ABSTRACT
Recently, Asparouhov and Muthén Structural Equation Modeling: A Multidisciplinary Journal, 28, 1-14, (2021a, 2021b) proposed a variant of the Wald test that uses Markov chain Monte Carlo machinery to generate a chi-square test statistic for frequentist inference. Because the test's composition does not rely on analytic expressions for sampling variation and covariation, it potentially provides a way to get honest significance tests in cases where the likelihood-based test statistic's assumptions break down (e.g., in small samples). The goal of this study is to use simulation to compare the new MCM Wald test to its maximum likelihood counterparts, with respect to both their type I error rate and power. Our simulation examined the test statistics across different levels of sample size, effect size, and degrees of freedom (test complexity). An additional goal was to assess the robustness of the MCMC Wald test with nonnormal data. The simulation results uniformly demonstrated that the MCMC Wald test was superior to the maximum likelihood test statistic, especially with small samples (e.g., sample sizes less than 150) and complex models (e.g., models with five or more predictors). This conclusion held for nonnormal data as well. Lastly, we provide a brief application to a real data example.
Subject(s)
Markov Chains , Monte Carlo Method , Humans , Likelihood Functions , Linear Models , Computer Simulation , Models, Statistical , Data Interpretation, Statistical , Sample SizeABSTRACT
OBJECTIVE: Although emerging studies examine the inverse relationship between body satisfaction and disordered eating for Black women, it has not been established how racially salient aspects of body satisfaction may have implications for eating behaviors and longitudinal health outcomes. METHOD: In a longitudinal sample of 455 Black women, we examined whether skin color satisfaction across ages 10-15 was directly related to adult health outcomes at age 40 (e.g., disordered eating, self-esteem, self-reported health, depressive symptoms, and cardiovascular risk). We also investigated the indirect impact of skin color satisfaction on adult health, mediated by body satisfaction, and binge eating. RESULTS: No significant direct or indirect effects of adolescent skin color satisfaction were observed for depressive symptoms or cardiovascular health outcomes. At ages 10 and 12, skin color satisfaction had negative and positive direct effects, respectively, on self-esteem. At age 15, greater skin color satisfaction was directly associated with greater self-reported health. Post hoc analyses revealed that when additionally accounting for adolescent body satisfaction, greater skin color satisfaction was indirectly associated with greater self-esteem and self-reported health, alongside lower cardiovascular risk. CONCLUSIONS: Although previous research suggests that in adolescence, Black girls' skin color satisfaction affects both body satisfaction and disordered eating behaviors, this association does not hold into midlife. Rather, post hoc analyses suggest that the lasting effects of adolescent skin color satisfaction are mediated by the longitudinal stability of body satisfaction, which in turn, is associated with adult health outcomes. (PsycInfo Database Record (c) 2024 APA, all rights reserved).
Subject(s)
Bulimia , Feeding and Eating Disorders , Adult , Humans , Female , Adolescent , Skin Pigmentation , Self Concept , Bulimia/psychology , Feeding and Eating Disorders/epidemiology , Personal Satisfaction , Outcome Assessment, Health Care , Body Image/psychologyABSTRACT
Estimating power for multilevel models is complex because there are many moving parts, several sources of variation to consider, and unique sample sizes at Level 1 and Level 2. Monte Carlo computer simulation is a flexible tool that has received considerable attention in the literature. However, much of the work to date has focused on very simple models with one predictor at each level and one cross-level interaction effect, and approaches that do not share this limitation require users to specify a large set of population parameters. The goal of this tutorial is to describe a flexible Monte Carlo approach that accommodates a broad class of multilevel regression models with continuous outcomes. Our tutorial makes three important contributions. First, it allows any number of within-cluster effects, between-cluster effects, covariate effects at either level, cross-level interactions, and random coefficients. Moreover, we do not assume orthogonal effects, and predictors can correlate at either level. Second, our approach accommodates models with multiple interaction effects, and it does so with exact expressions for the variances and covariances of product random variables. Finally, our strategy for deriving hypothetical population parameters does not require pilot or comparable data. Instead, we use intuitive variance-explained effect size expressions to reverse-engineer solutions for the regression coefficients and variance components. We describe a new R package mlmpower that computes these solutions and automates the process of generating artificial data sets and summarizing the simulation results. The online supplemental materials provide detailed vignettes that annotate the R scripts and resulting output. (PsycInfo Database Record (c) 2023 APA, all rights reserved).
ABSTRACT
The purpose of the current study was to test the longitudinal association between disordered eating symptoms (body dissatisfaction, drive for thinness and bulimia) in adolescence (ages 12, 14, 16, 18, 19) and adulthood (age 40) in a sample of 883 white and Black women. We also investigated moderation by race. Adolescent symptoms at each time point significantly predicted adulthood symptoms for the body dissatisfaction and drive for thinness subscales, for both Black and white women. Bulimia symptoms in adolescence predicted symptoms in adulthood; however, the effect was largely driven by white women. Although moderation was non-significant, among white women, bulimia symptoms at all adolescent time points predicted adulthood bulimia, but among Black women, only symptoms at ages 18 and 19 were predictive of adulthood bulimia. Results suggest that both Black and white women are susceptible to disordered eating and that symptoms emerging in adolescence can potentially follow women into midlife.
ABSTRACT
Composite scores are an exceptionally important psychometric tool for behavioral science research applications. A prototypical example occurs with self-report data, where researchers routinely use questionnaires with multiple items that tap into different features of a target construct. Item-level missing data are endemic to composite score applications. Many studies have investigated this issue, and the near-universal theme is that item-level missing data treatment is superior because it maximizes precision and power. However, item-level missing data handling can be challenging because missing data models become very complex and suffer from the same "curse of dimensionality" problem that plagues the estimation of psychometric models. A good deal of recent missing data literature has focused on advancing factored regression specifications that use a sequence of regression models to represent the multivariate distribution of a set of incomplete variables. The purpose of this paper is to describe and evaluate a factored specification for composite scores with incomplete item responses. We used a series of computer simulations to compare the proposed approach to gold standard multiple imputation and latent variable modeling approaches. Overall, the simulation results suggest that this new approach can be very effective, even under extreme conditions where the number of items is very large (or even exceeds) the sample size. A real data analysis illustrates the application of the method using software available on the internet. (PsycInfo Database Record (c) 2023 APA, all rights reserved).
ABSTRACT
The year 2022 is the 20th anniversary of Joseph Schafer and John Graham's paper titled "Missing data: Our view of the state of the art," currently the most highly cited paper in the history of Psychological Methods. Much has changed since 2002, as missing data methodologies have continually evolved and improved; the range of applications that are possible with modern missing data techniques has increased dramatically, and software options are light years ahead of where they were. This article provides an update on the state of the art that catalogs important innovations from the past two decades of missing data research. The paper addresses topics described in the original paper, including developments related to missing data theory, full information maximum likelihood, Bayesian estimation, multiple imputation, and models for missing not at random processes. The paper also describes newer factored regression specifications and missing data handling for multilevel models, both of which have been a focus of recent research. The paper concludes with a summary of the current software landscape and a discussion of several practical issues. (PsycInfo Database Record (c) 2023 APA, all rights reserved).
ABSTRACT
A growing body of literature has focused on missing data methods that factorize the joint distribution into a part representing the analysis model of interest and a part representing the distributions of the incomplete predictors. Relatively little is known about the utility of this method for multilevel models with interactive effects. This study presents a series of Monte Carlo computer simulations that investigates Bayesian and multiple imputation strategies based on factored regressions. When the model's distributional assumptions are satisfied, these methods generally produce nearly unbiased estimates and good coverage, with few exceptions. Severe misspecifications that arise from substantially non-normal distributions can introduce biased estimates and poor coverage. Follow-up simulations suggest that a Yeo-Johnson transformation can mitigate these biases. A real data example illustrates the methodology, and the paper suggests several avenues for future research.
Subject(s)
Bayes Theorem , Data Interpretation, Statistical , Regression Analysis , Multilevel Analysis , Computer SimulationABSTRACT
BACKGROUND: The U.S. Food and Drug Administration identifies abstinence and the absence of heavy drinking days as outcomes for pharmacotherapy trials for alcohol use disorder (AUD). However, many individuals with AUD struggle to achieve these outcomes, which may discourage them from seeking treatment. World Health Organization (WHO) risk drinking levels have garnered attention in the alcohol field as potential non-abstinent outcomes for AUD medication trials. Further, testing combination pharmacotherapy for AUD represents an important direction in the field, particularly using medications such as naltrexone and varenicline, which are approved for treating AUD and smoking, respectively. The objective of the current study was to test the utility of the WHO risk drinking levels as a drinking outcome in a randomized clinical trial of combined varenicline and naltrexone for smoking cessation and drinking reduction. These analyses provide additional tests of the efficacy of this combination treatment. METHODS: The current study is a secondary analysis of a phase 2, randomized, double-blind clinical trial, wherein participants (N = 165) who were daily smokers and heavy drinkers were randomly assigned to receive either 2 mg/day of varenicline plus 50 mg/day of naltrexone or 2 mg/day of varenicline plus placebo for 12 weeks. Medication effects on 1- and 2-level reductions in WHO risk drinking levels were assessed at 4, 8, and 12 weeks into the active medication period. RESULTS: In logistic growth curve models individuals receiving the combined treatment had greater reductions in WHO risk drinking levels than individuals taking varenicline alone when assessed at 4 weeks into the active medication period. Among individuals who were WHO high and very high risk drinkers at baseline, the largest effect sizes favoring combination treatment were at Week 4 for the WHO 2-level reduction outcome (Cohen's h = 0.202) and Week 12 for the WHO 1-level reduction outcome (Cohen's h = 0.244), although these effects did not reach statistical significance. CONCLUSIONS: These findings provide evidence that combined varenicline plus naltrexone treatment is effective at reducing WHO risk drinking levels, particularly among individuals who smoke cigarettes daily and drink heavily. These results add to a growing body of literature validating reductions in WHO risk drinking levels as outcomes of alcohol medication trials.
Subject(s)
Alcoholism , Naltrexone , Humans , Varenicline/therapeutic use , Naltrexone/therapeutic use , Double-Blind Method , Alcoholism/drug therapy , Alcohol Drinking/drug therapy , World Health Organization , Treatment OutcomeABSTRACT
Depression and anxiety are highly prevalent psychological disorders; our understanding of these conditions remains limited. Efforts to explain anxiety and depression have been constrained in part by binary classification systems. Dimensional approaches to understanding psychopathology may be more effective. The present study used latent profile analysis (LPA) to assess whether unique subgroups exist within a tri-level model of anxiety and depression. Participants (N=627) completed self-report questionnaires from which tri-level model factors were derived. LPA was conducted on those factors. A 4-profile model offered optimal fit to the data at baseline. This model was replicated at a second time point. Models derived included profiles labelled 'Mixed Fears,' 'Anxious Arousal,' 'Low Mood/Anhedonia,' and 'Sub-Clinical.' Profiles were validated at Time 1 using diagnostic status and clinical severity ratings associated with mood and anxiety presentations. Profiles demonstrated flexibility in accommodating breadth in clinical presentations and common comorbidities. Latent variable models may offer more ecologically valid approaches to understanding psychopathology.
ABSTRACT
Although it has been demonstrated that (a) body dissatisfaction and internalization of societal appearance standards contribute to disordered eating and (b) that internalization of societal appearance standards leads to decreased skin color satisfaction among Black women, it has not been established whether skin color dissatisfaction contributes to disordered eating among Black women or girls. The objective of the present study is to determine the influence of skin color satisfaction as a potential predictor for binge eating, and its effect through body image in Black girls during the vulnerable developmental period of adolescence. Using data from ten annual measurements in 1213 Black girls across ages 10-19, we sought to determine whether skin color satisfaction predicts Binge Eating Disorder (BED) risk and symptoms using pre-registered logistic and multilevel models. We found that lower skin color satisfaction at ages 13 and 14 significantly predicted greater odds of BED and lower skin color satisfaction at all ages predicted greater BED symptoms. Body satisfaction mediated the relationship between skin color satisfaction and BED symptoms. Our results suggest that skin color dissatisfaction is a novel component of body image for Black girls that is also related to binge eating.
Subject(s)
Binge-Eating Disorder , Bulimia , Feeding and Eating Disorders , Adolescent , Adult , Body Image/psychology , Child , Female , Humans , Personal Satisfaction , Prospective Studies , Skin Pigmentation , Young AdultABSTRACT
OBJECTIVE: Uveal melanoma, a rare eye cancer, presents potential vision loss and life threat. This prospective, longitudinal study interrogated the predictive utility of visual impairment, as moderated by optimism/pessimism, on depressive symptoms in 299 adults undergoing diagnostic evaluation. METHOD: Depressive symptoms (Center for Epidemiologic Studies Depression Scale), subjective (Measure of Outcome in Ocular Disease vision subscale) and objective (logarithm of the minimum angle of resolution) visual impairment, and optimism/pessimism (Life Orientation Test-Revised) were assessed before diagnostic evaluation and 1 week, 3 months, and 12 months after diagnosis. Multilevel modeling, with repeated measures (Level 1) nested within individuals (Level 2) and imputation of missing data (Blimp software), was performed. RESULTS: Depressive symptoms were significantly more elevated 1 week after diagnosis in cancer patients (n = 107) versus patients not diagnosed with cancer (n = 192). Higher subjective (but not objective) visual impairment predicted greater depressive symptoms (p < .001). Across the entire sample, the two-way (Optimism/Pessimism × Subjective Visual Impairment) interactions were statistically significant (ps < .05), but not the three-way interaction (with diagnosis). The positive association between subjective visual impairment and depressive symptoms was significant at low and moderate levels of optimism (ps < .001), but not at high optimism (p > .05). The association was significant at high and moderate levels (ps < .001), but not low (p > .05) levels of pessimism. CONCLUSIONS: Elevated depressive symptoms are evident in adults who do (vs. do not) receive a diagnosis of uveal melanoma but appear to remit within 3 months. Perceived impaired vision, especially coupled with low optimism or high pessimism, predicts depressive symptoms over time, with implications for intervention. (PsycInfo Database Record (c) 2021 APA, all rights reserved).
Subject(s)
Depression , Melanoma , Optimism , Pessimism , Uveal Neoplasms , Vision Disorders , Adult , Depression/epidemiology , Humans , Longitudinal Studies , Melanoma/diagnosis , Melanoma/psychology , Optimism/psychology , Pessimism/psychology , Prospective Studies , Uveal Neoplasms/diagnosis , Uveal Neoplasms/psychology , Vision Disorders/psychologyABSTRACT
Structural equation modeling (SEM) applications routinely employ a trilogy of significance tests that includes the likelihood ratio test, Wald test, and score test or modification index. Researchers use these tests to assess global model fit, evaluate whether individual estimates differ from zero, and identify potential sources of local misfit, respectively. This full cadre of significance testing options is not yet available for multiply imputed data sets, as methodologists have yet to develop a general score test for this context. Thus, the goal of this article is to outline a new score test for multiply imputed data. Consistent with its complete-data counterpart, this imputation-based score test provides an estimate of the familiar expected parameter change statistic. The new procedure is available in the R package semTools and naturally suited for identifying local misfit in SEM applications (i.e., a model modification index). The article uses a simulation study to assess the performance (Type I error rate, power) of the proposed score test relative to the score test produced by full information maximum likelihood (FIML) estimation. Due to the two-stage nature of multiple imputation, the score test exhibited slightly lower power than the corresponding FIML statistic in some situations but was generally well calibrated. (PsycInfo Database Record (c) 2020 APA, all rights reserved).
Subject(s)
Data Interpretation, Statistical , Latent Class Analysis , Models, Statistical , Psychology/methods , HumansABSTRACT
Despite the broad appeal of missing data handling approaches that assume a missing at random (MAR) mechanism (e.g., multiple imputation and maximum likelihood estimation), some very common analysis models in the behavioral science literature are known to cause bias-inducing problems for these approaches. Regression models with incomplete interactive or polynomial effects are a particularly important example because they are among the most common analyses in behavioral science research applications. In the context of single-level regression, fully Bayesian (model-based) imputation approaches have shown great promise with these popular analysis models. The purpose of this article is to extend model-based imputation to multilevel models with up to 3 levels, including functionality for mixtures of categorical and continuous variables. Computer simulation results suggest that this new approach can be quite effective when applied to multilevel models with random coefficients and interaction effects. In most scenarios that we examined, imputation-based parameter estimates were quite accurate and tracked closely with those of the complete data. The new procedure is available in the Blimp software application for macOS, Windows, and Linux, and the article includes a data analysis example illustrating its use. (PsycINFO Database Record (c) 2020 APA, all rights reserved).
Subject(s)
Data Interpretation, Statistical , Models, Statistical , Multilevel Analysis , Psychology/methods , Regression Analysis , HumansABSTRACT
Literature addressing missing data handling for random coefficient models is particularly scant, and the few studies to date have focused on the fully conditional specification framework and "reverse random coefficient" imputation. Although it has not received much attention in the literature, a joint modeling strategy that uses random within-cluster covariance matrices to preserve cluster-specific associations is a promising alternative for random coefficient analyses. This study is apparently the first to directly compare these procedures. Analytic results suggest that both imputation procedures can introduce bias-inducing incompatibilities with a random coefficient analysis model. Problems with fully conditional specification result from an incorrect distributional assumption, whereas joint imputation uses an underparameterized model that assumes uncorrelated intercepts and slopes. Monte Carlo simulations suggest that biases from these issues are tolerable if the missing data rate is 10% or lower and the sample is composed of at least 30 clusters with 15 observations per group. Furthermore, fully conditional specification tends to be superior with intraclass correlations that are typical of crosssectional data (e.g., ICC = .10), whereas the joint model is preferable with high values typical of longitudinal designs (e.g., ICC = .50).
Subject(s)
Data Interpretation, Statistical , Models, Statistical , Multilevel Analysis , HumansABSTRACT
In the present study, the predictors and outcomes associated with the trajectories of peer rejection were examined in a longitudinal sample of Italian children (338 boys, 269 girls) ages 10 to 14 years. Follow-up assessments included 60% of the original sample at age 16-17. Low, medium, and high rejection trajectory groups were identified using growth mixture models. Consistent with previous studies, we found that (a) being less prosocial and more physically aggressive at age 10 was characteristic of those children with the high rejection trajectory; (b) being less attractive was related to higher peer rejection from age 10 to 14; and (c) boys with a high rejection trajectory showed high levels of delinquency and anxiety-depression and low levels of academic aspiration at age 16-17, whereas girls with a high rejection trajectory showed low levels of academic aspiration and social competence at age 16-17. Our findings indicate the detrimental consequences of peer rejection on children's development and adjustment and shed light on the mechanisms that contribute to maintaining or worsening (e.g., being attractive, prosocial, and aggressive) a child's negative status (e.g., being rejected) within his or her peer group over time.
Subject(s)
Adolescent Behavior/physiology , Adolescent Development/physiology , Aggression/physiology , Beauty , Child Development/physiology , Peer Group , Rejection, Psychology , Social Adjustment , Social Behavior , Adolescent , Child , Female , Follow-Up Studies , Humans , Longitudinal Studies , Male , Models, Statistical , Psychological DistanceABSTRACT
Specialized imputation routines for multilevel data are widely available in software packages, but these methods are generally not equipped to handle a wide range of complexities that are typical of behavioral science data. In particular, existing imputation schemes differ in their ability to handle random slopes, categorical variables, differential relations at Level-1 and Level-2, and incomplete Level-2 variables. Given the limitations of existing imputation tools, the purpose of this manuscript is to describe a flexible imputation approach that can accommodate a diverse set of 2-level analysis problems that includes any of the aforementioned features. The procedure employs a fully conditional specification (also known as chained equations) approach with a latent variable formulation for handling incomplete categorical variables. Computer simulations suggest that the proposed procedure works quite well, with trivial biases in most cases. We provide a software program that implements the imputation strategy, and we use an artificial data set to illustrate its use. (PsycINFO Database Record
Subject(s)
Behavioral Research/methods , Data Interpretation, Statistical , Models, Statistical , Multilevel Analysis/methods , HumansABSTRACT
Multiple imputation has enjoyed widespread use in social science applications, yet the application of imputation-based inference to structural equation modeling has received virtually no attention in the literature. Thus, this study has 2 overarching goals: evaluate the application of Meng and Rubin's (1992) pooling procedure for likelihood ratio statistic to the SEM test of model fit, and explore the possibility of using this test statistic to define imputation-based versions of common fit indices such as the TLI, CFI, and RMSEA. Computer simulation results suggested that, when applied to a correctly specified model, the pooled likelihood ratio statistic performed well as a global test of model fit and was closely calibrated to the corresponding full information maximum likelihood (FIML) test statistic. However, when applied to misspecified models with high rates of missingness (30%-40%), the imputation-based test statistic generally exhibited lower power than that of FIML. Using the pooled test statistic to construct imputation-based versions of the TLI, CFI, and RMSEA worked well and produced indices that were well-calibrated with those of full information maximum likelihood estimation. This article gives Mplus and R code to implement the pooled test statistic, and it offers a number of recommendations for future research. (PsycINFO Database Record
Subject(s)
Data Interpretation, Statistical , Models, Psychological , Models, Statistical , Psychology/methods , HumansABSTRACT
For almost 50 years, psychologists have been theorizing about and measuring religiosity essentially the way Gordon Allport did, when he distinguished between intrinsic and extrinsic religiosity. However, there is a historical debate regarding what this scale actually measures, which items should be included, and how many factors or subscales exist. To provide more definitive answers, we estimated a series of confirmatory factor analysis models comparing four competing theories for how to score Gorsuch and McPherson's commonly used measure of intrinsic and extrinsic religiosity. We then formally investigated measurement invariance across U.S. Protestants, Irish Catholics, and Turkish Muslims and across U.S. Protestants, Catholics, and Muslims. We provide evidence that a five-item version of intrinsic religiosity is invariant across the U.S. samples and predicts less warmth toward atheists and gay men/lesbians, validating the scale. Our results suggest that a variation of Gorsuch and McPherson's measure may be appropriate for some but not all uses in cross-cultural research.
Subject(s)
Religion and Psychology , Adolescent , Adult , Cross-Cultural Comparison , Female , Humans , Male , Models, Psychological , Prejudice , Social Identification , Surveys and Questionnaires , Young AdultABSTRACT
In Ordinary Least Square regression, researchers often are interested in knowing whether a set of parameters is different from zero. With complete data, this could be achieved using the gain in prediction test, hierarchical multiple regression, or an omnibus F test. However, in substantive research scenarios, missing data often exist. In the context of multiple imputation, one of the current state-of-art missing data strategies, there are several different analogous multi-parameter tests of the joint significance of a set of parameters, and these multi-parameter test statistics can be referenced to various distributions to make statistical inferences. However, little is known about the performance of these tests, and virtually no research study has compared the Type 1 error rates and statistical power of these tests in scenarios that are typical of behavioral science data (e.g., small to moderate samples, etc.). This paper uses Monte Carlo simulation techniques to examine the performance of these multi-parameter test statistics for multiple imputation under a variety of realistic conditions. We provide a number of practical recommendations for substantive researchers based on the simulation results, and illustrate the calculation of these test statistics with an empirical example.
Subject(s)
Data Interpretation, Statistical , Multilevel Analysis , Multivariate Analysis , Academic Success , Adolescent , Behavioral Research/methods , Child Behavior Disorders/diagnosis , Computer Simulation , Factor Analysis, Statistical , Humans , Likelihood Functions , Monte Carlo Method , Reading , Regression Analysis , Risk , SoftwareABSTRACT
The last 20 years has seen an uptick in research on missing data problems, and most software applications now implement one or more sophisticated missing data handling routines (e.g., multiple imputation or maximum likelihood estimation). Despite their superior statistical properties (e.g., less stringent assumptions, greater accuracy and power), the adoption of these modern analytic approaches is not uniform in psychology and related disciplines. Thus, the primary goal of this manuscript is to describe and illustrate the application of multiple imputation. Although maximum likelihood estimation is perhaps the easiest method to use in practice, psychological data sets often feature complexities that are currently difficult to handle appropriately in the likelihood framework (e.g., mixtures of categorical and continuous variables), but relatively simple to treat with imputation. The paper describes a number of practical issues that clinical researchers are likely to encounter when applying multiple imputation, including mixtures of categorical and continuous variables, item-level missing data in questionnaires, significance testing, interaction effects, and multilevel missing data. Analysis examples illustrate imputation with software packages that are freely available on the internet.