In this work, we introduce a multiple-group longitudinal IRT model that accounts for skewed latent trait distributions. Our approach extends the model proposed by Santos et al. in 2022, which introduced a general class o...In this work, we introduce a multiple-group longitudinal IRT model that accounts for skewed latent trait distributions. Our approach extends the model proposed by Santos et al. in 2022, which introduced a general class of longitudinal IRT models. The latent traits follow a multivariate skew-normal distribution, induced by an antedependence structure with centered skew-normal errors. Additionally, latent mean trajectories are modeled using quadratic curves, while structured covariance matrices capture within-participant dependencies. A three-parameter probit model is employed for dichotomous items. Bayesian parameter estimation and model fit assessment are conducted through a hybrid MCMC algorithm, combining the FFBS sampler with Metropolis-Hastings steps. The model's effectiveness is demonstrated through an application to real data from the Longitudinal Study of the 2005 School Generation in Brazil (GERES project), where it outperforms the normal model by better capturing asymmetry in latent traits. A simulation study further supports its robustness across various test conditions.
This study addresses the challenge of doubly-censoring effects in longitudinal data structures, particularly within latent growth curve models (LGCMs). Censoring can severely bias estimates and inferences, distorting the...This study addresses the challenge of doubly-censoring effects in longitudinal data structures, particularly within latent growth curve models (LGCMs). Censoring can severely bias estimates and inferences, distorting the relationships between growth factors and covariates. To combat this issue, this study introduces the Generalized Tobit estimator (GBIT), an advancement of the conventional Tobit model, designed to handle mixed censoring effects in longitudinal data. The objectives of this study were threefold: (a) to develop GBIT for doubly-censored data, (b) to evaluate GBIT's performance in LGCMs under mixed censoring, and (c) to examine the impact of such censoring on covariate effects and outcomes within LGCMs. A Monte Carlo simulation was conducted to assess GBIT's effectiveness to handle doubly-censoring effects in the LGCM framework, demonstrating its ability to provide unbiased estimates even in the presence of significant censoring. Also, GBIT was applied for empirical data positing doubly-censoring effects, further supporting the use of GBIT, particularly in situations involving doubly-censored data.
Latent class (LC) analysis is a model-based clustering approach for categorical data, with a wide range of applications in the social sciences and beyond. When the data have a hierarchical structure, the multilevel LC mo...Latent class (LC) analysis is a model-based clustering approach for categorical data, with a wide range of applications in the social sciences and beyond. When the data have a hierarchical structure, the multilevel LC model can be used to account for higher-level dependencies between the units by means of a further categorical LC variable at the group level. The research interest of LC analysis typically lies in the relationship between the LCs and external covariates, or predictors. To estimate LC models with covariates, researchers can use the one-step approach, or the generally recommended stepwise estimators, which separate the estimation of the clustering model from the subsequent estimation of the regression model. The package multilevLCA has the most comprehensive set of model specifications and estimation approaches for this family of models in the open-source domain, estimating single- and multilevel LC models, with and without covariates, using the one-step and stepwise approaches.
In the last half of the twentieth century, psychology and neuroscience have experienced a renewed interest in intraindividual variation. To date, there are few quantitative methods to evaluate whether a population (betwe...In the last half of the twentieth century, psychology and neuroscience have experienced a renewed interest in intraindividual variation. To date, there are few quantitative methods to evaluate whether a population (between-person) structure is likely to hold for individual people, often referred to as ergodicity. We introduce a new network information theoretic metric, the ergodicity information index (EII), that quantifies the amount of information lost by representing all individuals with a between-person structure. A Monte Carlo simulation demonstrated that EII can effectively delineate between ergodic and nonergodic systems. A bootstrap test is derived to statistically determine whether the empirical data is likely generated from an ergodic process. When a process is identified as nonergodic, then it's possible that a mixture of groups exist. To evaluate whether groups exist, we develop an information theoretic clustering method to detect groups. Finally, two empirical examples are presented using intensive longitudinal data from personality and neuroscience domains. Both datasets were found to be nonergodic, and meaningful groupings were identified in each dataset. Subsequent analysis showed that some of these groups are ergodic, meaning that the individuals can be represented with a single population structure without significant loss of information. Notably, in the neuroscience data, we could correctly identify two clusters of individuals (young vs. older adults) measured by a pattern separation task that were related to hippocampal connectivity to the default mode network.
Growth curve modeling has been widely used in many disciplines to understand the trajectories of growth. Two popular forms utilized in the real-world analyses are the linear and quadratic growth curve models. These model...Growth curve modeling has been widely used in many disciplines to understand the trajectories of growth. Two popular forms utilized in the real-world analyses are the linear and quadratic growth curve models. These models operate on the assumption that measurements are conducted exactly at pre-set time or intervals. In essence, the reliability of these models is deeply tied to the punctuality and consistency of the data collection process. However, in real-world data collection, this assumption is often violated. Deviations from the ideal measurement schedule often emerge, resulting in measurement error in time and consequent biased responses. Our simulation findings indicate that such error can skew estimations, especially in quadratic GCM. To account for the measurement error in time, we introduce a Bayesian growth curve model to accommodate the error in the individual time values. We demonstrate the performance of the proposed approach through simulation studies. Furthermore, to illustrate its application in practice, we provide a real-data example, underscoring the practical benefits of the proposed model.
When using ecological momentary assessment data (EMA), missing data is pervasive as participant attrition is a common issue. Thus, any EMA study must have a missing data plan. In this paper, we discuss missingness in tim...When using ecological momentary assessment data (EMA), missing data is pervasive as participant attrition is a common issue. Thus, any EMA study must have a missing data plan. In this paper, we discuss missingness in time series analysis and the appropriate way to handle missing data when the data is modeled as an idiographic discrete time continuous measure state-space model. We found that Missing Completely at Random, Missing At Random, and Time-dependent Missing At Random data have less bias and variability than Autoregressive Time-dependent Missing At Random and Missing Not At Random. The Kalman filter excelled at handling missing data under most conditions. Contrary to the literature, we found that using a variety of methods, multiple imputations struggled to recover the parameters.
There is a growing interest of researchers in meta-analytic methods for comparing variances as a means to answer questions on between-group differences in variability. When measurements are fallible, however, the varianc...There is a growing interest of researchers in meta-analytic methods for comparing variances as a means to answer questions on between-group differences in variability. When measurements are fallible, however, the variance of an outcome reflects both the variance of the true scores and the error variance. Consequently, effect sizes based on variances, such as the log variability ratio (lnVR) or the log coefficient of variation ratio (lnCVR), may thus not only reflect between-group differences in the true-score variances but also differences in measurement reliability. In this article, we derive formulas to correct the lnVR and lnCVR and their sampling variances for between-group differences in reliability and evaluate their performance in simulation studies. We find that when the goal is to meta-analyze differences between the true-score variances and reliability differs between groups, our proposed corrections lead to accurate estimates of effect sizes and sampling variances in single studies, accurate estimates of the average effect and the between-study variance in random-effects meta-analysis, and adequate type I error rates for the significance test of the average effect. We discuss how to deal with problems arising from missing or imprecise group-specific reliability estimates in meta-analytic data sets and identify questions for further methodological research.
Work surrounding the replicability and generalizability of network models has increased in recent years, prompting debate on whether network properties can be expected to be consistent across samples. To date, certain me...Work surrounding the replicability and generalizability of network models has increased in recent years, prompting debate on whether network properties can be expected to be consistent across samples. To date, certain methodological practices may have contributed to observed inconsistencies, including use of single-item indicators and non-identical measurement tools. The current study used a resampling approach to disentangle the effects of sampling variability from scale variability when assessing network replicability in empirical data. Additionally, we explored whether consistencies in network characteristics were improved when more items were aggregated to estimate node scores, which we hypothesized should yield more representative measures of latent constructs. Overall, using different scales produced more variability in network properties than using different samples, but these discrepancies were markedly reduced with larger samples and greater node aggregation. Findings underscored the impact of aggregating items when estimating nodes: Multi-item indicators led to denser networks, higher network sensitivity, greater estimates of global strength, and greater levels of consistency in network properties (e.g., edge weights, centrality scores). Taken together, variability in network properties across samples may arise from poor measurement conditions; additionally, variability may reflect properties of the true network model and/or the measurement instrument. All data and syntax are openly available online (https://osf.io/m37q2/).
LOO (Leave-One-Out cross-validation) and WAIC (Widely Applicable Information Criterion) are widely used for model selection in Bayesian statistics. Most studies select the model with the smallest value based on point est...LOO (Leave-One-Out cross-validation) and WAIC (Widely Applicable Information Criterion) are widely used for model selection in Bayesian statistics. Most studies select the model with the smallest value based on point estimates, often without considering the differences in fit indices or the uncertainty of the estimates. To address this gap, we propose a sequential method for comparing models based on confidence intervals for or A simulation study was conducted to evaluate this method in selecting mixed-effects location-scale models (MELSMs). Our study revealed that the sequential methods, especially when using a 90% confidence interval, achieved a higher accuracy rate of model selection compared to the point method when the true model was simple, had a large magnitude of random intercept in the scale model, or had a large sample size. Models selected by the sequential methods demonstrated higher power, narrower credible interval width, smaller standard errors for the fixed effect in the location model, and lower bias for the random effect of the intercept in the location model. Differences between LOO and WAIC were significant only when the level-1 sample size was small, with LOO performing better when the true model had homogeneous or severe heterogeneity in residual variances.
Automated detection of facial emotions has been an interesting topic for multiple decades in social and behavioral research but is only possible very recently. In this tutorial, we review three popular artificial intelli...Automated detection of facial emotions has been an interesting topic for multiple decades in social and behavioral research but is only possible very recently. In this tutorial, we review three popular artificial intelligence based emotion detection programs that are accessible to R programmers: Google Cloud Vision, Amazon Rekognition, and Py-Feat. We present their advantages, disadvantages, and provide sample code so that researchers can immediately begin designing, collecting, and analyzing emotion data. Furthermore, we provide an introductory level explanation of the machine learning, deep learning, and computer vision algorithms that underlie most emotion detection programs in order to improve literacy of explainable artificial intelligence in the social and behavioral science literature.
Neural networks like variational autoencoders have been proposed as a statistical tool to fit item factor models to data. Advantages are that high dimensional models can be estimated more efficiently as compared to conve...Neural networks like variational autoencoders have been proposed as a statistical tool to fit item factor models to data. Advantages are that high dimensional models can be estimated more efficiently as compared to conventional approaches. In this study, we demonstrate advantages of a specific autoencoder as a tool for amortized joint maximum likelihood estimation of item factor models. Contrary to contemporary joint maximum likelihood estimation and marginal maximum likelihood estimation, no additional parameter constraints are necessary to ensure standard asymptotic theory to apply. In a simulation study, the performance of the autoencoder is compared to constrained joint maximum likelihood and various forms of marginal maximum likelihood under different distributions for the factor scores. Results show that the amortized joint maximum likelihood estimates of the factors scores are overall less biased as compared to the other approaches. We illustrate the use of the autoencoder in two real data examples.
Diagnostic classification models (DCMs) are psychometric models designed to classify examinees according to their proficiency or non-proficiency of specified latent attributes. Longitudinal DCMs have recently been develo...Diagnostic classification models (DCMs) are psychometric models designed to classify examinees according to their proficiency or non-proficiency of specified latent attributes. Longitudinal DCMs have recently been developed as psychometric models for modeling changes in examinee proficiency statuses over time. Currently, software programs for estimating longitudinal DCMs are limited in functionality and generality, expensive, or cumbersome for applied researchers. This manuscript describes and demonstrates a newly developed R package for estimating a general longitudinal DCM, the transition diagnostic classification model.
We propose interrater reliability coefficients for observational interdependent social network data, which are dyadic data from a network of interacting subjects that are observed by external raters. Using the social rel...We propose interrater reliability coefficients for observational interdependent social network data, which are dyadic data from a network of interacting subjects that are observed by external raters. Using the social relations model, dyadic scores of subjects' behaviors during these interactions can be decomposed into actor, partner, and relationship effects. These effects constitute different facets of theoretical interest about which researchers formulate research questions. Based on generalizability theory, we extended the social relations model with rater effects, resulting in a model that decomposes the variance of dyadic observational data into effects of actors, partners, relationships, raters, and their statistical interactions. We used the variances of these effects to define intraclass correlation coefficients (ICCs) that indicate the extent the actor, partner, and relationship effects can be generalized across external raters. We proposed Markov chain Monte Carlo estimation of a Bayesian hierarchical linear model to estimate the ICCs, and tested their bias and coverage in a simulation study. The method is illustrated using data on social mimicry.
Psychometric networks can be estimated using nodewise regression to estimate edge weights when the joint distribution is analytically difficult to derive or the estimation is too computationally intensive. The nodewise a...Psychometric networks can be estimated using nodewise regression to estimate edge weights when the joint distribution is analytically difficult to derive or the estimation is too computationally intensive. The nodewise approach runs generalized linear models with each node as the outcome. Two regression coefficients are obtained for each link, which need to be aggregated to obtain the edge weight (i.e., the conditional association). The nodewise approach has been shown to reveal the true graph structure. However, for continuous variables, the regression coefficients are scaled differently than the partial correlations, and therefore the nodewise approach may lead to different edge weights. Here, the aggregation of the two regression coefficients is crucial in obtaining the true partial correlation. We show that when the correlations of the two predictors with the control variables are different, averaging the regression coefficients leads to an asymptotically biased estimator of the partial correlation. This is likely to occur when a variable has a high correlation with other nodes in the network (e.g., variables in the same domain) and a lower correlation with another node (e.g., variables in a different domain). We discuss two different ways of aggregating the regression weights, which can obtain the true partial correlation: first, multiplying the weights and taking their square root, and second, rescaling the regression weight by the residual variances. The two latter estimators can recover the true network structure and edge weights.
In this tutorial, we clarify the distinction between estimated factor scores, which are weighted composites of observed variables, and true factor scores, which are unobservable values of the underlying latent variable....In this tutorial, we clarify the distinction between estimated factor scores, which are weighted composites of observed variables, and true factor scores, which are unobservable values of the underlying latent variable. Using an analogy with linear regression, we show how predicted values in linear regression share the properties of the most common type of factor score estimates, regression factor scores, computed from single-indicator and multiple indicator latent variable models. Using simulated data from 1- and 2-factor models, we also show how the amount of measurement error affects the reliability of regression factor scores, and compare the performance of regression factor scores with that of unweighted sum scores.
Time series analysis is increasingly popular across scientific domains. A key concept in time series analysis is stationarity, the stability of statistical properties of a time series. Understanding stationarity is cruci...Time series analysis is increasingly popular across scientific domains. A key concept in time series analysis is stationarity, the stability of statistical properties of a time series. Understanding stationarity is crucial to addressing frequent issues in time series analysis such as the consequences of failing to model non-stationarity, how to determine the mechanisms generating non-stationarity, and consequently how to model those mechanisms (i.e., by differencing or detrending). However, many empirical researchers have a limited understanding of stationarity, which can lead to the use of incorrect research practices and misleading substantive conclusions. In this paper, we address this problem by answering these questions in an accessible way. To this end, we study how researchers can use detrending and differencing to model trends in time series analysis. We show simulation the consequences of modeling trends inappropriately, and evaluate the performance of one popular approach to distinguish different trend types in empirical data. We present these results in an accessible way, providing an extensive introduction to key concepts in time series analysis, illustrated throughout with simple examples. Finally, we discuss a number of take-home messages and extensions to standard approaches, which directly address more complex time-series analysis problems encountered by empirical researchers.
Interest in identifying latent growth profiles to support the psychological and social-emotional development of individuals has translated into the widespread use of growth mixture models (GMMs). In most cases, GMMs are...Interest in identifying latent growth profiles to support the psychological and social-emotional development of individuals has translated into the widespread use of growth mixture models (GMMs). In most cases, GMMs are based on scores from item responses collected using survey scales or other measures. Research already shows that GMMs can be sensitive to departures from ideal modeling conditions and that growth model results outside of GMMs are sensitive to decisions about how item responses are scored, but the impact of scoring decisions on GMMs has never been investigated. We start to close that gap in the literature with the current study. Through empirical and Monte Carlo studies, we show that GMM results-including convergence, class enumeration, and latent growth trajectories within class-are extremely sensitive to seemingly arcane measurement decisions. Further, our results make clear that, because GMM latent classes are not known a priori, measurement models used to produce scores for use in GMMs are, almost by definition, misspecified because they cannot account for group membership. Misspecification of the measurement model then, in turn, biases GMM results. Practical implications of these results are discussed. Our findings raise serious concerns that many results in the current GMM literature may be driven, in part or whole, by measurement artifacts rather than substantive differences in developmental trends.
Multivariate Behav Res
· 2025 · PMID 39801265
·
Full text
When studying effect heterogeneity between different subgroups (i.e., moderation), researchers are frequently interested in the mediation mechanisms underlying the heterogeneity, that is, the mediated moderation. For ass...When studying effect heterogeneity between different subgroups (i.e., moderation), researchers are frequently interested in the mediation mechanisms underlying the heterogeneity, that is, the mediated moderation. For assessing mediated moderation, conventional methods typically require parametric models to define mediated moderation, which has limitations when parametric models may be misspecified and when causal interpretation is of interest. For causal interpretations about mediation, causal mediation analysis is increasingly popular but is underdeveloped for mediated moderation analysis. In this study, we extend the causal mediation literature, and we propose a novel method for mediated moderation analysis. Using the potential outcomes framework, we obtain two causal estimands that decompose the total moderation: (i) the mediated moderation attributable to a mediator and (ii) the remaining moderation unattributable to the mediator. We also develop a multiply robust estimation method for the mediated moderation analysis, which can incorporate machine learning methods in the inference of the causal estimands. We evaluate the proposed method through simulations. We illustrate the proposed mediated moderation analysis by assessing the mediation mechanism that underlies the gender difference in the effect of a preventive intervention on adolescent behavioral outcomes.