycles existed in both the time-varying predictor series (X) and the time-varying outcome series (Y) but were ignored, the effects of the within- and between-person components of X on Y tended to be biased, as were their corresponding standard errors. The direction and magnitude of the bias depended on the phase difference between the cycles in the two series. In cases where cycles existed in only one series but were ignored, the standard errors of the regression coefficients for the within- and between-person components of X tended to be biased, and the direction and magnitude of bias depended on which series contained cyclical components.