The current study proposed a new model, termed the cross-classified multiple membership latent variable regression (CCMM-LVR) model, to be utilized for multiple membership data structures (for example, in the presence of student mobility across schools) that provides an extension to the three-level latent variable regression model (HM3-LVR). The HM3-LVR model is beneficial for testing more flexible, directional hypotheses about growth trajectory parameters and handles pure clustering of participants within higher-level units. However, the HM3-LVR model involves the assumption that students remain in the same cluster (school) throughout the duration of the time period of interest. The CCMM-LVR model, on the other hand, appropriately models the participants’ changing clusters over time. The first purpose of this study was to demonstrate use and interpretation of the CCMM-LVR model and its parameters with a large-scale longitudinal dataset that had a multiple membership data structure (i.e., student mobility). The impact of ignoring mobility in the real data was investigated by comparing parameter estimates, standard error estimates, and model fit indices for the two estimating models (CCMM-LVR and HM3-LVR). The second purpose of the dissertation was to conduct a simulation study to try to understand the source of potential differences between the two estimating models and find out which model’s estimates were closer to the truth given the conditions investigated. The manipulated conditions in the simulation study included the mobility rate, number of clustering units, number of individuals (i.e., students) per cluster (here, school), and number of measurement occasions per individual. The outcomes investigated in the simulation study included relative parameter bias, relative standard error bias, root mean square error, and coverage rates of the 95% credible intervals. Substantial bias was found across conditions for both models, but the CCMM-LVR model resulted in the least amount of relative parameter bias and more efficient estimates of the parameters, especially for larger numbers of clustering units. The results of the real data and simulation studies are discussed, along with the implications for applied researchers for when to consider using the CCMM-LVR model versus the misspecified HM3-LVR model. / text

This study was undertaken to evaluate gender differences in predictors of substance use in clinic-referred, Hispanic American adolescents with substance use disorders. Individual (disruptive behavior disorders, depression) and family variables (family conflict, parental monitoring) were evaluated as predictors of the initial level and change over time in marijuana use, and gender was evaluated as a moderator of the associations. The study involved an analysis of an existing dataset of 113 Hispanic American adolescents (93 boys; age 12 to 17) referred for outpatient treatment for substance abuse and their parental guardian. Participants and parental guardians completed questionnaires and a structured interview to report on predictor variables at baseline and marijuana use at baseline and 3-, 6-, 12-, and 18-months post-baseline. Latent growth curve modeling was conducted to evaluate the study hypotheses. Adolescents reported high levels of marijuana use at baseline and relatively stable levels of marijuana use over time. Treatment and gender effects influenced the marijuana use trajectory. Girls exhibited more impaired psychosocial functioning than boys, including worse disruptive behavior problems and depression and lower levels of parental monitoring. Depression was negatively associated with marijuana use longitudinally. Overall, individual and family risk factors are associated with adolescent marijuana use in complex ways. Implications for intervention are discussed.

The current study evaluated a multiple-membership growth curve model that can be used to model growth in student achievement, in the presence of student mobility. The purpose of the study was to investigate the impact of ignoring multiple school membership when modeling student achievement across time. Part one of the study consisted of an analysis of real longitudinal student achievement data. This real data analysis compared parameter estimates, standard error estimates, and model-fit statistics obtained from a growth curve model that ignores multiple membership, to those obtained from a growth model that accounts for multiple school membership via the MMREM approach. Part two of the study consisted of a simulation study designed to determine the impact of ignoring multiple membership and the accuracy of parameter estimates obtained under the two modeling approaches, under a series of data conditions. The goal of the study was to assess the importance of incorporating a more flexible MMREM approach when modeling students’ academic achievement across time. Overall, the results of the current study indicated that the Cross-classified multiple membership growth curve model (CCMM-GCM) may provide more accurate parameter estimates than competing approaches for a number of data conditions. Both modeling approaches, however, yielded substantially biased estimates of parameters for some experimental conditions. Overall, results demonstrate that incorporating student mobility into achievement growth modeling can result in more accurate estimates of schools effects. / text

Using the perspective of sexual script theory (Gagnon & Simon, 1973) and growth curve modeling, this study examined whether characteristics of parents and parent-adolescent connectedness influence change in lifetime number of sexual partners from adolescence through young adulthood. Living in a blended family, having at least one college-educated parent and on-time parent-adolescent sexual communication positively predicted later lifetime number of sexual partners. Parent religiosity and parent-adolescent connectedness negatively predicted later lifetime number of sexual partners. Parent-adolescent sexual communication that focused on negative consequences of sex and parent disapproval of adolescent sexual activity were not significant in the overall model. Control variables included adolescent race/ethnicity, gender, physical maturity, marriage history, virginity pledge history, and expectations of positive consequences of sex. Physical maturity and gender were not significant in the overall model. In conclusion, parents have significant and far-reaching influence on their children's later sexual behavior. This study extended research in the field by examining lifetime number of sexual partners across four time points, which allowed observation of change in this outcome variable with age and accounted for the nested nature of the data. / Master of Science

Adolescence is a period of heightened risk for substance use and heightened vulnerability to substance exposure. Yet, little is known about how socioeconomic status (SES) influences adolescent decision making and behavior across development to add to these risks. This prospective longitudinal study used latent growth curve modeling (GCM) to examine the contributions of SES on adolescent delay discounting and substance use in a sample of 167 adolescents (52% male). Confirmatory factor analysis (CFA) was used to compute SES factor scores across three waves using a composite of parent and spouse education years and combined annual household income. Adolescent delay discounting and substance use were measured annually across three waves. The main goal of this study is to examine how SES may explain individual differences in growth trajectories of delay discounting and substance use. We used parallel process growth curve modeling with SES as a time-varying and time-invariant covariate to examine the associations between adolescent SES, delay discounting, and substance use onset as well as frequency. These results reveal that delay discounting exhibits a declining linear trend across adolescent development whereas cigarette, alcohol, marijuana, and polysubstance use exhibit increasing linear trends across adolescent development. Furthermore, low SES (as a time-invariant covariate) may lead to earlier onset adolescent alcohol and polysubstance use by way of heightened levels of delay discounting. These findings suggest that delay discounting interventions may be a promising avenue for reducing socioeconomic disparities in early onset alcohol and polysubstance use, while delay discounting development is still underway. / Master of Science / Adolescence is a period of heightened risk for substance use and heightened vulnerability to the effects of substances. Yet, little is known about how socioeconomic status (SES) influences adolescent decision making and behavior to add to these risks. This study used latent growth curve modeling (GCM) to examine the role of SES on adolescent decision making and substance use in a sample of 167 adolescents (52% male). Confirmatory factor analysis (CFA) was used to compute SES factor scores across three time points using an average of parent and spouse education years and income. Adolescent delay discounting and substance use were measured annually across three time points. The main goal of this study is to examine how SES may explain individual differences in delay discounting and substance use across adolescence. We used parallel process growth curve modeling with SES as a time-varying and time-invariant covariate to examine the links between adolescent SES, delay discounting, and substance use age of onset and frequency. These results reveal that delay discounting shows linear decreases in growth across adolescence whereas cigarette, alcohol, marijuana, and polysubstance use show increasing linear growth across adolescence. Additionally, low SES may lead to earlier onset adolescent alcohol and polysubstance use by way of heightened levels of delay discounting. These findings suggest that delay discounting interventions may help reduce socioeconomic differences in early onset alcohol and polysubstance use, while delay discounting development is still in progress.

abstract: Research on the consequences of gang membership is limited mainly to the study of crime and victimization. This gives the narrow impression that the effects of gang membership do not cascade into other life domains. This dissertation conceptualized gang membership as a snare in the life-course that disrupts progression in conventional life domains. National Longitudinal Survey of Youth Cohort of 1997 (NLSY97) data were used to examine the effects of adolescent gang membership on the nature and patterns of educational attainment and employment over a 12-year period in the life-course. Variants of propensity score weighting were used to assess the effects of gang joining on a range of outcomes pertaining to educational attainment and employment. The key findings in this dissertation include: (1) selection adjustments partially or fully confounded the effects of gang joining; despite this (2) gang joiners had 70 percent the odds of earning a high school diploma and 42 percent the odds of earning a 4-year college degree than matched individuals who avoided gangs; (3) at the 11-year mark, the effect of gang joining on educational attainment exceeded one-half year; (4) gang joiners made up for proximate deficits in high school graduation and college matriculation, but gaps in 4-year college degree and overall educational attainment gained throughout the study; (5) gang joiners were less likely to be employed and more likely to not participate in the labor force, and these differences accelerated toward the end of the study; (6) gang joiners spent an additional one-third of a year jobless relative to their matched counterparts; and (7) the cumulative effect of gang joining on annual income exceeded $14,000, which was explained by the patterning of joblessness rather than the quality of jobs. The theoretical and policy implications of these findings, as well as directions for future research, are addressed in the concluding chapter of this dissertation. / Dissertation/Thesis / Ph.D. Criminology and Criminal Justice 2012

The therapeutic alliance is a core element of successful treatment in therapy. Recent literature has explored variables that predict the alliance at various time points during therapy, but few studies have explored how the alliance develops over time and the factors that influence its rate of change. The current study addresses these questions by estimating latent growth-curve models to analyze how male and female partners' alliance scores develop over time and how adverse childhood experiences (ACEs) impact the development of the alliance during the first six sessions of therapy. Results from these analyses show that neither men nor women's ACEs had a significant effect on the rate of change in the alliance. A discussion of the attachment implications of these findings, as well as the limitations of this study and potential directions for future research are then presented.

Across time, the influence of parents and peers appears to change. Early in life, parents have a stronger influence on the development of youth than do their peers. This, however, will change as an individual ages. Using longitudinal data from the Marion County (Oregon) Youth Survey (1964-1979), I examine the influence of parents or delinquent association, drug use and arrest. Analysis generated through latent growth curve modeling show that although parental influence appears to decrease significantly later in life, parental attachment delays the formation of delinquent peer networks, thereby indirectly reducing the total number of arrests. Even so, reductions in parental influence over time were associated with a significantly accelerated rate of acquiring delinquent peers and hence, with an increased frequency of arrest and drug use. The available evidence thus suggests that parental attachment has initial inhibitory effects on the formation of peer networks but only limited long-term developmental effects.

No description available.

This research examines the viability of a need-based approach that models the age-trajectories of healthcare utilization. We propose a fundamentally different way of treating age in modeling healthcare use. Rather than treating age as a need indicator, we refocus modeling efforts to predicting the age-trajectories of healthcare use. Using inpatient hospital utilization data from the Discharge Abstract Database, first, we model the age-trajectories of the rate of hospital use employing a common functional form. Second, we assess variation in these age-trajectories using growth curve modeling. Third, we explain variation in these age-trajectories using census variables. Our analysis shows that the regional variation in the age-trajectories of the rate of inpatient hospital use is sufficient to justify this method, and could be partially explained using census variables. This indicates that modeling age-trajectories of healthcare use is advantageous, and the current need-based approach may benefit from this new modeling strategy.