This study investigates the dynamics of semantic associations by exploring the interplay between continuity and direction in a geometric semantic space. While acknowledging the role of continuity in guiding associations, our work introduces Direction as a crucial factor influencing transitions. Conceptually, we define the stream of associations as movement along a sequence of objects, with attention amplifying dissimilarity and progressing in the direction of maximal resolution, conceptualized as the most "stretched" direction. The core of our methodological innovation lies in the introduction of a unique adaptation of discrete Ricci curvature to measure the direction of maximal resolution, tailored specifically to a hypergraph framework. By reinterpreting traditional curvature concepts within this context, we provide a novel quantitative approach to understanding semantic transitions. Empirically, our investigation involves a categorical fluency task where participants name animals, allowing us to construct a hypergraph for transition analysis. We evaluate two hypotheses: the relationship between edge "stretchiness" and transition probability, and the enhanced explanatory power of considering Similarity + Direction over similarity alone. Our model challenges the standard view by proposing that the stream of thought moves in the direction of maximal resolution. By introducing the concept of Ricci curvature in a hypernetwork, we offer a novel tool for quantifying resolution and demonstrate its practical application in the context of semantic space. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Psychological theories are often expressed verbally using natural language, which may lead to varying interpretations of the phenomenon under study. This potential confusion can be mitigated by formalizing verbal theories using mathematical language, which can help in defining, analyzing, and interpreting one's hypotheses in quantitative terms. Differential equations (DEs) are a class of models in the dynamical systems framework, particularly suited to many dynamic theories in psychology. However, there is a lack of tools for translating verbal theories into DE systems. To facilitate this translation, we introduce SimDE (https://simde.ucdavis.edu/), an open-access R Shiny application that allows users to specify a DE model and then simulate the trajectories of each variable over time. SimDE provides an interface to simulate a range of DE models, with features such as: (a) first- or second-order DEs (e.g., exponential, oscillatory), (b) models with or without a dynamic error term (ordinary or stochastic DEs), (c) models with coupling dynamics. Users have the flexibility of plotting these systems in order to see the pattern of changes over time and determine the appropriateness of the model for the phenomenon they are trying to study. The goal of our app is to serve as a tool for researchers who want to explore DE models for their psychological theories before they even collect data. It can also help researchers to study the implicit assumptions of their systems defined with such DEs and further refine them as needed. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Thanks to the popularity of smartphones with high-quality cameras and social media platforms, an exceptional amount of image data is generated and shared daily. This visual data can provide unprecedented insights into daily life and can be used to help answer research questions in psychology. However, the traditional methods used to analyze visual data are burdensome and are either time-intensive (e.g., content analysis) or require technical training (e.g., developing and training deep learning models). Zero-shot learning, where a pretrained model is used without any additional training, requires less technical expertise and may be a particularly attractive method for psychology researchers aiming to analyze image data. In this tutorial, we aim to provide an overview and step-by-step guide on how to analyze visual data with zero-shot learning. Specifically, we demonstrate how to use two popular models (Contrastive Language-Image Pretraining and Large Language and Vision Assistant) to identify a beverage in an image from a data set where we manipulated the type of beverage present, the setting, and the prominence of the beverage in the image (foreground, midground, background). To guide researchers through this process, we provide open code and data on GitHub and as a Google Colab notebook. Finally, we discuss how to interpret and report accuracy, how to create a validation data set, what steps need to be taken to implement the models with new data, and discuss future challenges and limitations of the method. To conclude, zero-shot learning requires less technical expertise and may be a particularly attractive method for psychology researchers aiming to analyze image data. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Latent growth curve (LGC) models, implemented through structural equation modeling, are widely used to analyze developmental and learning trajectories. Model selection in LGC often relies on goodness-of-fit indices (e.g., χ², Akaike information criterion, and root-mean-square error of approximation), but these metrics fail to assess the temporal constancy, or stability of parameters, an important aspect when forecasting longitudinal data. Addressing this gap, we propose a novel parameter constancy test (PCT) tailored for LGC models. This test evaluates internal constancy, identifies potential breakpoints, helps determine the minimal number of measurement waves needed for reliable modeling, and is also useful for comparing different explanatory models of the analyzed data. To validate this approach, we applied PCT to real-world data, comparing the widely used quadratic function model with the negative exponential model and other nonlinear functions. The results reveal that the negative exponential model, unlike the quadratic function, consistently exhibits parameter constancy even with fewer sampling waves, making it particularly suitable for longitudinal analysis. Additionally, PCT highlights how inappropriate model selection or instability may lead to misinterpretations, particularly in evaluating interventions or extrapolating beyond observed time frames. Our findings emphasize the dual importance of statistical fit and parameter constancy in selecting LGC models. By integrating PCT into standard practice, researchers can better ensure model consistency, optimize resource allocation, and avoid erroneous conclusions in developmental and learning studies. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Factor analysis and item response theory models are conceptually equivalent and feature interchangeable parameters; however, they differ in their estimation techniques. Item response theory typically employs full-information techniques, such as marginal maximum likelihood (MML), whereas ordinal factor analysis relies on limited-information techniques, like weighted least squares mean and variance adjusted and unweighted least squares mean and variance adjusted. Previous studies comparing these techniques have produced conflicting results without clear explanations. Moreover, there is limited guidance on the optimal use of limited-information techniques and the effects of nonnormal distributions, leaving critical gaps in understanding. This study addresses these gaps through a comprehensive Monte Carlo simulation that incorporates diverse nonnormal distributions and reevaluates approaches to handling nonconvergent solutions. The results show that the case-wise exclusion approach, commonly used in prior research, unfairly penalizes high-convergence techniques, such as MML. In contrast, the dataset-wise exclusion approach, which removes all datasets with nonconvergent solutions, enables fairer comparisons and highlights MML's superior performance under most conditions. Additionally, while skewed leptokurtic distributions confirm the expected effects of normality violations, other nonnormal distributions yield unexpected results, cautioning against generalizing findings from one type of nonnormality to all. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

This article introduces autoregressive (AR) linear models to psychology students and researchers through a step-by-step approach using SPSS and R. Despite their relevance, AR models remain underutilized in behavioral sciences, possibly due to conceptual challenges and difficulties interpreting autocorrelation and seasonality. Our aim is to simplify their implementation by presenting time series models as special cases of linear regression, using accessible language and practical examples. The article illustrates AR estimation using real data, incorporating lagged values as predictors of the dependent variable. Residual diagnostics, a frequently overlooked aspect in applied research, receive special attention, including figures and statistical tests. As Kmenta (1971) demonstrated, serially correlated residuals can lead to artificially low p values for the parameter estimates, potentially resulting in explanatory variables being deemed significant when they truly are not. To promote understanding, we offer intuitive visualizations and clear decision rules for model building, lag selection, and seasonality detection. We compare polynomial and AR models using the confounding test. The data set and annotated R and SPSS scripts are included to support replication and help readers learn basic syntax. We also discuss conceptual and practical limitations of moving average, integration (I), and exponential smoothing models, emphasizing the practical advantages of AR-only models in psychological contexts. Throughout, we stress the importance of aligning statistical models with theoretical assumptions and the temporal structure of data. By combining step-by-step explanations, visual guidance, and real-data applications, this tutorial provides a practical foundation for incorporating AR models into applied psychological research. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

The development and evaluation of text classifiers in psychology depends on rigorous manual coding. Yet, the evaluation of manual coding and computational algorithms is usually considered separately. This is problematic because developing high-validity classifiers is a repeated process of identifying, explaining, and addressing conceptual and measurement issues during both the manual coding and classifier development stages. To address this problem, we introduce the Repeated Adjustment of Measurement Protocols (RAMP) method for developing high-validity text classifiers in psychology. The RAMP method has three stages: manual coding, classifier development, and integrative evaluation. These stages integrate the best practices of content analysis (manual coding), data science (classifier development), and psychology (integrative evaluation). Central to this integration is the concept of an inference loop, defined as the process of maximizing validity through repeated adjustments to concepts and constructs, guided by push-back from the empirical data. Inference loops operate both within each stage of the method and across related studies. We illustrate RAMP through a case study, where we manually coded 21,815 sentences for misunderstanding (Krippendorff's α = .79), and developed a rule-based classifier (Matthews correlation coefficient [MCC] = 0.22), a supervised machine learning classifier (Bidirectional Encoder Representations From Transformers; MCC = 0.69) and a large language model classifier (GPT-4o; MCC = 0.47). By integrating manual coding and classifier development stages, we were able to identify and address a concept validity problem with misunderstandings. RAMP advances existing methods by operationalizing validity as an ongoing dynamic process, where concepts and constructs are repeatedly adjusted toward increasingly widespread intersubjective agreement on their utility. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Machine learning (ML) has become a common approach for estimating propensity scores (PSs) for quasi-experimental research using matching, weighting, or stratification on the PS. This systematic review examined 179 applications of ML for PS estimation across different fields, such as health, education, social sciences, and business over 40 years. The results show that the gradient boosting machine (GBM) is the most frequently used method, followed by random forest. Classification and regression trees, neural networks, and the super learner were also used in more than 5% of studies. The most frequently used packages to estimate PSs were twang, gbm, and randomforest in the R statistical software. The review identified that critical steps of the propensity score analysis are frequently underreported. Covariate balance evaluation was not reported by 48.04% of articles. Also, improper use of values for covariate balance evaluation was identified in 13.97% of the studies. Only 22.8% of studies performed a sensitivity analysis. Many hyperparameter configurations were used for ML methods, but only 46.9% of studies reported the hyperparameters used. A set of guidelines for reporting the use of ML for PS estimation is provided. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Worldwide, trends of globalization, demographic changes, and rising social mobility are modifying the composition of the pools of applicants competing for jobs, educational places, and other opportunities for which selection tools are used. This study investigates the impact of changes in applicant-pool composition (operationalized as the proportion of a focal group within the applicant pool, ) on the predictive validity of a selection tool (). We demonstrate that there is a direct relationship between and that is shaped by several parameters identified by a proposed equation. To investigate this relationship in the field, we conducted an empirical study based on more than 130,000 observations gathered from a real-life, high-stakes selection context. The results indicate that, under certain conditions, a given change in can have a beneficial effect on , whereas, under different conditions, it has a detrimental effect. Furthermore, the results suggest that the proposed equation can be used to explain the performance of a selection tool, under real conditions in the field, where the simplifying assumption underlying the proposed equation does not always hold. The present findings will enable decision-makers to explore the possible impact of anticipated changes in applicant-pool composition on the predictive validity of their selection tools, helping them make better informed decisions regarding the best one to use. Additional practical implications involve explaining differences between the predictive validity of different selection tools, or variations in the predictive validity of one particular tool depending on country, year, academic discipline, or occupation, for instance. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Mediation analysis is widely used in psychology to assess how an independent variable transmits its causal effect on an outcome both directly and indirectly through intermediary variables known as mediators. Causal mediation analysis addresses numerous criticisms of product-of-coefficients approach, often regarded as the primary method for estimating indirect effects in psychological research. However, navigating causal mediation analysis, especially in settings with multiple mediators, can be challenging for those unfamiliar with its concepts, assumptions, and estimation strategies. In this tutorial, we therefore offer a comprehensive guide to conducting causal mediation analysis with two mediators across three data-generating mechanisms: setups with causally dependent mediators, independent mediators, and noncausally dependent mediators. For each of these mechanisms, we provide formal mathematical definitions and assumptions for the natural direct and indirect effects, along with less technical explanations of these concepts. We also provide R and Stata codes for estimating the natural direct effect, the joint natural indirect effect, and the path-specific natural indirect effects using four different estimators: the imputation approach, the extended imputation approach, the inverse probability weighted approach, and the extended quasi-Bayesian Monte Carlo approach. Additionally, we illustrate each of these methods with examples from the International Dating Violence Study. This tutorial aims to equip applied researchers in psychology with all the necessary tools to conduct causal mediation analysis involving two mediators across various multiple mediators setups. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

The interpretation of correlation coefficients has invoked considerable discussion over many decades. One interpretive procedure is to use the coefficient of determination-the squared correlation coefficient-to index variance accounted for in one variable by variance in the other variable. A second interpretive procedure is to construct binomial effect size displays that involve dichotomizing continuous dependent variables. The present goal is to present a third interpretive procedure, with tutorial, to estimate probabilistic (dis)advantages implied by correlation coefficients and construct gain-probability diagrams. The proposed procedure does not involve dichotomizing continuous dependent variables, thereby losing information. In addition, the proposed procedure extends well to comparing correlation coefficients and facilitates subtle and nuanced implications that can enhance theoretical specificity. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Educational disparities are rooted in, and perpetuate, social inequalities across multiple dimensions such as race, socioeconomic status, and geography. To reduce disparities, most intervention strategies focus on a single domain and frequently evaluate their effectiveness by using causal decomposition analysis. However, a growing body of research suggests that single-domain interventions may be insufficient for individuals marginalized on multiple fronts. While interventions across multiple domains are increasingly proposed, there is limited guidance on appropriate methods for evaluating their effectiveness. To address this gap, we develop an extended causal decomposition analysis that simultaneously targets multiple causally ordered intervening factors, allowing for the assessment of their synergistic effects. These scenarios often involve challenges related to model misspecification because of complex interactions among group categories, intervening factors, and their confounders with the outcome. To mitigate these challenges, we introduce a triply robust estimator that leverages machine-learning techniques to address potential model misspecification. We apply our method to a cohort of students from the High School Longitudinal Study (HSLS:09), focusing on math achievement disparities between Black, Hispanic, and White high schoolers. Specifically, we examine how two sequential interventions-equalizing the proportion of students who attend high-performing schools and equalizing enrollment in Algebra I by ninth grade across racial groups-may reduce these disparities. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Missing not at random (MNAR) and nonnormal data are challenging to handle. Traditional missing data analytical techniques such as full information maximum likelihood (FIML) estimation may fail with nonnormal data as they are built on normal distribution assumptions. Two-stage robust estimation (TSRE) does manage nonnormal data, but both FIML and TSRE are less explored in longitudinal studies under MNAR conditions with nonnormal distributions. Unlike traditional statistical approaches, machine learning approaches do not require distributional assumptions about the data. More importantly, they have shown promise for MNAR data; however, their application in longitudinal studies, addressing both missing at random (MAR) and MNAR scenarios, is also underexplored. This study utilizes Monte Carlo simulations to assess and compare the effectiveness of six analytical techniques for missing data within the growth curve modeling framework. These techniques include traditional approaches like FIML and TSRE, machine learning approaches by single imputation (K-nearest neighbors and missForest), and machine learning approaches by multiple imputation (micecart and miceForest). We investigate the influence of sample size, missing data rate, missing data mechanism, and data distribution on the accuracy and efficiency of model estimation. Our findings indicate that FIML is most effective for MNAR data among the tested approaches. TSRE excels in handling MAR data, while missForest is only advantageous in limited conditions with a combination of very skewed distributions, very large sample sizes (e.g., ≥ 1,000), and low missing data rates. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Interaction effects are common in the behavioral sciences, especially in psychology, as they help explore how various factors influence human behavior. This article introduces a factored regression framework designed to estimate latent variable interactions and nonlinear effects, providing a flexible approach for modeling complex data structures, accommodating diverse data types, and handling missing data on any variable. The factored regression framework also allows graphical diagnostics to probe interactions effectively. Monte Carlo simulations were conducted to compare the performance of factored regression with existing maximum likelihood methods, such as latent moderated structural equations and product indicators. Results indicate that factored regression performs comparably to, if not better than, these traditional methods. The factored regression framework is implemented in Blimp software, offering an accessible and user-friendly syntax for specifying the models. Through practical examples and syntax excerpts, this article demonstrates the application of factored regression for estimating latent interactions, making it more approachable for a wide audience in behavioral research. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Mediation modeling using longitudinal data is an exciting field that captures the interrelations in dynamic changes, such as mediated changes, over time. Even though discrete-time vector autoregressive approaches are commonly used to estimate indirect effects in longitudinal data, they have known limitations due to the dependency of inferential results on the time intervals between successive occasions and the assumption of regular spacing between measurements. Continuous-time vector autoregressive models have been proposed as an alternative to address these issues. Previous work in the area (e.g., Deboeck & Preacher, 2015; Ryan & Hamaker, 2021) has shown how the direct, indirect, and total effects, for a range of time-interval values, can be calculated using parameters estimated from continuous-time vector autoregressive models for causal inferential purposes. However, both standardized effects size measures and methods for calculating the uncertainty around the direct, indirect, and total effects in continuous-time mediation have yet to be explored. Drawing from the mediation model literature, we present and compare results using the delta, Monte Carlo, and parametric bootstrap methods to calculate SEs and confidence intervals for the direct, indirect, and total effects in continuous-time mediation for inferential purposes. Options to automate these inferential procedures and facilitate interpretations are available in the cTMed R package. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Reliability is a key concept in psychology that has been broadly studied since the introduction of Cronbach's α, which is a measure of internal consistency. Despite its importance, reliability has been relatively understudied when dealing with intensive longitudinal data. Although intensive longitudinal measurements are often considered more ecologically valid and less prone to recall bias than survey data collected using traditional methods, there is no warranty that they are more reliable. Hence, empirical researchers need tools to study and report the reliability of the scales used in intensive longitudinal research. In recent years, psychologists have proposed different approaches to estimate the reliability of scales and items used when studying psychological dynamics. However, it is unclear how these approaches compare to one another, making it difficult to determine what options researchers have given a particular data set and specific research questions. Specifically, these approaches estimate reliability indices based on different statistical models, such as linear multilevel analysis, vector autoregressive models, and dynamic factor models. Furthermore, while some methods involve estimating one reliability index for the scores that applies to the whole sample, others estimate person-specific reliability indices. This wide variety of approaches can provoke some confusion. In this article, we aim to bridge this gap by reviewing and highlighting the similarities and differences of different methods used to estimate the reliability of intensive longitudinal data. We also showcase their application with empirical data. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Meta-analytic structural equation modeling (MASEM) allows a researcher to simultaneously examine multiple relations among variables by fitting a structural equation model to summary statistics from multiple studies. Consider, for example, a mediation model with a predictor (), mediator (), and outcome variable (). In such a model, can be a dichotomous variable, allowing researchers to examine the direct and indirect effects of an intervention as in randomized controlled trials (RCTs). However, the natural choice of a meta-analysis of RCTs would involve standardized mean differences as effect sizes, whereas MASEM requires correlation matrices as input. This can be solved by converting standardized mean differences (Cohen's or Hedges' ) to point-biserial correlations (). Possible conversion formulas vary across publications and conversion tools, and it is unclear which one is most appropriate for use in MASEM. The aim of this article is to describe and evaluate several conversions of standardized mean differences to point-biserial correlations in the context of RCTs. We investigate the impact of the usage of various conversions on MASEM parameter estimation using the R package metaSEM in a simulation study, varying the ratio of group sample sizes, number of primary studies, sample sizes, and missingness. The results show that a relatively unknown -to- conversion generally performs best. However, this conversion formula is not implemented in the mainstream conversion tools. We developed a user-friendly web application entitled Effect Size Calculator and Converter (https://hdejonge.shinyapps.io/ESCACO) that converts the user's primary study statistics into an effect size suitable for use in MASEM. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Models where some part of a mediation is moderated (conditional process models) are commonly used in psychology research, allowing for better understanding of when the process by which a focal predictor affects an outcome through a mediator depends on moderating variables. Methodological developments in conditional process analysis have focused on between-subject designs. However, two-instance repeated-measures designs, where each subject is measured twice: once in each of two instances, are also very common. Research on how to statistically test mediation, moderation, and conditional process models in these designs has lagged behind. Judd et al. (2001) introduced a piecewise method for testing for mediation, that Montoya and Hayes (2017) then translated to a path-analytic approach, quantifying the indirect effect. Moderation analysis in these designs has been described by Judd et al. (2001, 1996), and Montoya (2018). The generalization to conditional process analysis remains incomplete. I propose a general conditional process model for two-instance repeated-measures designs with one moderator and one mediator. Simplifications of this general model correspond to more commonly used moderated mediation models, such as first-stage and second-stage conditional process analysis. An applied example shows both how to conduct the analysis using MEMORE, a free and easy-to-use macro for SPSS and SAS, and how to interpret the results of such an analysis. Alternative methods for evaluating moderated mediation in two-instance repeated-measures designs using multilevel approaches are also discussed. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Using instruments comprising ordered responses to items is ubiquitous for studying many constructs of interest. However, using such an item response format may lead to items with response categories infrequently endorsed or unendorsed completely. In maximum likelihood estimation, this results in nonexisting estimates for thresholds. This work focuses on a Bayesian estimation approach to counter this issue. The issue changes from the existence of an estimate to how to effectively construct threshold priors. The proposed prior specification reconceptualizes the threshold prior as prior to the probability of each response category, which is an easier metric to manipulate while maintaining the necessary ordering constraints on the thresholds. The resulting induced-prior is more communicable, and we demonstrate comparable statistical efficiency with existing threshold priors. Evidence is provided using a simulated data set, a Monte Carlo simulation study, and an example multigroup item-factor model analysis. All analyses demonstrate how at least a relatively informative threshold prior is necessary to avoid inefficient posterior sampling and increase confidence in the coverage rates of posterior credible intervals. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Intervention studies in psychology often have a partially nested design (PND): after individuals are assigned to study arms, individuals in a treatment arm are subsequently assigned to clusters (e.g., therapists/therapy groups) to receive treatment, whereas individuals in a control arm are unclustered. Given the presence of clustering in the treatment arm, it can be of interest to examine the heterogeneity of treatment effects across the clusters; but this is challenging in PNDs. First, in defining a causal effect of treatment for a specific cluster, it is unclear how the treatment and control outcomes should be compared, as the clustering is absent in the control arm. Although it may be tempting to compare outcomes between a specific cluster and the entire control arm, this crude comparison may not represent a causal effect even in PNDs with randomized treatment assignments, as the cluster assignment may be nonrandomized (elaborated in this study). In this study, we develop methods to define, identify, and estimate the causal effects of treatment across specific clusters in a PND where the treatment and/or cluster assignment may be nonrandomized. Using the principal stratification approach and potential outcomes framework, we define causal estimands for the cluster-specific treatment effects in two scenarios: (a) no-interference and (b) within-cluster interference. We identify the effects under the principal ignorability assumption. For estimation, we provide a multiply-robust method that can protect against misspecification in a nuisance model and can incorporate machine learning methods in the nuisance model estimation. We evaluate the estimators' performance through simulations and illustrate the application using an empirical PND example. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Recent reviews report that about 80% of empirical factor analyses are applied to Likert-type responses and that it is exceedingly common to treat Likert-type item responses as continuous. However, traditional model fit index cutoffs like the root-mean-square error of approximation ≤ .06 or comparative fit index ≥ .95 were derived to have 90+% sensitivity to misspecification with continuous responses. A disconnect therefore emerges whereby traditional methodological guidelines assume continuous responses whereas empirical data often contain Likert-type responses. We provide an illustrative simulation study to show that this disconnect is not innocuous-the sensitivity of traditional cutoffs to misspecification is close to 100% with continuous responses but can fall considerably if 5-point Likert responses are treated as continuous in some conditions. In other conditions, the reverse may occur, and traditional cutoffs may be too strict. Generally, applying traditional cutoffs to Likert-type responses can adversely impact conclusions about fit adequacy. This article aims to address this prevalent issue by extending the dynamic fit index (DFI) framework to accommodate Likert-type responses. DFI is a simulation-based method that was initially intended to address changes in cutoff sensitivity to misspecification because of model characteristics (e.g., number of items, strength of loadings). Here, we propose extending DFI so that it also accounts for data characteristics (e.g., number of Likert scale points, response distribution). Two simulations are included to demonstrate that-with 5-point Likert-type responses-the proposed method (a) improves upon traditional cutoffs, (b) improves upon DFI cutoffs based on multivariate normality, and (c) consistently maintains 90+% sensitivity to misspecification. (PsycInfo Database Record (c) 2025 APA, all rights reserved).