The effective sample size in Bayesian information criterion for level-specific fixed and random-effect selection in a two-level nested model.

Popular statistical software provides the Bayesian information criterion (BIC) for multi-level models or linear mixed models. However, it has been observed that the combination of statistical literature and software documentation has led to discrepancies in the formulas of the BIC and uncertainties as to the proper use of the BIC in selecting a multi-level model with respect to level-specific fixed and random effects. These discrepancies and uncertainties result from different specifications of sample size in the BIC's penalty term for multi-level models. In this study, we derive the BIC's penalty term for level-specific fixed- and random-effect selection in a two-level nested design. In this new version of BIC, called BIC E 1 , this penalty term is decomposed into two parts if the random-effect variance-covariance matrix has full rank: (a) a term with the log of average sample size per cluster and (b) the total number of parameters times the log of the total number of clusters. Furthermore, we derive the new version of BIC, called BIC E 2 , in the presence of redundant random effects. We show that the derived formulae, BIC E 1 and BIC E 2 , adhere to empirical values via numerical demonstration and that BIC E ( E indicating either E 1 or E 2 ) is the best global selection criterion, as it performs at least as well as BIC with the total sample size and BIC with the number of clusters across various multi-level conditions through a simulation study. In addition, the use of BIC E 1 is illustrated with a textbook example dataset.

Using Auxiliary Item Information in the Item Parameter Estimation of a Graded Response Model for a Small to Medium Sample Size: Empirical Versus Hierarchical Bayes Estimation.

Marginal maximum likelihood estimation (MMLE) is commonly used for item response theory item parameter estimation. However, sufficiently large sample sizes are not always possible when studying rare populations. In this paper, empirical Bayes and hierarchical Bayes are presented as alternatives to MMLE in small sample sizes, using auxiliary item information to estimate the item parameters of a graded response model with higher accuracy. Empirical Bayes and hierarchical Bayes methods are compared with MMLE to determine under what conditions these Bayes methods can outperform MMLE, and to determine if hierarchical Bayes can act as an acceptable alternative to MMLE in conditions where MMLE is unable to converge. In addition, empirical Bayes and hierarchical Bayes methods are compared to show how hierarchical Bayes can result in estimates of posterior variance with greater accuracy than empirical Bayes by acknowledging the uncertainty of item parameter estimates. The proposed methods were evaluated via a simulation study. Simulation results showed that hierarchical Bayes methods can be acceptable alternatives to MMLE under various testing conditions, and we provide a guideline to indicate which methods would be recommended in different research situations. R functions are provided to implement these proposed methods.

Modeling variability in treatment effects for cluster randomized controlled trials using by-variable smooth functions in a generalized additive mixed model.

Variability in treatment effects is common in intervention studies using cluster randomized controlled trial (C-RCT) designs. Such variability is often examined in multilevel modeling (MLM) to understand how treatment effects (TRT) differ based on the level of a covariate (COV), called TRT [Formula: see text] COV. In detecting TRT [Formula: see text] COV effects using MLM, relationships between covariates and outcomes are assumed to vary across clusters linearly. However, this linearity assumption may not hold in all applications and an incorrect assumption may lead to biased statistical inference about TRT [Formula: see text] COV effects. In this study, we present generalized additive mixed model (GAMM) specifications in which cluster-specific functional relationships between covariates and outcomes can be modeled using by-variable smooth functions. In addition, the implementation for GAMM specifications is explained using the mgcv R package (Wood, 2021). The usefulness of the GAMM specifications is illustrated using intervention data from a C-RCT. Results of simulation studies showed that parameters and by-variable smooth functions were recovered well in various multilevel designs and the misspecification of the relationship between covariates and outcomes led to biased estimates of TRT [Formula: see text] COV effects. Furthermore, this study evaluated the extent to which the GAMM can be treated as an alternative model to MLM in the presence of a linear relationship.

Incorporating Functional Response Time Effects into a Signal Detection Theory Model.

Signal detection theory (SDT; Tanner & Swets in Psychological Review 61:401-409, 1954) is a dominant modeling framework used for evaluating the accuracy of diagnostic systems that seek to distinguish signal from noise in psychology. Although the use of response time data in psychometric models has increased in recent years, the incorporation of response time data into SDT models remains a relatively underexplored approach to distinguishing signal from noise. Functional response time effects are hypothesized in SDT models, based on findings from other related psychometric models with response time data. In this study, an SDT model is extended to incorporate functional response time effects using smooth functions and to include all sources of variability in SDT model parameters across trials, participants, and items in the experimental data. The extended SDT model with smooth functions is formulated as a generalized linear mixed-effects model and implemented in the gamm4 R package. The extended model is illustrated using recognition memory data to understand how conversational language is remembered. Accuracy of parameter estimates and the importance of modeling variability in detecting the experimental condition effects and functional response time effects are shown in conditions similar to the empirical data set via a simulation study. In addition, the type 1 error rate of the test for a smooth function of response time is evaluated.

Space-time modeling of intensive binary time series eye-tracking data using a generalized additive logistic regression model.

Eye-tracking has emerged as a popular method for empirical studies of cognitive processes across multiple substantive research areas. Eye-tracking systems are capable of automatically generating fixation-location data over time at high temporal resolution. Often, the researcher obtains a binary measure of whether or not, at each point in time, the participant is fixating on a critical interest area or object in the real world or in a computerized display. Eye-tracking data are characterized by spatial-temporal correlations and random variability, driven by multiple fine-grained observations taken over small time intervals (e.g., every 10 ms). Ignoring these data complexities leads to biased inferences for the covariates of interest such as experimental condition effects. This article presents a novel application of a generalized additive logistic regression model for intensive binary time series eye-tracking data from a between- and within-subjects experimental design. The model is formulated as a generalized additive mixed model (GAMM) and implemented in the mgcv R package. The generalized additive logistic regression model was illustrated using an empirical data set aimed at understanding the accommodation of regional accents in spoken language processing. Accuracy of parameter estimates and the importance of modeling the spatial-temporal correlations in detecting the experimental condition effects were shown in conditions similar to our empirical data set via a simulation study. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Modelling multilevel nonlinear treatment-by-covariate interactions in cluster randomized controlled trials using a generalized additive mixed model.

A cluster randomized controlled trial (C-RCT) is common in educational intervention studies. Multilevel modelling (MLM) is a dominant analytic method to evaluate treatment effects in a C-RCT. In most MLM applications intended to detect an interaction effect, a single interaction effect (called a conflated effect) is considered instead of level-specific interaction effects in a multilevel design (called unconflated multilevel interaction effects), and the linear interaction effect is modelled. In this paper we present a generalized additive mixed model (GAMM) that allows an unconflated multilevel interaction to be estimated without assuming a prespecified form of the interaction. R code is provided to estimate the model parameters using maximum likelihood estimation and to visualize the nonlinear treatment-by-covariate interaction. The usefulness of the model is illustrated using instructional intervention data from a C-RCT. Results of simulation studies showed that the GAMM outperformed an alternative approach to recover an unconflated logistic multilevel interaction. In addition, the parameter recovery of the GAMM was relatively satisfactory in multilevel designs found in educational intervention studies, except when the number of clusters, cluster sizes, and intraclass correlations were small. When modelling a linear multilevel treatment-by-covariate interaction in the presence of a nonlinear effect, biased estimates (such as overestimated standard errors and overestimated random effect variances) and incorrect predictions of the unconflated multilevel interaction were found.

Level-specific residuals and diagnostic measures, plots, and tests for random effects selection in multilevel and mixed models.

Multilevel data structures are often found in multiple substantive research areas, and multilevel models (MLMs) have been widely used to allow for such multilevel data structures. One important step when applying MLM is the selection of an optimal set of random effects to account for variability and heteroscedasticity in multilevel data. Literature reviews on current practices in applying MLM showed that diagnostic plots are only rarely used for model selection and for model checking. In this study, possible random effects and a generic description of the random effects were provided to guide researchers to select necessary random effects. In addition, based on extensive literature reviews, level-specific diagnostic plots were presented using various kinds of level-specific residuals, and diagnostic measures and statistical tests were suggested to select a set of random effects. Existing and newly proposed methods were illustrated using two data sets: a cross-sectional data set and a longitudinal data set. Along with the illustration, we discuss the methods and provide guidelines to select necessary random effects in model-building steps. R code was provided for the analyses.

Modeling Multivariate Count Time Series Data with a Vector Poisson Log-Normal Additive Model: Applications to Testing Treatment Effects in Single-Case Designs.

In education and psychology, single-case designs (SCDs) have been used to detect treatment effects using time series data in the presence or absence of intervention. One popular design variant of SCDs is a multiple-baseline design for multiple outcomes, which often collects outcomes with some form of a count. A Poisson model is a natural choice for the count outcome. However, the assumption of the Poisson model that the outcome variable's mean is equal to its variance is often violated in SCDs, as the variance is often larger than the mean (called overdispersion). In addition, when multiple outcomes are from the same participant, it is likely that they are correlated. In this paper, we present a vector Poisson log-normal additive (V-PLN-A) model to deal with (a) change processes (auto- and cross-correlations and data-driven trend) and (b) correlation and overdispersion in multivariate count time series. A multivariate normal distribution was adapted to account for correlation among multiple outcomes as well as possible overdispersion. The V-PLN-A model was applied to an educational intervention study to test treatment effects. Simulation study results showed that parameter recovery of the V-PLN-A model was satisfactory in a large number of timepoints using Bayesian analysis, and that ignoring change processes and overdispersion led to biased estimates of the treatment effects.

A Markov Mixed-Effect Multinomial Logistic Regression Model for Nominal Repeated Measures with an Application to Syntactic Self-Priming Effects.

Syntactic priming effects have been investigated for several decades in psycholinguistics and the cognitive sciences to understand the cognitive mechanisms that support language production and comprehension. The question of whether speakers prime themselves is central to adjudicating between two theories of syntactic priming, activation-based theories and expectation-based theories. However, there is a lack of a statistical model to investigate the two different theories when nominal repeated measures are obtained from multiple participants and items. This paper presents a Markov mixed-effect multinomial logistic regression model in which there are fixed and random effects for own-category lags and cross-category lags in a multivariate structure and there are category-specific crossed random effects (random person and item effects). The model is illustrated with experimental data that investigates the average and participant-specific deviations in syntactic self-priming effects. Results of the model suggest that evidence of self-priming is consistent with the predictions of activation-based theories. Accuracy of parameter estimates and precision is evaluated via a simulation study using Bayesian analysis.

Multilevel Reliability Measures of Latent Scores Within an Item Response Theory Framework.

This paper evaluated multilevel reliability measures in two-level nested designs (e.g., students nested within teachers) within an item response theory framework. A simulation study was implemented to investigate the behavior of the multilevel reliability measures and the uncertainty associated with the measures in various multilevel designs regarding the number of clusters, cluster sizes, and intraclass correlations (ICCs), and in different test lengths, for two parameterizations of multilevel item response models with separate item discriminations or the same item discrimination over levels. Marginal maximum likelihood estimation (MMLE)-multiple imputation and Bayesian analysis were employed to evaluate the accuracy of the multilevel reliability measures and the empirical coverage rates of Monte Carlo (MC) confidence or credible intervals. Considering the accuracy of the multilevel reliability measures and the empirical coverage rate of the intervals, the results lead us to generally recommend MMLE-multiple imputation. In the model with separate item discriminations over levels, marginally acceptable accuracy of the multilevel reliability measures and empirical coverage rate of the MC confidence intervals were found in a limited condition, 200 clusters, 30 cluster size, .2 ICC, and 40 items, in MMLE-multiple imputation. In the model with the same item discrimination over levels, the accuracy of the multilevel reliability measures and the empirical coverage rate of the MC confidence intervals were acceptable in all multilevel designs we considered with 40 items under MMLE-multiple imputation. We discuss these findings and provide guidelines for reporting multilevel reliability measures.