Matching Items (15)
Filtering by
- Genre: Doctoral Dissertation
Description
Although the issue of factorial invariance has received increasing attention in the literature, the focus is typically on differences in factor structure across groups that are directly observed, such as those denoted by sex or ethnicity. While establishing factorial invariance across observed groups is a requisite step in making meaningful cross-group comparisons, failure to attend to possible sources of latent class heterogeneity in the form of class-based differences in factor structure has the potential to compromise conclusions with respect to observed groups and may result in misguided attempts at instrument development and theory refinement. The present studies examined the sensitivity of two widely used confirmatory factor analytic model fit indices, the chi-square test of model fit and RMSEA, to latent class differences in factor structure. Two primary questions were addressed. The first of these concerned the impact of latent class differences in factor loadings with respect to model fit in a single sample reflecting a mixture of classes. The second question concerned the impact of latent class differences in configural structure on tests of factorial invariance across observed groups. The results suggest that both indices are highly insensitive to class-based differences in factor loadings. Across sample size conditions, models with medium (0.2) sized loading differences were rejected by the chi-square test of model fit at rates just slightly higher than the nominal .05 rate of rejection that would be expected under a true null hypothesis. While rates of rejection increased somewhat when the magnitude of loading difference increased, even the largest sample size with equal class representation and the most extreme violations of loading invariance only had rejection rates of approximately 60%. RMSEA was also insensitive to class-based differences in factor loadings, with mean values across conditions suggesting a degree of fit that would generally be regarded as exceptionally good in practice. In contrast, both indices were sensitive to class-based differences in configural structure in the context of a multiple group analysis in which each observed group was a mixture of classes. However, preliminary evidence suggests that this sensitivity may contingent on the form of the cross-group model misspecification.
ContributorsBlackwell, Kimberly Carol (Author) / Millsap, Roger E (Thesis advisor) / Aiken, Leona S. (Committee member) / Enders, Craig K. (Committee member) / Mackinnon, David P (Committee member) / Arizona State University (Publisher)
Created2011
Description
Research has shown that a developmental process of maturing out of alcohol involvement occurs during young adulthood, and that this process is related to both young adult role transitions (e.g., marriage) and personality developmental (e.g., decreased disinhibition and neuroticism). The current study extended past research by testing whether protective marriage and personality effects on maturing out were stronger among more severe late adolescent drinkers, and whether protective marriage effects were stronger among those who experienced more personality development. Parental alcoholism and gender were tested as moderators of marriage, personality, and late adolescent drinking effects on maturing out; and as distal predictors mediated by these effects. Participants were a subsample (N = 844; 51% children of alcoholics; 53% male, 71% non-Hispanic Caucasian, 27% Hispanic; Chassin, Barrera, Bech, & Kossak-Fuller, 1992) from a larger longitudinal study of familial alcoholism. Hypotheses were tested with latent growth models characterizing alcohol consumption and drinking consequence trajectories from late adolescence to adulthood (age 17-40). Past findings were replicated by showing protective effects of becoming married, sensation-seeking reductions, and neuroticism reductions on the drinking trajectories. Moderation tests showed that protective marriage effects on the drinking trajectories were stronger among those with higher pre-marriage drinking in late adolescence (i.e., higher growth intercepts). This might reflect role socialization mechanisms such that more severe drinking produces more conflict with the demands of new roles (i.e., role incompatibility), thus requiring greater drinking reductions to resolve this conflict. In contrast, little evidence was found for moderation of personality effects by late adolescent drinking or for moderation of marriage effects by personality. Parental alcoholism findings suggested complex moderated mediation pathways. Parental alcoholism predicted less drinking reduction through decreasing the likelihood of marriage (mediation) and muting marriage's effect on the drinking trajectories (moderation), but parental alcoholism also predicted more drinking reduction through increasing initial drinking in late adolescence (mediation). The current study provides new insights into naturally occurring processes of recovery during young adulthood and suggests that developmentally-tailored interventions for young adults could harness these natural recovery processes (e.g., by integrating role incompatibility themes and addressing factors that block role effects among those with familial alcoholism).
ContributorsLee, Matthew R. (Author) / Chassin, Laurie (Thesis advisor) / Corbin, William R. (Committee member) / Mackinnon, David P (Committee member) / Presson, Clark C. (Committee member) / Arizona State University (Publisher)
Created2013
Description
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.
ContributorsCham, Hei Ning (Author) / Tein, Jenn-Yun (Thesis advisor) / Enders, Stephen G (Thesis advisor) / Enders, Craig K. (Committee member) / Mackinnon, David P (Committee member) / Arizona State University (Publisher)
Created2013
Description
Fibromyalgia (FM) is a chronic pain condition characterized by debilitating fatigue. This study examined the dynamic relation between interpersonal enjoyment and fatigue in 102 partnered and 74 unpartnered women with FM. Participants provided three daily ratings for 21 days. They rated their fatigue in late morning and at the end of the day. Both partnered and unpartnered participants reported their interpersonal enjoyment in the combined familial, friendship, and work domains (COMBINED domain) in the afternoon. Additionally, partnered participants reported their interpersonal enjoyment in the spousal domain. The study was guided by three hypotheses at the within-person level, based on daily diaries: (1) elevated late morning fatigue would predict diminished afternoon interpersonal enjoyment; (2) diminished interpersonal enjoyment would predict elevated end-of-day fatigue; (3) interpersonal enjoyment would mediate the late morning to end-of-day fatigue relationship. In cross-level models, the study explored whether individual differences (between-person) in late morning fatigue and afternoon interpersonal enjoyment would moderate within-person relations from late morning fatigue to afternoon interpersonal enjoyment, and from afternoon interpersonal enjoyment to end-of-day fatigue. Furthermore, it explored whether the hypothesized relationships at the within-person level would also emerge at the between-person level (between-person mediation models). Multilevel structural equation modeling and multilevel modeling were employed for model testing, separately for partnered and unpartnered participants. Within-person mediation models supported that on high fatigue mornings, afternoon interpersonal enjoyment was dampened in the spousal and combined domains in partnered and unpartnered samples. Moreover, low afternoon interpersonal enjoyment in both the spousal and combined domains predicted elevated end-of-day fatigue. Afternoon interpersonal enjoyment mediated the relationship of late morning to end-of-day fatigue in the combined domain but in not the spousal domain. Cross-level moderation analyses showed that individual differences in afternoon spousal enjoyment moderated the day-to-day relation between afternoon spousal enjoyment and end-of-day fatigue. Finally, the mediational chain was not observed at the between-person level. These findings suggest that preserving interpersonal enjoyment in non-spousal relations limits within-day increases in FM fatigue. They highlight the importance of examining domain-specificity in interpersonal enjoyment when studying fatigue, and suggest that targeting enjoyment in social relations may improve the efficacy of existing treatments.
ContributorsYeung, Wan (Author) / Aiken, Leona S. (Thesis advisor) / Davis, Mary C. (Thesis advisor) / Mackinnon, David P (Committee member) / Zautra, Alex J (Committee member) / Arizona State University (Publisher)
Created2013
Description
Coarsely grouped counts or frequencies are commonly used in the behavioral sciences. Grouped count and grouped frequency (GCGF) that are used as outcome variables often violate the assumptions of linear regression as well as models designed for categorical outcomes; there is no analytic model that is designed specifically to accommodate GCGF outcomes. The purpose of this dissertation was to compare the statistical performance of four regression models (linear regression, Poisson regression, ordinal logistic regression, and beta regression) that can be used when the outcome is a GCGF variable. A simulation study was used to determine the power, type I error, and confidence interval (CI) coverage rates for these models under different conditions. Mean structure, variance structure, effect size, continuous or binary predictor, and sample size were included in the factorial design. Mean structures reflected either a linear relationship or an exponential relationship between the predictor and the outcome. Variance structures reflected homoscedastic (as in linear regression), heteroscedastic (monotonically increasing) or heteroscedastic (increasing then decreasing) variance. Small to medium, large, and very large effect sizes were examined. Sample sizes were 100, 200, 500, and 1000. Results of the simulation study showed that ordinal logistic regression produced type I error, statistical power, and CI coverage rates that were consistently within acceptable limits. Linear regression produced type I error and statistical power that were within acceptable limits, but CI coverage was too low for several conditions important to the analysis of counts and frequencies. Poisson regression and beta regression displayed inflated type I error, low statistical power, and low CI coverage rates for nearly all conditions. All models produced unbiased estimates of the regression coefficient. Based on the statistical performance of the four models, ordinal logistic regression seems to be the preferred method for analyzing GCGF outcomes. Linear regression also performed well, but CI coverage was too low for conditions with an exponential mean structure and/or heteroscedastic variance. Some aspects of model prediction, such as model fit, were not assessed here; more research is necessary to determine which statistical model best captures the unique properties of GCGF outcomes.
ContributorsCoxe, Stefany (Author) / Aiken, Leona S. (Thesis advisor) / West, Stephen G. (Thesis advisor) / Mackinnon, David P (Committee member) / Reiser, Mark R. (Committee member) / Arizona State University (Publisher)
Created2012
Description
Mediation analysis is a statistical approach that examines the effect of a treatment (e.g., prevention program) on an outcome (e.g., substance use) achieved by targeting and changing one or more intervening variables (e.g., peer drug use norms). The increased use of prevention intervention programs with outcomes measured at multiple time points following the intervention requires multilevel modeling techniques to account for clustering in the data. Estimating multilevel mediation models, in which all the variables are measured at individual level (Level 1), poses several challenges to researchers. The first challenge is to conceptualize a multilevel mediation model by clarifying the underlying statistical assumptions and implications of those assumptions on cluster-level (Level-2) covariance structure. A second challenge is that variables measured at Level 1 potentially contain both between- and within-cluster variation making interpretation of multilevel analysis difficult. As a result, multilevel mediation analyses may yield coefficient estimates that are composites of coefficient estimates at different levels if proper centering is not used. This dissertation addresses these two challenges. Study 1 discusses the concept of a correctly specified multilevel mediation model by examining the underlying statistical assumptions and implication of those assumptions on Level-2 covariance structure. Further, Study 1 presents analytical results showing algebraic relationships between the population parameters in a correctly specified multilevel mediation model. Study 2 extends previous work on centering in multilevel mediation analysis. First, different centering methods in multilevel analysis including centering within cluster with the cluster mean as a Level-2 predictor of intercept (CWC2) are discussed. Next, application of the CWC2 strategy to accommodate multilevel mediation models is explained. It is shown that the CWC2 centering strategy separates the between- and within-cluster mediated effects. Next, Study 2 discusses assumptions underlying a correctly specified CWC2 multilevel mediation model and defines between- and within-cluster mediated effects. In addition, analytical results for the algebraic relationships between the population parameters in a CWC2 multilevel mediation model are presented. Finally, Study 2 shows results of a simulation study conducted to verify derived algebraic relationships empirically.
ContributorsTofighi, Davood (Author) / West, Stephen G. (Thesis advisor) / Mackinnon, David P (Thesis advisor) / Enders, Craig C (Committee member) / Millsap, Roger E (Committee member) / Arizona State University (Publisher)
Created2010
Description
Falls are a public health concern for older adults with or without cognitive impairment, including clinical Mild Cognitive Impairment (MCI) and Alzheimer’s Disease (AD) dementia. Executive function (EF) is linked to falls and is notably impaired in individuals with MCI and AD dementia. However, it is unclear which EF assessments are associated with falls in older adults with intact cognition, MCI, and AD dementia. Apolipoprotein E4 (ApoE4) increases the risk of MCI and AD dementia, with limited information suggesting decreased EF and gait impairment in ApoE4 carriers. The purpose of this study was to 1) investigate the relationships between EF assessments and falls (fall history and future fall) by cognitive status (intact cognition, MCI, AD dementia) and 2) determine if ApoE4 moderates the relationship between EF and falls across cognitive status. EF assessments included Digit Span (DS), Trail Making Test (ΔTMT), Stroop Interference Test, Controlled Oral Word Association assessments of phonemic and semantic fluency (COWAC), and Clock Drawing Test. Binary logistic regression was used with secondary data to analyze the EF and falls relationship. Results showed a significant association of the ΔTMT with fall history but may not be appropriate for use with people with AD dementia due to a floor effect. Significant interaction effects were shown between AD dementia and DS, ΔTMT, and EF as a latent variable, where individuals with a fall history had better EF performance. The interaction effects are influenced by the lower percentage of reported falls in people with severe AD dementia. In the second aim, a confounding effect exposed a potential four-way interaction where higher EF in ApoE4 non-carriers with AD dementia experienced a future fall without a fall history. Overall, more research is needed to determine which EF assessments are best suited for fall risk assessment and whether ApoE4 plays a role in the relationship between EF and falls. Since the low percentage of reported falls in people with severe AD dementia exposed a differing trend between EF and falls, more research is needed to develop valid and reliable tools for collecting fall data in individuals with severe AD dementia.
ContributorsDelgado, Ferdinand (Author) / Der Ananian, Cheryl (Thesis advisor) / Yu, Fang (Thesis advisor) / Mackinnon, David P (Committee member) / Peterson, Daniel S (Committee member) / Ofori, Edward (Committee member) / Arizona State University (Publisher)
Created2023
Description
Mediation analysis is used to investigate how an independent variable, X, is related to an outcome variable, Y, through a mediator variable, M (MacKinnon, 2008). If X represents a randomized intervention it is difficult to make a cause and effect inference regarding indirect effects without making no unmeasured confounding assumptions using the potential outcomes framework (Holland, 1988; MacKinnon, 2008; Robins & Greenland, 1992; VanderWeele, 2015), using longitudinal data to determine the temporal order of M and Y (MacKinnon, 2008), or both. The goals of this dissertation were to (1) define all indirect and direct effects in a three-wave longitudinal mediation model using the causal mediation formula (Pearl, 2012), (2) analytically compare traditional estimators (ANCOVA, difference score, and residualized change score) to the potential outcomes-defined indirect effects, and (3) use a Monte Carlo simulation to compare the performance of regression and potential outcomes-based methods for estimating longitudinal indirect effects and apply the methods to an empirical dataset. The results of the causal mediation formula revealed the potential outcomes definitions of indirect effects are equivalent to the product of coefficient estimators in a three-wave longitudinal mediation model with linear and additive relations. It was demonstrated with analytical comparisons that the ANCOVA, difference score, and residualized change score models’ estimates of two time-specific indirect effects differ as a function of the respective mediator-outcome relations at each time point. The traditional model that performed the best in terms of the evaluation criteria in the Monte Carlo study was the ANCOVA model and the potential outcomes model that performed the best in terms of the evaluation criteria was sequential G-estimation. Implications and future directions are discussed.
ContributorsValente, Matthew J (Author) / Mackinnon, David P (Thesis advisor) / West, Stephen G. (Committee member) / Grimm, Keving (Committee member) / Chassin, Laurie (Committee member) / Arizona State University (Publisher)
Created2018
Description
The goal of diagnostic assessment is to discriminate between groups. In many cases, a binary decision is made conditional on a cut score from a continuous scale. Psychometric methods can improve assessment by modeling a latent variable using item response theory (IRT), and IRT scores can subsequently be used to determine a cut score using receiver operating characteristic (ROC) curves. Psychometric methods provide reliable and interpretable scores, but the prediction of the diagnosis is not the primary product of the measurement process. In contrast, machine learning methods, such as regularization or binary recursive partitioning, can build a model from the assessment items to predict the probability of diagnosis. Machine learning predicts the diagnosis directly, but does not provide an inferential framework to explain why item responses are related to the diagnosis. It remains unclear whether psychometric and machine learning methods have comparable accuracy or if one method is preferable in some situations. In this study, Monte Carlo simulation methods were used to compare psychometric and machine learning methods on diagnostic classification accuracy. Results suggest that classification accuracy of psychometric models depends on the diagnostic-test correlation and prevalence of diagnosis. Also, machine learning methods that reduce prediction error have inflated specificity and very low sensitivity compared to the data-generating model, especially when prevalence is low. Finally, machine learning methods that use ROC curves to determine probability thresholds have comparable classification accuracy to the psychometric models as sample size, number of items, and number of item categories increase. Therefore, results suggest that machine learning models could provide a viable alternative for classification in diagnostic assessments. Strengths and limitations for each of the methods are discussed, and future directions are considered.
ContributorsGonzález, Oscar (Author) / Mackinnon, David P (Thesis advisor) / Edwards, Michael C (Thesis advisor) / Grimm, Kevin J. (Committee member) / Zheng, Yi (Committee member) / Arizona State University (Publisher)
Created2018
Description
Research and theory in social psychology and related fields indicates that people simultaneously hold many cultural identities. And it is well evidenced across relevant fields (e.g., sociology, marketing, economics) that salient identities are instrumental in a variety of cognitive and behavioral processes, including decision-making. It is not, however, well understood how the relative salience of various cultural identities factors into the process of making identity-relevant choices, particularly ones that require an actor to choose between conflicting sets of cultural values or beliefs. It is also unclear whether the source of that salience (e.g., chronic or situational) is meaningful in this regard. The current research makes novel predictions concerning the roles of cultural identity centrality and cultural identity situational salience in three distinct aspects of the decision-making process: Direction of decision, speed of decision, and emotion related to decision. In doing so, the research highlights two under-researched forms of culture (i.e., political and religious) and uses as the focal dependent variable a decision-making scenario that forces participants to choose between the values of their religious and political cultures and, to some degree, behave in an identity-inconsistent manner. Results indicate main effects of Christian identity centrality and democrat identity centrality on preference for traditional versus gender-neutral (i.e., non-traditional/progressive) restrooms after statistically controlling for covariates. Additionally, results show a significant main effect of democrat identity centrality and a significant interaction effect of Christian and democrat identity centrality on positive emotion linked to the decision. Post hoc analyses further reveal a significant quadratic relationship between Christian identity centrality and emotion related to the decision. There was no effect of situational strength of democrat identity salience on the decision. Neither centrality or situational strength had any effect on the speed with which participants made their decisions. This research theoretically and empirically advances the study of cultural psychology and carries important implications for identity research and judgment and decision-making across a variety of fields, including management, behavioral economics, and marketing.
ContributorsBarbour, Joseph Eugene (Author) / Cohen, Adam B. (Thesis advisor) / Kenrick, Douglas T. (Committee member) / Mackinnon, David P (Committee member) / Mandel, Naomi (Committee member) / Arizona State University (Publisher)
Created2019