Many observational studies assessing the effects of treatments or exposures are limited to comparisons between treatment users and nonusers or exposed and unexposed participants at study entry. However, the underlying and etiologically relevant exposure may gradually increase over time before reaching some plateau. This amount of time required for this latent cumulative exposure to reach a maximum hazard will be referred to as the "lag", coming from the concept that the association between exposure and outcome is lagged or delayed. Accounting for the lag is essential when analyzing exposure-response associations adequately. My challenge was to simultaneously estimate the lag-time and the exposure's lagged-association with the outcome at plateau. In this dissertation, I draw an analogy with the pharmacokinetic one-compartment model (OCM). OCM describes the accumulation of a medication in the body based on an exponential cumulative density function whose rate of increase is defined by a half-life parameter. Upon discontinuation, the OCM assumes that a medication will eliminate at the same half-life rate. The decline, for my purposes, can be interpreted as the time to return to a null effect of exposure, which occurs at roughly 4-5 half-lives. My methods model the association of a latent exposure and dichotomous outcome using a half-life of effect, similar to the OCM, in longitudinal analyses of single and repeated exposures. I derive profile likelihood-based algorithms to estimate of the upper limit of association simultaneously with the rate of latent exposure growth towards or away from plateau. Lastly, I extend this approach to allow different half-life parameters for incline and decline. Using simulations, I analyze the performance of my approach by comparing bias and coverage of the estimates for the half-life and effect parameters. With data from the Black Women's Health Study Cohort (a prospective cohort of 59,000 women followed 1995-2015), I show that prolonged cigarette smoking is associated with a maximum hazard of cardiovascular disease (CVD) at 2.5 times the hazard of never smokers. Additionally, I estimate that it takes about 7 years of smoking cessation for an individual's hazard of CVD to decrease by 50%. / 2020-11-13T00:00:00Z

Interest in mediation analysis has increased over time, with particular excitement in the social and behavioral sciences. A mediator is defined as an intermediate in the causal sequence between an independent and dependent variable. Previous research has demonstrated that the cross-sectional form of mediation analysis is inherently flawed, evidenced by the inability of the cross-sectional mediation model to account for temporal precedence and estimation of the indirect effect being biased in nearly all situations. For these reasons, a longitudinal model is recommended. However, a method for determining the exact time points to measure the variables used in mediation analysis has not been adequately examined. In this study, we examined methods for determining an appropriate time lag when designing a mediation study. The methods implemented include correlation analysis, the quadratic and exponential forms of the lag as a moderator approach, and knot estimation using basis splines. The data for the study was simulated for three distinct trends generated using a linear piecewise model, a sigmoid model, and a sigmoid piecewise model. Additionally, two sampling approaches, an intense sampling approach and a three-measure approach, were examined as well as six sample sizes and three effect sizes for the total effect on the outcome. The estimation methods were additionally compared by considering different types of error structures used in data generation as well as by examining equal and unequal time lag lengths between the predictor and mediator, and the mediator and outcome. The intent of the study is to provide methods so that researchers can estimate the best time to evaluate mediator and outcome measurements that will be used in mediation analysis. The results from this study showed that the best estimation method varied depending on the lag being estimated, the sampling approach, and the length of the lag. However, the knot estimation approach worked reasonably well in most scenarios considered even with small sample sizes of 5 or 10 per group. The findings from this study have the potential to improve study design for research implementing longitudinal mediation analysis by reducing bias in the estimate of the indirect effect when adequate time points are used.

<p>In many spatio-temporal applications a vector of covariates is measured alongside a spatio-temporal response. In such cases, the purpose of the statistical model is to quantify the change, in expectation or otherwise, in the response due to a change in the predictors while adequately accounting for the spatio-temporal structure of the response, the predictors, or both. The most common approach for building such a model is to confine the relationship between the response and the predictors to a single spatio-temporal coordinate. For spatio-temporal problems, however, the relationship between the response and predictors may not be so confined. For example, spatial models are often used to quantify the effect of pollution exposure on mortality. Yet, an unknown lag exists between time of exposure to pollutants and mortality. Furthermore, due to mobility and atmospheric movement, a spatial lag between pollution concentration and mortality may also exist (e.g. subjects may live in the suburbs where pollution levels are low but work in the city where pollution levels are high).</p><p>The contribution of this thesis is to propose a hierarchical modeling framework which captures complex spatio-temporal relationships between responses and covariates. Specifically, the models proposed here use kernels to capture spatial and/or temporal lagged effects. Several forms of kernels are proposed with varying degrees of complexity. In each case, however, the kernels are assumed to be parametric with parameters that are easily interpretable and estimable from the data. Full distributional results are given for the Gaussian setting along with consequences of model misspecification. The methods are shown to be effective in understanding the complex relationship between responses and covariates through various simulated examples and analyses of physical data sets.</p> / Dissertation

None

No description available.

Master of Science / Department of Family Studies and Human Services / Jared R. Anderson / The present study tests the reciprocal relationships between marital attitudes, time spent together, and marital satisfaction in a sample of newlyweds (N = 1220). Using cognitive dissonance theory and the investment model, this study tests two sets of theoretically derived unidirectional pathways to provide empirical evidence for longitudinal associations between these three constructs. Two separate models were tested, one for husbands, and one for wives. For both husbands and wives, results of the panel models indicated significant autoregressive paths among all three variables from T1 to T2, T2 to T3, and T1 to T3. Among husbands, significant cross-lagged paths emerged between T1 marital satisfaction and T2 marital attitudes, T1 marital satisfaction and T2 time spent together, T1 time spent together and T2 marital satisfaction, T2 time spent together and T3 marital satisfaction, and T2 time spent together and T3 marital attitudes. For wives, cross-lagged paths between T1 marital satisfaction and T2 time spent together, T1 time spent together and T2 marital satisfaction, and T2 time spent together and T3 marital satisfaction were found to be significant. Bootstrap test for indirect effects resulted in no significant mediating effects in the models. The results of these models are discussed and implications for future research and intervention are given.

The purpose of this study was to take a multidimensional perspective to prosocial behavior and self-regulation by analyzing longitudinal, bidirectional relations between prosocial behavior toward strangers, friends, and family members and behavioral, cognitive, and emotional dimensions of self-regulation across adolescence. Participants included reports from 500 adolescents (age Time 1 = 12, Time 2 = 14, Time 3 = 16, Time 4 = 18; 52% female, 77% European American) taking part in the Flourishing Families Project. Nine cross-lagged panel models were conducted analyzing longitudinal associations between each target of prosocial behavior and each dimension of self-regulation. Results revealed that in early adolescence, prosocial behavior toward strangers and cognitive self-regulation were bidirectionally related. Prosocial behavior toward strangers was significantly associated with cognitive self-regulation from age 12 to age 18 and cognitive self-regulation was significantly associated with prosocial behavior toward friends across adolescence. Further, behavioral and emotional self-regulation were significantly related to prosocial behavior toward family from age 12 to age 18. Gender was significantly associated with initial levels of study variables but was not significantly relate to patterns of association. Discussion focuses on how findings fit into existent theory and research.

A number of preschool children engage in consistent problem behaviors that place them at-risk for developing emotional and behavioral disorders. These problem behaviors have been associated with poorer short and long-term outcomes for young children. Teacher-child relationships (i.e. conflict and closeness) and positive interactions between teachers and children may be reciprocally associated with problem behavior (e.g. teacher-child closeness reducing problem behavior). The purpose of this study was to explore the longitudinal bi-directional relations between teacher-child relationships, teacher-child interactions, and problem behavior over a single school year. Using a cross-lagged panel model, data from a larger randomized control trial of the BEST in CLASS program, was examined across three time points, and differences based on intervention participation were examined. Results indicated that there were far fewer paths in the business as usual group compared to the BEST in CLASS group. As expected, in the comparison group, problem behavior at Time 1 predicted lower levels of closeness at Time 2, however, this relation was not significant in the BEST in CLASS group. In the BEST in CLASS model problem behavior at Time 1 negatively predicted Time 2 positive interactions and there was a cross-lagged association with problem behavior at Time 1 predicting higher levels of conflict at Time, which in turn predicted higher levels of problem behavior at Time 3. Additional findings, limitations and implications for intervention work, practice, and policy are discussed.

Dans un contexte d'amélioration des moteurs aéronautiques en termes de consommation et de pollution, les simulations numériques apparaissent comme un outil intéressant pour mieux comprendre et modéliser les phénomènes turbulents qui se produisent dans les turbomachines. La simulation aux grandes échelles (SGE) d’un étage de turbomachine à des conditions réalistes (nombre de Mach, nombre de Reynolds…) reste toutefois hors de portée dans le cadre industriel. La méthode chorochronique, aujourd’hui largement utilisée pour les calculs URANS, permet de réduire le coût des simulations numériques, mais elle implique de stocker le signal aux frontières du domaine pendant une période complète de l’écoulement. Le stockage direct de l’information étant exclu étant donné la taille des maillages et les pas de temps mis en jeu, la solution la plus courante actuellement est de décomposer le signal sous la forme de séries de Fourier. Cette solution ne retient du signal qu’une fréquence fondamentale (la fréquence de passage de la roue opposée) et un nombre limité d’harmoniques. Dans le cadre d’une SGE, elle implique donc une grande perte d’énergie, et le filtrage des phénomènes décorrélés de la vitesse de rotation comme par exemple un lâcher tourbillonnaire. Le remplacement de la décomposition en séries de Fourier par une décomposition aux valeurs propres (POD pour Proper Orthogonal Decomposition) permet de stocker le signal aux interfaces sans faire d’hypothèse sur les fréquences contenues dans le signal et donc de réduire la perte d’énergie liée à l’utilisation d’un modèle réduit. La compression s’effectue en supprimant les plus petites valeurs singulières et les vecteurs associés. Cette nouvelle méthode est validée sur la simulation URANS d'étages de turbomachines et comparée aux conditions classiques utilisant les séries de Fourier et à des calculs de références contenant plusieurs aubes par roue. Elle est ensuite appliquée à la simulation aux grandes échelles de l'écoulement d'un cylindre. Les erreurs causées par l'hypothèse chorochronique et par la compression sont séparées et on montre que l'utilisation de la POD permet de réduire de moitié le filtrage des fluctuations de vitesses par rapport aux séries de Fourier pour un même taux de compression. Enfin, la simulation aux grandes échelles d'un étage de turbomachine avec des conditions chorochroniques POD est réalisée afin de valider la méthode dans le cadre d'une configuration industrielle. / The more and more restrictive standards in terms of fuel consumption and pollution for aircraft engines lead to a constant improvement of their design. Numerical simulations appear as an interesting tool for a better understanding and modeling of the turbulent phenomena which occur in turbomachinery. The large-eddy simulation (LES) of a turbomachinery stage at realistic conditions (Mach number, Reynolds number...) remains out of reach for industrial congurations. The phase-lagged method, widely used for unsteady Reynolds-averaged Navier--Stockes (URANS) calculations, is a good candidate to reduce the computational cost. However, it needs to store the signal at all the boundaries over a full passage of the opposite blade. A direct storage of the information being excluded given the size of the mesh grid and timesteps involved, the most used solution currently is to decompose the signal into Fourier series. This solution retains the fundamental frequency of the signal (the opposite blade passage frequency) and a limited number of harmonics. In the frame of a LES, as the spectra are broadband, it implies a loss of energy. Replacing the Fourier series decomposition by a proper orthogonal decomposition (POD) allows the storage of the signal at the interfaces without making any assumptions on the frequency content of the signal, and helps to reduce the loss of energy caused by the phase lagged method. The compression is done by removing the smallest singular values and the associated vectors. This new method is first validated on the URANS simulations of turbomachinery stages and compared with Fourier series-based conditions and references calculations with multiple blades per row. It is then applied to the large eddy simulation of the flow around a cylinder. The error caused by the phase-lagged assumption and compression are separated and it is showed that the use of the POD allows to halve the filtering of the velocity fluctuations with respect to the Fourier series, for a given compression rate. Finally, the large eddy simulation of a compressor stage with POD phase-lagged conditions is carried out to validate the method for realistic turbomachinery configurations.

abstract: The current study investigated the dynamic interplay of career decision ambiguity tolerance and career indecision over three assessment times in a sample of college students (n=583). While the previous research has repeatedly shown an association of career decision ambiguity tolerance with career indecision, the direction of this association has not been adequately assessed with longitudinal investigation. It was hypothesized in this study that there is a reciprocal pattern of career decision ambiguity tolerance leading to subsequent career indecision and career indecision leading to subsequent career decision ambiguity tolerance. Using a cross-lagged panel design, this study found support for the reciprocal pattern that aversion with ambiguity led to increased negative experience, choice anxiety, and lack of readiness in career decision making, while negative experience, choice anxiety, and lack of readiness led to increased aversion with ambiguity as well. Additionally, this study revealed that choice anxiety and readiness for career decision making led to increased interests in new information. The key findings were discussed with respect to the theoretical and clinical implications for career counseling along with limitations and suggestions for future research. / Dissertation/Thesis / Doctoral Dissertation Counseling Psychology 2017