Baurley JW, Ervin CM, Witkiewitz K
… +3 more, Claus E, Levy M, McMahan CS
Multivariate Behav Res
· 2026 Jun · PMID 42377435
·
Publisher ↗
Alcohol use disorder (AUD) research faces significant challenges in capturing individual heterogeneity and complex temporal patterns in drinking behaviors. Standard statistical methods fail to account for within-person v...Alcohol use disorder (AUD) research faces significant challenges in capturing individual heterogeneity and complex temporal patterns in drinking behaviors. Standard statistical methods fail to account for within-person variability and between-person differences, while existing machine learning algorithms are not designed for hierarchical, longitudinal data structures common in AUD research. We developed a comprehensive R package implementing 30 Bayesian machine learning functions specifically designed for alcohol use research, spanning interpretable linear and logistic regression to flexible Bayesian additive regression trees (BART), all with mixed-effects and time-trend extensions. We demonstrate the package capabilities using alcohol use data from two studies: the ABQDrinQ longitudinal cohort ( = 190) and the COMBINE clinical trial ( = 1,383). Key findings include strong associations between concurrent substance use and alcohol consumption (nicotine use associated with 13.5% increase in drinks and 14 times higher odds of drinking), discovery of nonlinear age effects on drinking variability (peak at ages 25-30), and high-accuracy daily predictions (median correlation 0.82, median absolute error 1.0 drinks). The Bayesian framework provides uncertainty quantification essential for both research and clinical applications, while the range of algorithms allows researchers to navigate the complexity-interpretability tradeoff. While developed for alcohol research, the methodological framework addresses statistical challenges common across substance use research.
Multivariate Behav Res
· 2026 Jun · PMID 42364132
·
Publisher ↗
Current models for assessing response accuracy and response times in typically overlook variations in speed and ability within individuals. Instead, they often treat these as residual variances, missing the dynamic chan...Current models for assessing response accuracy and response times in typically overlook variations in speed and ability within individuals. Instead, they often treat these as residual variances, missing the dynamic changes in individual performance throughout a test. Additionally, the influence of item position and the ordinal nature of response accuracy remain underexplored. This paper introduces a comprehensive modelling framework that integrates item responses, response times, and item position to better understand skill acquisition and latent speed changes. Our approach uses the Bivariate Generalised Linear Item Response Theory (B-GLIRT) model, capturing the dual impact of ability on response accuracy and the interplay between ability and speed on response times. We extend this model by incorporating random effects for item and individual-specific variations. The proposed model further explores how item positioning affects test performance and provides diagnostic insights into individual differences. The paper also discusses parameter estimation, model identification, and applications to real-world data, illustrating the practical implications of our findings in computer-based learning assessments.
Multivariate Behav Res
· 2026 Jun · PMID 42329748
·
Publisher ↗
Measuring daily romantic relationship quality is important for understanding conflict, support, and satisfaction processes in near real-time. Although there now exists a great deal of research on determinants and outcome...Measuring daily romantic relationship quality is important for understanding conflict, support, and satisfaction processes in near real-time. Although there now exists a great deal of research on determinants and outcomes of daily relationship quality, these investigations often rely on several untested assumptions regarding sources of, and consistency in, variability. In this study, we tested the variance components for daily relationship quality, how consistent these measurements are, and whether they are impacted by individual differences in attachment security. Six daily, nonexperimental self-reports of relationship satisfaction and functioning (e.g., conflict) from 101 couples were analyzed using generalizability theory. Results demonstrate a large amount of residual in daily reports of relationship quality, yet significant portions of variance attributable to person- and day-level processes. Variance proportions also changed depending on the index of relationship functioning examined, and individual differences in attachment security moderated these results. Lastly, daily reports of relationship satisfaction are relatively stable when submitted to models using persons as the object of measurement. These findings have implications for planning and interpreting future diary and ecological momentary assessment/experience sampling method studies.
Multivariate Behav Res
· 2026 Jun · PMID 42308021
·
Publisher ↗
Recognizing that complex networks of skills typically exhibit hierarchical and modular organization, this article presents a Modularized Higher-Order Diagnostic Classification Model (MHO-DCM) designed to capture hierarch...Recognizing that complex networks of skills typically exhibit hierarchical and modular organization, this article presents a Modularized Higher-Order Diagnostic Classification Model (MHO-DCM) designed to capture hierarchical relationships among attributes organized into clustered subdomains. Central to the proposed method is a representation of attribute hierarchies in which attributes are grouped into cognitively coherent subgraphs nested within a single higher-order ability continuum. We adopt a nominal response model framework in item response theory and leverage standard maximum likelihood estimation (MLE). In parallel, we demonstrate that sequential higher-order latent structural models can likewise be implemented in a modularized fashion within an MLE framework. The performance of the proposed models is examined through simulation studies assessing parameter recovery, classification accuracy, and null rejection rates of goodness-of-fit measures. An empirical demonstration showcases how the framework can be applied in practice, highlighting its advantages in flexibility, interpretability, and the richer diagnostic insights it affords.
Multivariate Behav Res
· 2026 Jun · PMID 42273875
·
Publisher ↗
Generalizability is a perennial concern in randomized studies. While randomized studies are the gold standard for establishing causality, study samples are rarely representative of broader populations due to factors such...Generalizability is a perennial concern in randomized studies. While randomized studies are the gold standard for establishing causality, study samples are rarely representative of broader populations due to factors such as convenience sampling, participant self-selection, and researchers' inclusion or exclusion criteria. To strengthen the generalizability of randomized experiments, the framework of causal effect generalizability provides a solution. However, existing methods require accessing representative individual-level data from the target population, which is often unavailable due to limited resources, data access restrictions, or privacy concerns. In this paper, we develop a novel method to generalize causal effects using only summary statistics on covariates from the target population. We illustrate the estimator using a real-world study by generalizing the impact of a climate change behavioral intervention from the study sample to a broader population. By avoiding the need for individual-level data from the target population, our method offers a practical tool for generalizing causal findings from randomized studies. We hope that the proposed method helps build more accurate theories and enhance the policy relevance of behavioral and psychological research.
Sun RW, Chang F, Yang W
… +2 more, Cheung SF, Cheung SH
Multivariate Behav Res
· 2026 Jun · PMID 42227640
·
Publisher ↗
Standardization is used in many common methods in psychology to enhance the interpretability of results. For example, the so-called "betas" are usually reported in structural equation modeling (SEM). However, there are t...Standardization is used in many common methods in psychology to enhance the interpretability of results. For example, the so-called "betas" are usually reported in structural equation modeling (SEM). However, there are three situations in which standardization, as is usually conducted by SEM programs, makes the results difficult to interpret and sometimes even misleading: standardizing dummy variables, standardizing the product term in moderation, and standardizing a variable that is already measured on a meaningful unit. Another problem is the confidence interval of the standardized results: It is either not reported or computed using methods known to be biased or suboptimal. The R package betaselectr was developed to help users get standardized coefficients properly in the three situations above, along with appropriate standard errors and confidence intervals that take into account the sampling errors in the standard deviations used to do the standardization, and not only in structural equation modeling but also in multiple regression.
Multivariate Behav Res
· 2026 May · PMID 42187198
·
Publisher ↗
Meta-analysts frequently encounter missing covariate values, which can complicate valid estimation of meta-regression models. In practice, missing data are managed often through ad hoc deletion approaches, which can redu...Meta-analysts frequently encounter missing covariate values, which can complicate valid estimation of meta-regression models. In practice, missing data are managed often through ad hoc deletion approaches, which can reduce the validity of statistical inferences. More advanced missing data handling approaches such as multiple imputation (MI) remain underutilized, particularly in meta-analyses with dependent effect sizes within studies. This study expands the use of MI techniques for handling missing covariates in such contexts. Specifically, this study introduces adapted multilevel MI approaches that are expected to better accommodate the structure of meta-analytic data with dependent effect sizes and associated model. The study presents Monte Carlo computer simulations that compare the performance of different MI techniques including single-level and multilevel agnostic MI methods as well as multilevel substantive model-based MI and ad hoc deletion approaches. The results generally supported the use of the MI approach over deletion approaches when the dependent structure is well specified in the imputation procedure. This study demonstrates the feasibility of the use of MI techniques and underscores the importance of incorporating the hierarchical structure into the meta-regression model when analyzing dependent effect sizes with missing covariate values. Implications of the study and directions for future research are discussed.
Multivariate Behav Res
· 2026 May · PMID 42138153
·
Publisher ↗
Recently, there have been growing efforts in developing fair algorithms for treatment effect estimation and optimal treatment recommendations to mitigate discriminatory biases against disadvantaged groups. While most of...Recently, there have been growing efforts in developing fair algorithms for treatment effect estimation and optimal treatment recommendations to mitigate discriminatory biases against disadvantaged groups. While most of this work has focused on addressing discrimination due to individual-level sensitive variables (e.g., race/ethnicity), it overlooks the broader impact of societal structures and cultural norms (e.g., structural racism) beyond the individual level. In this paper, we formalize the concept of multilevel fairness for estimating heterogeneous treatment effects to improve fairness in optimal policies. Specifically, we propose a general framework for the estimation of conditional average treatment effects under multilevel fairness constraints that incorporate sensitive variables from multiple structural levels. Using this framework, we analyze the tradeoff between fairness and the maximum achievable utility by the optimal policy. We evaluate the effectiveness of our framework through a simulation study and a real data study on advanced math courses using data from the High School Longitudinal Study of 2009.
Langenberg B, Helm JL, McCabe CJ
… +2 more, Günther T, Mayer A
Multivariate Behav Res
· 2026 May · PMID 42105287
·
Publisher ↗
This article demonstrates the application of residual dynamic structural equation modeling (RDSEM) for analyzing custom contrasts in experimental factorial designs. Previous applications of RDSEM have often focused on ec...This article demonstrates the application of residual dynamic structural equation modeling (RDSEM) for analyzing custom contrasts in experimental factorial designs. Previous applications of RDSEM have often focused on ecological momentary assessment and daily diary data. However, RDSEM was explicitly developed for intensive longitudinal data more generally, including settings with very short time intervals between observations. Beyond these types of studies, RDSEM is also well suited for analyzing data from laboratory studies such as eye-tracking or reaction time experiments. We compare three analytic approaches, namely analysis of variance, linear mixed models, and RDSEM, emphasizing the unique advantages of RDSEM. Although often applied to momentary assessment data, RDSEM proves highly effective for experimental analysis, offering the ability to integrate both time-varying and time-invariant covariates, model autoregressive effects, and capture interindividual differences in residual variances / intraindividual variability. These strengths arise from RDSEM's integration of time-series, multilevel, and latent variable modeling, all implemented through Bayesian estimation.
In standardized tests, examinees are likely to engage in either one or more following test behaviors: solution behavior, rapid guessing behavior, cheating behavior, nonresponse behavior, etc. Examinees do not always resp...In standardized tests, examinees are likely to engage in either one or more following test behaviors: solution behavior, rapid guessing behavior, cheating behavior, nonresponse behavior, etc. Examinees do not always response all items with solution behavior due to various reasons (such as time constraint or low motivation). Aside from solution behavior, rapid guessing, cheating or nonresponse behavior can result in aberrant responses and inaccurate estimates of examinees' ability or trait, as well as item parameters, thus undermining the validity and fairness of the test. To address this issue, this paper aims to propose an IRTree model to that simultaneously considers rapid guessing, cheating and nonresponse behaviors in order to model the various behaviors exhibited by examinees. The proposed model offers a notable improvement over previous studies, as it provides additional classifications for examinee behaviors at both item and examinee levels. Furthermore, it is the first model to separate and simultaneously model guessing and cheating. Two real data sets are utilized to demonstrate the reasonableness and superiority of the proposed model. Subsequently, two simulation studies based on these real data sets are conducted to validate, revealing that it provide more precise estimates of person and item parameters compared to existing models, and explored the boundary condition of model application.
Sherlock P, Mansolf M, Hofheimer J
… +9 more, Hockett CW, O'Connor TG, Roubinov D, Graff JC, Lai JS, Bush NR, Wright RJ, Chiu YM, program collaborators for Environmental Influences on Child Health Outcomes
Multivariate Behav Res
· 2026 May · PMID 42063395
·
Full text
The goal of this study was to investigate the contextual nature of prenatal depression (PND) and postpartum depression (PPD). We report an investigation of maternal PND and PPD using nonrandomly clustered data from 8,936...The goal of this study was to investigate the contextual nature of prenatal depression (PND) and postpartum depression (PPD). We report an investigation of maternal PND and PPD using nonrandomly clustered data from 8,936 mothers in 16 cohorts in the Environmental influences on Child Health Outcomes (ECHO) Cohort. We used a recursive partitioning algorithm (ctree) to account for differential effects arising from predictor and effect heterogeneity across cohort to identify contexts, and the characteristics comprising these contexts, associated with PND and PPD. Consistent with Bronfenbrenner's bioecological systems framework, risk factors for PND and PPD are heterogeneously and synergistically present in context: the same variables did not consistently present as risk factors across groups. Findings from this study support the idea that the presence or absence of medical (and sociodemographic) risks do not have simple associations with perinatal depression. To that end, no univariate predictors of PND and PPD can be said to constitute risk absent the broader context of mothers' lives. Thus, a more useful research question than "which variables increase risk for PPD?" is "in what contexts, or in which subgroups, does a particular risk or set of risks contribute to perinatal depression?"
Multivariate Behav Res
· 2026 May · PMID 42063392
·
Publisher ↗
This paper considers the problem of causal mediation analysis (CMA) when the outcome, mediator, or both are modeled as latent variables that are measured with error from multiple indicators. Traditional structural equati...This paper considers the problem of causal mediation analysis (CMA) when the outcome, mediator, or both are modeled as latent variables that are measured with error from multiple indicators. Traditional structural equation modeling approaches rely on restrictive parametric assumptions and struggle to capture nonlinear relationships or interactions among covariates, outcomes, and mediators; additionally, accounting for measurement error is difficult when nonlinearities are present. We address these challenges using Bayesian causal mediation forests to model the structural relationships, which were shown by Linero and Zhang to perform well for estimating mediation effects. To infer the latent variables, we consider (a) a full hierarchical Bayesian model and (b) an approximation based on composite scores that is easier to implement. We evaluate our approach through simulation experiments and apply our methodology to data from the Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE) study, which is a publicly available, large-scale, multi-site randomized controlled trial designed to evaluate the effectiveness of cognitive training interventions in order to improve the cognitive abilities of older adults. We find a nonlinear relationship between the ability to perform daily activities (the outcome) and reasoning ability (the mediator), with reasoning training improving the outcome through the mediator.
Multivariate Behav Res
· 2026 May · PMID 42063352
·
Publisher ↗
It is well-known that bifactor structures are over-represented as preferred solutions in measurement modeling. This study explores the extent to which unmodeled clustering of observations in larger social or organization...It is well-known that bifactor structures are over-represented as preferred solutions in measurement modeling. This study explores the extent to which unmodeled clustering of observations in larger social or organization units (e.g., students clustered in schools) offers a partial explanation for this phenomenon. We investigate the overlap between bifactor confirmatory factor models and multilevel confirmatory factor analysis fit to identically structured data formats. Structural symmetries between these models are identified, leading to a series of postulates regarding expected differences across modeling frameworks. Next, through simulation and empirical data analysis, we demonstrate that bifactor solutions can emerge as artifacts of conflated level-1 and level-2 effects when clustering is ignored, causing invalid interpretations of factors. Specifically, results suggest that when bifactor models are fit to clustered data with one level-2 factor and multiple level-1 factors, general factor loadings are typically inflated, leading to greater support for the misspecified bifactor solution. We encourage researchers to consider multilevel measurement models as alternative explanations for bifactor solutions, so factors are accurately interpreted at the correct level of analysis.
Establishing the correct partial measurement invariance model is crucial for ensuring unbiased comparisons of relationships between latent variables across multiple groups. While traditional approaches rely on detecting...Establishing the correct partial measurement invariance model is crucial for ensuring unbiased comparisons of relationships between latent variables across multiple groups. While traditional approaches rely on detecting noninvariant items followed by estimation of structural relationships, more recently, approaches that estimate latent parameters without prior knowledge of anchor items have been developed. Specifically, regularization and alignment are powerful approaches that can be used to estimate multiple group structural models. This study compares a traditional sequential search based on multiple-group CFA (MGCFA) to alignment, lasso, elastic net, and ridge regression for estimating the correlation and means between latent variables without pre-specifying anchor items. In the simulation study, we varied the percentage, magnitude, and pattern of noninvariance, sample size, number of indicators, and correlation value and evaluated the bias and efficiency of the methods in terms of the recovery of the factor correlation, means, and item parameters for a two-group model. Results indicated that elastic net led to less biased and more efficient estimates under higher proportions of noninvariance, while alignment performed better under low to modest noninvariance. We provide recommendations for researchers estimating latent correlations and means under different levels of measurement invariance.
The last two decades have seen a dramatic increase in using intensive longitudinal data to capture psychological processes. Intensive longitudinal data allow researchers to study intraindividual change and variability. M...The last two decades have seen a dramatic increase in using intensive longitudinal data to capture psychological processes. Intensive longitudinal data allow researchers to study intraindividual change and variability. Multiple modeling approaches have been developed to examine these dynamics in a process as it unfolds over time. What is often not considered in these models are factors that can influence the dynamics of the given process. In this article, we describe a state space model to examine the dynamics of daily affect and combine it with covariates that moderate the parameters describing such dynamics. In our approach, the moderators represent sentiment values from open responses to a daily questionnaire. Unlike standard Likert-type measures, open text allows individuals to express their thoughts and feelings without numerical or wording restrictions. We apply natural language processing to quantify positive and negative sentiment associated with such written responses reported daily. The implemented model utilizes a functional relationship between the variability in dynamic parameters and a time-varying covariate. The target of the moderator covariates are the autoregressive and cross-lag parameters in a vector autoregressive model. We find that such moderation effects from the covariates are small, yet robust. However, when the covariates are specified as predictors of the process instead of the dynamic parameters, their effects are strong. Our analyses show that, overall, qualitative measures are valuable to help understand dynamic processes.
Zhang L, Rahal C, Kanopka K
… +3 more, Ulitzsch E, Zhang Z, Domingue BW
Multivariate Behav Res
· 2026 Mar · PMID 41910334
·
Publisher ↗
Confirmatory Factor Analysis (CFA) has been widely used to assess the fit of theoretical measurement models to observed data. We introduce the InterModel Vigorish (IMV) to the field; a predictive fit index that offers no...Confirmatory Factor Analysis (CFA) has been widely used to assess the fit of theoretical measurement models to observed data. We introduce the InterModel Vigorish (IMV) to the field; a predictive fit index that offers novel perspectives for model comparison. The IMV complements traditional fit indices by offering additional information to support model evaluation, with a particular emphasis on a model's generalizability to the hold-out data. It also yields an interpretable and intuitive metric that facilitates meaningful comparisons. We extend it into the CFA framework with binary outcomes and conduct four simulation studies to evaluate its effectiveness. The simulation results suggest that IMV effectively gauges model misspecification, offering insights both at the scale and item levels. As designed, it is insensitive to changes in sample size. By focusing on predictive accuracy, the IMV discourages overfitting. It also enables item-level comparisons, offering richer diagnostic information. To facilitate the practical application of IMV, we offer an empirical example that demonstrates its efficacy in applied research. The paper is accompanied by an R package to further advance the use of the IMV in the CFA space.
Multivariate Behav Res
· 2026 Mar · PMID 41910251
·
Publisher ↗
Conventional growth curve models, often fitted to sum or mean scores of scale responses, do not account for potential changes in item measurement unrelated to construct growth (i.e. differential item functioning; DIF). A...Conventional growth curve models, often fitted to sum or mean scores of scale responses, do not account for potential changes in item measurement unrelated to construct growth (i.e. differential item functioning; DIF). An untested assumption is that the construct is stably measured over time. When this assumption is incorrect, estimates of construct change obtained from conventional growth models may be biased. To address this issue, we recently proposed a new, flexible second-order growth model based upon a longitudinal extension of moderated nonlinear factor analysis (MNLFA; Chen & Bauer, 2024) that allows for DIF from categorical or continuous covariates that may or may not vary over time (e.g. sex, age, age sex). Further, we applied Bayesian regularization to evaluate DIF effects across multiple sources simultaneously without imposing item equality assumptions (i.e. anchor items). In this paper, we present a simulation study to validate the model's performance in detecting DIF over time and between groups. Results indicate that the proposed approach effectively detects DIF without predetermined anchor items and avoids the biased growth estimates consistently observed for conventional models fitted to mean scores. We demonstrate the utility of the method in an empirical example on child externalizing behaviors.