Psychological methods最新文献

In current practice, intervention scientists applying the multiphase optimization strategy (MOST) with a 2^k factorial optimization trial use a component screening approach (CSA) to select intervention components for inclusion in an optimized intervention. In this approach, scientists review all estimated main effects and interactions to identify the important ones based on a fixed threshold, and then base decisions about component selection on these important effects. We propose an alternative posterior expected value approach based on Bayesian decision theory. This new approach aims to be easier to apply and more readily extensible to a variety of intervention optimization problems. We used Monte Carlo simulation to evaluate the performance of a posterior expected value approach and CSA (automated for simulation purposes) relative to two benchmarks: random component selection, and the classical treatment package approach. We found that both the posterior expected value approach and CSA yielded substantial performance gains relative to the benchmarks. We also found that the posterior expected value approach outperformed CSA modestly but consistently in terms of overall accuracy, sensitivity, and specificity, across a wide range of realistic variations in simulated factorial optimization trials. We discuss implications for intervention optimization and promising future directions in the use of posterior expected value to make decisions in MOST. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Integrating regularization methods into structural equation modeling is gaining increasing popularity. The purpose of regularization is to improve variable selection, model estimation, and prediction accuracy. In this study, we aim to: (a) compare Bayesian regularization methods for exploring covariate effects in multiple-indicators multiple-causes models, (b) examine the sensitivity of results to hyperparameter settings of penalty priors, and (c) investigate prediction accuracy through cross-validation. The Bayesian regularization methods examined included: ridge, lasso, adaptive lasso, spike-and-slab prior (SSP) and its variants, and horseshoe and its variants. Sparse solutions were developed for the structural coefficient matrix that contained only a small portion of nonzero path coefficients characterizing the effects of selected covariates on the latent variable. Results from the simulation study showed that compared to diffuse priors, penalty priors were advantageous in handling small sample sizes and collinearity among covariates. Priors with only the global penalty (ridge and lasso) yielded higher model convergence rates and power, whereas priors with both the global and local penalties (horseshoe and SSP) provided more accurate parameter estimates for medium and large covariate effects. The horseshoe and SSP improved accuracy in predicting factor scores, while achieving more parsimonious models. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Psychologists tend to rely on verbal descriptions of the environment over time, using terms like "unpredictable," "variable," and "unstable." These terms are often open to different interpretations. This ambiguity blurs the match between constructs and measures, which creates confusion and inconsistency across studies. To better characterize the environment, the field needs a shared framework that organizes descriptions of the environment over time in clear terms: as statistical definitions. Here, we first present such a framework, drawing on theory developed in other disciplines, such as biology, anthropology, ecology, and economics. Then we apply our framework by quantifying "unpredictability" in a publicly available, longitudinal data set of crime rates in New York City (NYC) across 15 years. This case study shows that the correlations between different "unpredictability statistics" across regions are only moderate. This means that regions within NYC rank differently on unpredictability depending on which definition is used and at which spatial scale the statistics are computed. Additionally, we explore associations between unpredictability statistics and measures of unemployment, poverty, and educational attainment derived from publicly available NYC survey data. In our case study, these measures are associated with mean levels in crime rates but hardly with unpredictability in crime rates. Our case study illustrates the merits of using a formal framework for disentangling different properties of the environment. To facilitate the use of our framework, we provide a friendly, step-by-step guide for identifying the structure of the environment in repeated measures data sets. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

The coefficient of determination, R², also called the explained variance, is often taken as a proportional measure of the relative determination of model on outcome. However, while R² has some attractive statistical properties, its reliance on squared variations (variances) may limit its use as an easily interpretable descriptive statistic of that determination. Here, the properties of this coefficient on the squared scale are discussed and generalized to three relative measures on the original scale. These generalizations can all be expressed as transformations of R², and alternatives can therefore also be calculated by plugging in related estimates, such as the adjusted R². The third coefficient, new for this article, and here termed the CoD_SD (the coefficient of determination in terms of standard deviations), or R_π (R-pi), equals R²/(R²+1-R²). It is argued that this coefficient most usefully captures the relative determination of the model. When the contribution of the error is c times that of the model, the CoD_SD equals 1/(1 + c), while R² equals 1/(1 + c²). (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Joint modeling of decisions and neural activation poses the potential to provide significant advances in linking brain and behavior. However, methods of joint modeling have been limited by difficulties in estimation, often due to high dimensionality and simultaneous estimation challenges. In the current article, we propose a method of model estimation that draws on state-of-the-art Bayesian hierarchical modeling techniques and uses factor analysis as a means of dimensionality reduction and inference at the group level. This hierarchical factor approach can adopt any model for the individual and distill the relationships of its parameters across individuals through a factor structure. We demonstrate the significant dimensionality reduction gained by factor analysis and good parameter recovery, and illustrate a variety of factor loading constraints that can be used for different purposes and research questions, as well as three applications of the method to previously analyzed data. We conclude that this method provides a flexible and usable approach with interpretable outcomes that are primarily data-driven, in contrast to the largely hypothesis-driven methods often used in joint modeling. Although we focus on joint modeling methods, this model-based estimation approach could be used for any high dimensional modeling problem. We provide open-source code and accompanying tutorial documentation to make the method accessible to any researchers. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Propensity score analysis (PSA) is a prominent method to alleviate selection bias in observational studies, but missing data in covariates is prevalent and must be dealt with during propensity score estimation. Through Monte Carlo simulations, this study evaluates the use of imputation methods based on multiple random forests algorithms to handle missing data in covariates: multivariate imputation by chained equations-random forest (Caliber), proximity imputation (PI), and missForest. The results indicated that PI and missForest outperformed other methods with respect to bias of average treatment effect regardless of sample size and missing mechanisms. A demonstration of these five methods with PSA to evaluate the effect of participation in center-based care on children's reading ability is provided using data from the Early Childhood Longitudinal Study, Kindergarten Class of 2010-2011. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

We demonstrate that all conventional meta-analyses of correlation coefficients are biased, explain why, and offer solutions. Because the standard errors of the correlation coefficients depend on the size of the coefficient, inverse-variance weighted averages will be biased even under ideal meta-analytical conditions (i.e., absence of publication bias, p-hacking, or other biases). Transformation to Fisher's z often greatly reduces these biases but still does not mitigate them entirely. Although all are small-sample biases (n < 200), they will often have practical consequences in psychology where the typical sample size of correlational studies is 86. We offer two solutions: the well-known Fisher's z-transformation and new small-sample adjustment of Fisher's that renders any remaining bias scientifically trivial. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

A currently overlooked application of the latent curve model (LCM) is its use in assessing the consequences of development patterns of change-that is as a predictor of distal outcomes. However, there are additional complications for appropriately specifying and interpreting the distal outcome LCM. Here, we develop a general framework for understanding the sensitivity of the distal outcome LCM to the choice of time coding, focusing on the regressions of the distal outcome on the latent growth factors. Using artificial and real-data examples, we highlight the unexpected changes in the regression of the slope factor which stand in contrast to prior work on time coding effects, and develop a framework for estimating the distal outcome LCM at a point in the trajectory-known as the aperture-which maximizes the interpretability of the effects. We also outline a prioritization approach developed for assessing incremental validity to obtain consistently interpretable estimates of the effect of the slope. Throughout, we emphasize practical steps for understanding these changing predictive effects, including graphical approaches for assessing regions of significance similar to those used to probe interaction effects. We conclude by providing recommendations for applied research using these models and outline an agenda for future work in this area. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Following an extensive simulation study comparing the operating characteristics of three different procedures used for establishing equivalence (the frequentist "TOST," the Bayesian "HDI-ROPE," and the Bayes factor interval null procedure), Linde et al. (2021) conclude with the recommendation that "researchers rely more on the Bayes factor interval null approach for quantifying evidence for equivalence" (p. 1). We redo the simulation study of Linde et al. (2021) in its entirety but with the different procedures calibrated to have the same predetermined maximum Type I error rate. Our results suggest that, when calibrated in this way, the Bayes factor, HDI-ROPE, and frequentist equivalence tests all have similar-almost exactly-Type II error rates. In general any advocating for frequentist testing as better or worse than Bayesian testing in terms of empirical findings seems dubious at best. If one decides on which underlying principle to subscribe to in tackling a given problem, then the method follows naturally. Bearing in mind that each procedure can be reverse-engineered from the others (at least approximately), trying to use empirical performance to argue for 1 approach over another seems like tilting at windmills. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

[Correction Notice: An Erratum for this article was reported in Vol 29(3) of Psychological Methods (see record 2025-28068-002). In the article, the copyright attribution was incorrectly listed, and the Creative Commons CC BY 4.0 license disclaimer was incorrectly omitted from the author note. The correct copyright is "© 2023 The Author(s)," and the omitted disclaimer is below: Open Access funding provided by University College London: This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0; https://creativecommons.org/licenses/by/ 4.0). This license permits copying and redistributing the work in any medium or format, as well as adapting the material for any purpose, even commercially.] Theories are among the most important tools of science. Lewin (1943) already noted "There is nothing as practical as a good theory." Although psychologists discussed problems of theory in their discipline for a long time, weak theories are still widespread in most subfields. One possible reason for this is that psychologists lack the tools to systematically assess the quality of their theories. Thagard (1989) developed a computational model for formal theory evaluation based on the concept of explanatory coherence. However, there are possible improvements to Thagard's (1989) model and it is not available in software that psychologists typically use. Therefore, we developed a new implementation of explanatory coherence based on the Ising model. We demonstrate the capabilities of this new Ising model of Explanatory Coherence (IMEC) on several examples from psychology and other sciences. In addition, we implemented it in the R-package IMEC to assist scientists in evaluating the quality of their theories in practice. (PsycInfo Database Record (c) 2024 APA, all rights reserved).