A Practical Guide to Variable Selection in Structural Equation Models with Regularized MIMIC Models

Artificial intelligence to evaluate the impact of urban green and blue spaces on chlorophyll-a concentrations

Urbanization is accelerating rapidly, highlighting the critical role of aligning with sustainable development goals, urban green and blue spaces (UGS and UBS). These spaces play a crucial role in enhancing the health and well-being of city residents in terms of ecology. Acknowledging the importance of urban ecology, this study utilizes Sentinel-2A data and support vector machine classification, aimed to identify UGS and UBS. To examine the connections between UGS and UBS, specific indices, spectral bands, and textures were calculated. Additionally, the concentration of chlorophyll, a vital indicator of ecological health, was assessed using three indices. Structural equation modeling was employed to elucidate the relationship between UGS and UBS and their impact on chlorophyll concentration for the years 2017 and 2023. In the 2017 model, UGS exhibited a positive path coefficient (0.25) with chlorophyll-a, indicating that an increase in UGS is associated with an increase in chlorophyll levels. Conversely, in 2023, the path coefficient turned negative (- 0.83), presenting a stark contrast to the 2017 model. This shift suggests potential environmental or urban development changes, such as alterations in the quality or type of urban green spaces, potentially including more non-native or ornamental plants that contribute less to overall chlorophyll levels. UGS can be subjected to pollution, soil compaction, and other stressors that reduce plant health. Similarly, the UBS showed an increase in its path coefficient from - 0.99 in 2017 to - 1.8 in 2023, suggesting improvements such as cleaner water or urban planning strategies aimed at reducing water pollution. The consistent negative relationship across both years suggests that urban water bodies are not contributing to Chl levels due to complex interactions of water bodies with their urban surroundings. However, further research is essential to delve into these dynamics and comprehend the implications for urban ecological planning and sustainability.

The impact of neuropsychiatric burden on Restless Legs Syndrome (RLS) disease severity

OBJECTIVE

In patients with Restless Legs Syndrome (RLS), neuropsychiatric comorbidities like anxiety, depression, and somatization are common, yet the precise connection between somatization and RLS severity remains unclear. This study aims to elucidate the influence of neuropsychiatric comorbidities on RLS severity, focusing particularly on the role of somatization.

METHODS

This cross-sectional analytical study was conducted at a tertiary hospital. All 113 RLS patients who followed in neurology clinic for at least a year were invited, and 87 participated. Data collection included sociodemographic details, the International Restless Legs Syndrome Study Group rating scale (IRLS), the Beck Depression Inventory, Beck Anxiety Scale, and Somatization Scale. Elastic-net regularized path analysis was used as the statistical method.

RESULTS

Among the 87 participants (70.1 % female, mean age 52.5 ± 13.2 years), the mean duration of RLS diagnosis was 4.95 ± 4.53 years. Univariate statistics revealed positive correlations among RLS severity, anxiety, depression, and somatization. Path analysis showed that somatization was associated with RLS severity (p = 0.014). Anxiety had no direct effect on RLS severity but influenced it indirectly through its positive association with somatization (p < 0.001). Depression was found to have no effect on RLS severity, either directly or through somatization.

CONCLUSIONS

The relationship between anxiety and RLS severity is mediated by somatization. Furthermore, the association between RLS severity and somatization appears to be more significant than previously recognized, highlighting the importance of considering somatization in addressing the neuropsychiatric burden of RLS patients.

Optimizing epilepsy treatment: the impact of circadian rhythms and medication timing on conversion rates and survival

BACKGROUND

The progression from isolated seizures to status epilepticus (SE) is a critical clinical issue. This study explores the influence of circadian rhythms on this transition and assesses the impact of medication timing on SE conversion rates and patient survival.

AIM

To determine the circadian patterns in the transition from isolated seizures to SE and to evaluate the efficacy of medication timing in reducing this conversion and improving survival outcomes.

DESIGN AND METHODS

Utilizing the eICU Collaborative Research Database v2.0, a retrospective analysis was performed on patients at risk of SE conversion. The study analyzed the correlation between SE conversion timing and AEDs administration in relation to circadian rhythms, using a Logit model to evaluate the impact of medication timing on SE conversion and survival.

RESULTS

The transition from isolated seizures to SE showed distinct circadian patterns, with a delayed acrophase. Early night-time AEDs administration significantly reduced conversion rates. Medication timing also influenced survival rates, with higher survival during specific periods.

CONCLUSION

Circadian rhythms significantly affect the transition from isolated seizures to SE. Timely AEDs administration is crucial for reducing conversions and improving survival. A chronotherapeutic approach aligning AEDs administration with individual circadian vulnerabilities could advance epilepsy management in ICU settings. Future research should focus on personalized medication strategies that utilize circadian rhythms to optimize treatment effects.

Abiotic Factors Modulating Metabolite Composition in Brown Algae (Phaeophyceae): Ecological Impacts and Opportunities for Bioprospecting of Bioactive Compounds

Brown algae are vital structural elements and contributors to biodiversity in marine ecosystems. These organisms adapt to various environmental challenges by producing primary and secondary metabolites crucial for their survival, defense, and resilience. Besides their ecological role, these diverse metabolites have potential for biotechnological applications in industries including pharmaceuticals, cosmetics, and food. A literature review was conducted encompassing studies from 2014-2024, evaluating the effects of hydrodynamics, temperature, light, nutrients, seasonality, and salinity on the chemical profiles of various Phaeophyceae algae species. Thirty original articles spanning 69 species from the Sargassaceae, Dictyotaceae, Fucaceae, and Scytosiphonaceae families were analyzed and systematically arranged, with a focus on methodologies and key findings. This review furthers ecological discussions on each environmental factor and explores the biotechnological potential of metabolites such as polysaccharides, fatty acids, phenolics, diterpenes, and pigments. The information in this work is beneficial for metabolite bioprospecting and in vitro cultivation models as well as indoor and outdoor cultivation studies.

Statistical and machine learning models for location-specific crop yield prediction using weather indices

Crop yield prediction gains growing importance for all stakeholders in agriculture. Since the growth and development of crops are fully connected with many weather factors, it is inevitable to incorporate meteorological information into yield prediction mechanism. The changes in climate-yield relationship are more pronounced at a local level than across relatively large regions. Hence, district or sub-region-level modeling may be an appropriate approach. To obtain a location- and crop-specific model, different models with different functional forms have to be explored. This systematic review aims to discuss research papers related to statistical and machine-learning models commonly used to predict crop yield using weather factors. It was found that Artificial Neural Network (ANN) and Multiple Linear Regression were the most applied models. Support Vector Regression (SVR) model has a high success ratio as it performed well in most of the cases. The optimization options in ANN and SVR models allow us to tune models to specific patterns of association between weather conditions of a location and crop yield. ANN model can be trained using different activation functions with optimized learning rate and number of hidden layer neurons. Similarly, the SVR model can be trained with different kernel functions and various combinations of hyperparameters. Penalized regression models namely, LASSO and Elastic Net are better alternatives to simple linear regression. The nonlinear machine learning models namely, SVR and ANN were found to perform better in most of the cases which indicates there exists a nonlinear complex association between crop yield and weather factors.

Affordability, negative experiences, perceived racism, and health care system distrust among black American women aged 45 and over

Black Americans (AA) face a confluence of challenges when seeking care including unaffordable costs, negative experiences with providers, racism, and distrust in the healthcare system. This study utilized linear regressions and mediation analysis to explore the interconnectedness of these challenges within a community-based sample of 313 AA women aged 45 and older. Approximately 23% of participants reported affordability problems, while 44% had a negative experience with a provider. In the initial linear regression model excluding perceived racism, higher levels of distrust were observed among women reporting affordability problems (β = 2.66; p = 0.003) or negative experiences with a healthcare provider (β = 3.02; p = <0.001). However, upon including perceived racism in the model, it emerged as a significant predictor of distrust (β = 0.81; p = < 0.001), attenuating the relationships between affordability and distrust (β = 1.74; p = 0.030) and negative experience with a provider and distrust (β = 1.79; p = 0.009). Mediation analysis indicated that perceived racism mediated approximately 35% and 41% of the relationships between affordability and distrust and negative experience with a provider and distrust, respectively. These findings underscore the critical imperative of addressing racism in the efforts to mitigate racial disparities in healthcare. Future research should explore the applicability of these findings to other marginalized populations.

Bayesian factor selection in a hybrid approach to confirmatory factor analysis

To investigate latent structures of measured variables, various factor structures are used for confirmatory factor analysis, including higher-order models and more flexible bifactor models. In practice, measured variables may also have relatively small or moderate non-zero loadings on multiple group factors, which form cross loadings. The selection of correct and 'identifiable' latent structures is important to evaluate an impact of constructs of interest in the confirmatory factor analysis model. Herein, we first discuss the identifiability condition that allows several cross loadings of the models with underlying bifactor structures. Then, we implement Bayesian variable selection allowing cross loadings on bifactor structures using the spike and slab prior. Our approaches evaluate the inclusion probability for all group factor loadings and utilize known underlying structural information, making our approaches not entirely exploratory. Through a Monte Carlo study, we demonstrate that our methods can provide more accurately identified results than other available methods. For the application, the SF-12 version 2 scale, a self-report health-related quality of life survey is used. The model selected by our proposed methods is more parsimonious and has a better fit index compared to other models including the ridge prior selection and strict bifactor model.

Gray and White Matter Metrics Demonstrate Distinct and Complementary Prediction of Differences in Cognitive Performance in Children: Findings from ABCD (N = 11,876)

Individual differences in cognitive performance in childhood are a key predictor of significant life outcomes such as educational attainment and mental health. Differences in cognitive ability are governed in part by variations in brain structure. However, studies commonly focus on either gray or white matter metrics in humans, leaving open the key question as to whether gray or white matter microstructure plays distinct or complementary roles supporting cognitive performance. To compare the role of gray and white matter in supporting cognitive performance, we used regularized structural equation models to predict cognitive performance with gray and white matter measures. Specifically, we compared how gray matter (volume, cortical thickness, and surface area) and white matter measures (volume, fractional anisotropy, and mean diffusivity) predicted individual differences in cognitive performance. The models were tested in 11,876 children (ABCD Study; 5,680 female, 6,196 male) at 10 years old. We found that gray and white matter metrics bring partly nonoverlapping information to predict cognitive performance. The models with only gray or white matter explained respectively 15.4 and 12.4% of the variance in cognitive performance, while the combined model explained 19.0%. Zooming in, we additionally found that different metrics within gray and white matter had different predictive power and that the tracts/regions that were most predictive of cognitive performance differed across metrics. These results show that studies focusing on a single metric in either gray or white matter to study the link between brain structure and cognitive performance are missing a key part of the equation.

Grey and white matter metrics demonstrate distinct and complementary prediction of differences in cognitive performance in children: Findings from ABCD (N= 11 876)

Individual differences in cognitive performance in childhood are a key predictor of significant life outcomes such as educational attainment and mental health. Differences in cognitive ability are governed in part by variations in brain structure. However, studies commonly focus on either grey or white matter metrics in humans, leaving open the key question as to whether grey or white matter microstructure play distinct or complementary roles supporting cognitive performance. To compare the role of grey and white matter in supporting cognitive performance, we used regularized structural equation models to predict cognitive performance with grey and white matter measures. Specifically, we compared how grey matter (volume, cortical thickness and surface area) and white matter measures (volume, fractional anisotropy and mean diffusivity) predicted individual differences in cognitive performance. The models were tested in 11,876 children (ABCD Study, 5680 female; 6196 male) at 10 years old. We found that grey and white matter metrics bring partly non-overlapping information to predict cognitive performance. The models with only grey or white matter explained respectively 15.4% and 12.4% of the variance in cognitive performance, while the combined model explained 19.0%. Zooming in we additionally found that different metrics within grey and white matter had different predictive power, and that the tracts/regions that were most predictive of cognitive performance differed across metric. These results show that studies focusing on a single metric in either grey or white matter to study the link between brain structure and cognitive performance are missing a key part of the equation.

The positive-negative-competence (PNC) model of psychological responses to representations of robots

Robots are becoming an increasingly prominent part of society. Despite their growing importance, there exists no overarching model that synthesizes people's psychological reactions to robots and identifies what factors shape them. To address this, we created a taxonomy of affective, cognitive and behavioural processes in response to a comprehensive stimulus sample depicting robots from 28 domains of human activity (for example, education, hospitality and industry) and examined its individual difference predictors. Across seven studies that tested 9,274 UK and US participants recruited via online panels, we used a data-driven approach combining qualitative and quantitative techniques to develop the positive-negative-competence model, which categorizes all psychological processes in response to the stimulus sample into three dimensions: positive, negative and competence-related. We also established the main individual difference predictors of these dimensions and examined the mechanisms for each predictor. Overall, this research provides an in-depth understanding of psychological functioning regarding representations of robots.

The Hitchhiker's guide to longitudinal models: A primer on model selection for repeated-measures methods

Longitudinal data are becoming increasingly available in developmental neuroimaging. To maximize the promise of this wealth of information on how biology, behavior, and cognition change over time, there is a need to incorporate broad and rigorous training in longitudinal methods into the repertoire of developmental neuroscientists. Fortunately, these models have an incredibly rich tradition in the broader developmental sciences that we can draw from. Here, we provide a primer on longitudinal models, written in a beginner-friendly (and slightly irreverent) manner, with a particular focus on selecting among different modeling frameworks (e.g., multilevel versus latent curve models) to build the theoretical model of development a researcher wishes to test. Our aims are three-fold: (1) lay out a heuristic framework for longitudinal model selection, (2) build a repository of references that ground each model in its tradition of methodological development and practical implementation with a focus on connecting researchers to resources outside traditional neuroimaging journals, and (3) provide practical resources in the form of a codebook companion demonstrating how to fit these models. These resources together aim to enhance training for the next generation of developmental neuroscientists by providing a solid foundation for future forays into advanced modeling applications.

Comparing Bayesian Variable Selection to Lasso Approaches for Applications in Psychology

In the current paper, we review existing tools for solving variable selection problems in psychology. Modern regularization methods such as lasso regression have recently been introduced in the field and are incorporated into popular methodologies, such as network analysis. However, several recognized limitations of lasso regularization may limit its suitability for psychological research. In this paper, we compare the properties of lasso approaches used for variable selection to Bayesian variable selection approaches. In particular we highlight advantages of stochastic search variable selection (SSVS), that make it well suited for variable selection applications in psychology. We demonstrate these advantages and contrast SSVS with lasso type penalization in an application to predict depression symptoms in a large sample and an accompanying simulation study. We investigate the effects of sample size, effect size, and patterns of correlation among predictors on rates of correct and false inclusion and bias in the estimates. SSVS as investigated here is reasonably computationally efficient and powerful to detect moderate effects in small sample sizes (or small effects in moderate sample sizes), while protecting against false inclusion and without over-penalizing true effects. We recommend SSVS as a flexible framework that is well-suited for the field, discuss limitations, and suggest directions for future development.

Predictors of Psychological Strain and Allostatic Load in Teachers: Examining the Long-Term Effects of Biopsychosocial Risk and Protective Factors Using a LASSO Regression Approach

Teacher stress significantly challenges teachers' health, teaching quality, and students' motivation and achievement. Thus, it is crucial to identify factors that effectively prevent it. Using a LASSO regression approach, we examined which factors predict teachers' psychological strain and allostatic load over two years. The study included 42 teachers (28 female, Mage = 39.66, SD = 11.99) and three measurement time points: At baseline, we assessed teachers' (a) self-reports (i.e., on personality, coping styles, and psychological strain), (b) behavioral data (i.e., videotaped lessons), and (c) allostatic load (i.e., body mass index, blood pressure, and hair cortisol concentration). At 1- and 2-year follow-ups, psychological strain and allostatic load biomarkers were reassessed. Neuroticism and perceived student disruptions at baseline emerged as the most significant risk factors regarding teachers' psychological strain two years later, while a positive core self-evaluation was the most important protective factor. Perceived support from other teachers and the school administration as well as adaptive coping styles were protective factors against allostatic load after two years. The findings suggest that teachers' psychological strain and allostatic load do not primarily originate from objective classroom conditions but are attributable to teachers' idiosyncratic perception of this environment through the lens of personality and coping strategies.

Listening to and Believing Derogatory and Threatening Voices

BACKGROUND AND HYPOTHESIS

A plausible cause of distress for voice hearers is listening to and believing the threats and criticisms heard. Qualitative research indicates that patients have understandable reasons to listen. This study aimed to develop the understanding of distress using this listening and believing framework. Measures were developed of listening and believing voices and the reasons, and associations with distress tested.

STUDY DESIGN

A cross-sectional study of patients hearing derogatory and threatening voices (N = 591). Listening and Believing-Assessment and Listening and Believing-Reasons item pools were completed, and assessments of distress. Exploratory and confirmatory factor analyses and structural equation modeling (SEM) were conducted.

STUDY RESULTS

52% (n = 307) of participants believed their voices most or all the time. Listening and believing had 4 factors: active listening, passive listening, believing, and disregarding. Higher levels of believing, active listening, and particularly passive listening were associated with higher levels of anxiety, depression, and voice distress. Reasons for listening and believing formed 7 factors: to better understand the threat; being too worn down to resist; to learn something insightful; being alone with time to listen; voices trying to capture attention; voices sounding like real people; and voices sounding like known people. Each type of reason was associated with active listening, passive listening, and believing. SEM showed that feeling worn down in particular accounted for listening and believing. Test-retest reliability of measures was excellent.

CONCLUSIONS

A framework of listening and believing negative voices has the potential to inform the understanding and treatment of voice distress.

Pathways to primate hip function

Understanding how diverse locomotor repertoires evolved in anthropoid primates is key to reconstructing the clade's evolution. Locomotor behaviour is often inferred from proximal femur morphology, yet the relationship of femoral variation to locomotor diversity is poorly understood. Extant acrobatic primates have greater ranges of hip joint mobility-particularly abduction-than those using more stereotyped locomotion, but how bony morphologies of the femur and pelvis interact to produce different locomotor abilities is unknown. We conducted hypothesis-driven path analyses via regularized structural equation modelling (SEM) to determine which morphological traits are the strongest predictors of hip abduction in anthropoid primates. Seven femoral morphological traits and two hip abduction measures were obtained from 25 primate species, split into broad locomotor and taxonomic groups. Through variable selection and fit testing techniques, insignificant predictors were removed to create the most parsimonious final models. Some morphological predictors, such as femur shaft length and neck-shaft angle, were important across models. Different trait combinations best predicted hip abduction by locomotor or taxonomic group, demonstrating group-specific linkages among morphology, mobility and behaviour. Our study illustrates the strength of SEM for identifying biologically important relationships between morphology and performance, which will have future applications for palaeobiological and biomechanical studies.

Assessment of Neuromuscular and Psychological Function in People with Recurrent Neck Pain during a Period of Remission: Cross-Sectional and Longitudinal Analyses

The aim of this study was to examine for the presence of differences in neuromuscular and psychological function in individuals with recurrent neck pain (RNP) or chronic neck pain (CNP) following a whiplash trauma compared to healthy controls. A secondary aim was to examine whether neuromuscular characteristics together with psychological features in people with RNP were predictive of future painful episodes. Multiple features were assessed including neck disability, kinesiophobia, quality of life, cervical kinematics, proprioception, activity of superficial neck flexor muscles, maximum neck flexion and extension strength, and perceived exertion during submaximal contractions. Overall, those with RNP (n = 22) and CNP (n = 8) presented with higher neck disability, greater kinesiophobia, lower quality of life, slower and irregular neck movements, and less neck strength compared to controls (n = 15). Prediction analysis in the RNP group revealed that a higher number of previous pain episodes within the last 12 months along with lower neck flexion strength were predictors of higher neck disability at a 6-month follow-up. This preliminary study shows that participants with RNP presented with some degree of altered neuromuscular features and poorer psychological function with respect to healthy controls and these features were similar to those with CNP. Neck flexor weakness was predictive of future neck disability.

Structure and stability of symptoms in first episode psychosis: a longitudinal network approach

Early psychosis is characterised by heterogeneity in illness trajectories, where outcomes remain poor for many. Understanding psychosis symptoms and their relation to illness outcomes, from a novel network perspective, may help to delineate psychopathology within early psychosis and identify pivotal targets for intervention. Using network modelling in first episode psychosis (FEP), this study aimed to identify: (a) key central and bridge symptoms most influential in symptom networks, and (b) examine the structure and stability of the networks at baseline and 12-month follow-up. Data on 1027 participants with FEP were taken from the National EDEN longitudinal study and used to create regularised partial correlation networks using the 'EBICglasso' algorithm for positive, negative, and depressive symptoms at baseline and at 12-months. Centrality and bridge estimations were computed using a permutation-based network comparison test. Depression featured as a central symptom in both the baseline and 12-month networks. Conceptual disorganisation, stereotyped thinking, along with hallucinations and suspiciousness featured as key bridge symptoms across the networks. The network comparison test revealed that the strength and bridge centralities did not differ significantly between the two networks (C = 0.096153; p = 0.22297). However, the network structure and connectedness differed significantly from baseline to follow-up (M = 0.16405, p = <0.0001; S = 0.74536, p = 0.02), with several associations between psychosis and depressive items differing significantly by 12 months. Depressive symptoms, in addition to symptoms of thought disturbance (e.g. conceptual disorganisation and stereotyped thinking), may be examples of important, under-recognized treatment targets in early psychosis, which may have the potential to lead to global symptom improvements and better recovery.

Tutorial on the Use of the regsem Package in R

Sparse estimation through regularization is gaining popularity in psychological research. Such techniques penalize the complexity of the model and could perform variable/path selection in an automatic way, and thus are particularly useful in models that have small parameter-to-sample-size ratios. This paper gives a detailed tutorial of the R package regsem, which implements regularization for structural equation models. Example R code is also provided to highlight the key arguments of implementing regularized structural equation models in this package. The tutorial ends by discussing remedies of some known drawbacks of a popular type of regularization, computational methods supported by the package that can improve the selection result, and some other practical issues such as dealing with missing data and categorical variables.

Hippocampal Subfields and Limbic White Matter Jointly Predict Learning Rate in Older Adults

Age-related memory impairments have been linked to differences in structural brain parameters, including cerebral white matter (WM) microstructure and hippocampal (HC) volume, but their combined influences are rarely investigated. In a population-based sample of 337 older participants aged 61-82 years (Mage = 69.66, SDage = 3.92 years), we modeled the independent and joint effects of limbic WM microstructure and HC subfield volumes on verbal learning. Participants completed a verbal learning task of recall over five repeated trials and underwent magnetic resonance imaging (MRI), including structural and diffusion scans. We segmented three HC subregions on high-resolution MRI data and sampled mean fractional anisotropy (FA) from bilateral limbic WM tracts identified via deterministic fiber tractography. Using structural equation modeling, we evaluated the associations between learning rate and latent factors representing FA sampled from limbic WM tracts, and HC subfield volumes, and their latent interaction. Results showed limbic WM and the interaction of HC and WM-but not HC volume alone-predicted verbal learning rates. Model decomposition revealed HC volume is only positively associated with learning rate in individuals with higher WM anisotropy. We conclude that the structural characteristics of limbic WM regions and HC volume jointly contribute to verbal learning in older adults.

A Comparison of Regularized Maximum-Likelihood, Regularized 2-Stage Least Squares, and Maximum-Likelihood Estimation with Misspecified Models, Small Samples, and Weak Factor Structure

Several structural equation modeling estimation methods have recently been developed to alleviate problems associated with model misspecification. Two of the more popular such approaches are 2-stage least squares and regularization methods. Prior work examining the performance of these estimators has generally focused on problems with adequately sized samples and relatively large factor loadings. In contrast, relatively little research has been conducted comparing these estimation techniques with small samples and weak loadings, though both conditions are not uncommon in the multivariate modeling. The current simulation study focused on comparing these relatively new structural estimation methods for misspecified models (e.g., misspecified interactions and cross-loadings) with small samples and relatively weak factor loadings. Results indicated that regularized 2-stage least squares estimation performed better compared to the regularized structural equation modeling framework for small samples and with weak factor loadings. Implications and guidelines for applied researchers are presented.

Exploratory factor analysis with structured residuals for brain network data

Dimension reduction is widely used and often necessary to make network analyses and their interpretation tractable by reducing high-dimensional data to a small number of underlying variables. Techniques such as exploratory factor analysis (EFA) are used by neuroscientists to reduce measurements from a large number of brain regions to a tractable number of factors. However, dimension reduction often ignores relevant a priori knowledge about the structure of the data. For example, it is well established that the brain is highly symmetric. In this paper, we (a) show the adverse consequences of ignoring a priori structure in factor analysis, (b) propose a technique to accommodate structure in EFA by using structured residuals (EFAST), and (c) apply this technique to three large and varied brain-imaging network datasets, demonstrating the superior fit and interpretability of our approach. We provide an R software package to enable researchers to apply EFAST to other suitable datasets.

Prior Sensitivity in Bayesian Structural Equation Modeling for Sparse Factor Loading Structures

Bayesian structural equation modeling (BSEM) is a flexible tool for the exploration and estimation of sparse factor loading structures; that is, most cross-loading entries are zero and only a few important cross-loadings are nonzero. The current investigation was focused on the BSEM with small-variance normal distribution priors (BSEM-N) for both variable selection and model estimation. The prior sensitivity in BSEM-N was explored in factor analysis models with sparse loading structures through a simulation study (Study 1) and an empirical example (Study 2). Study 1 examined the prior sensitivity in BSEM-N based on the model fit, population model recovery, true and false positive rates, and parameter estimation. Seven shrinkage priors on cross-loadings and five noninformative/vague priors on other model parameters were examined. Study 2 provided a real data example to illustrate the impact of various priors on model fit and parameter selection and estimation. Results indicated that when the 95% credible intervals of shrinkage priors barely covered the population cross-loading values, it resulted in the best balance between true and false positives. If the goal is to perform variable selection, a sparse cross-loading structure is required, preferably with a minimal number of nontrivial cross-loadings and relatively high primary loading values. To improve parameter estimates, a relatively large prior variance is preferred. When cross-loadings are relatively large, BSEM-N with zero-mean priors is not recommended for the estimation of cross-loadings and factor correlations.

Crystallized and fluid intelligence are predicted by microstructure of specific white-matter tracts

Studies of the neural basis of intelligence have focused on comparing brain imaging variables with global scales instead of the cognitive domains integrating these scales or quotients. Here, the relation between mean tract-based fractional anisotropy (mTBFA) and intelligence indices was explored. Deterministic tractography was performed using a regions of interest approach for 10 white-matter fascicles along which the mTBFA was calculated. The study sample included 83 healthy individuals from the second wave of the Cuban Human Brain Mapping Project, whose WAIS-III intelligence quotients and indices were obtained. Inspired by the "Watershed model" of intelligence, we employed a regularized hierarchical Multiple Indicator, Multiple Causes model (MIMIC), to assess the association of mTBFA with intelligence scores, as mediated by latent variables summarizing the indices. Regularized MIMIC, used due to the limited sample size, selected relevant mTBFA by means of an elastic net penalty and achieved good fits to the data. Two latent variables were necessary to describe the indices: Fluid intelligence (Perceptual Organization and Processing Speed indices) and Crystallized Intelligence (Verbal Comprehension and Working Memory indices). Regularized MIMIC revealed effects of the forceps minor tract on crystallized intelligence and of the superior longitudinal fasciculus on fluid intelligence. The model also detected the significant effect of age on both latent variables.

Improving the assessment of measurement invariance: Using regularization to select anchor items and identify differential item functioning

A common challenge in the behavioral sciences is evaluating measurement invariance, or whether the measurement properties of a scale are consistent for individuals from different groups. Measurement invariance fails when differential item functioning (DIF) exists, that is, when item responses relate to the latent variable differently across groups. To identify DIF in a scale, many data-driven procedures iteratively test for DIF one item at a time while assuming other items have no DIF. The DIF-free items are used to anchor the scale of the latent variable across groups, identifying the model. A major drawback to these iterative testing procedures is that they can fail to select the correct anchor items and identify true DIF, particularly when DIF is present in many items. We propose an alternative method for selecting anchors and identifying DIF. Namely, we use regularization, a machine learning technique that imposes a penalty function during estimation to remove parameters that have little impact on the fit of the model. We focus specifically here on a lasso penalty for group differences in the item parameters within the two-parameter logistic item response theory model. We compare lasso regularization with the more commonly used likelihood ratio test method in a 2-group DIF analysis. Simulation and empirical results show that when large amounts of DIF are present and sample sizes are large, lasso regularization has far better control of Type I error than the likelihood ratio test method with little decrement in power. This provides strong evidence that lasso regularization is a promising alternative for testing DIF and selecting anchors. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Structural brain imaging correlates of general intelligence in UK Biobank

The associations between indices of brain structure and measured intelligence are unclear. This is partly because the evidence to-date comes from mostly small and heterogeneous studies. Here, we report brain structure-intelligence associations on a large sample from the UK Biobank study. The overall N = 29,004, with N = 18,426 participants providing both brain MRI and at least one cognitive test, and a complete four-test battery with MRI data available in a minimum N = 7201, depending upon the MRI measure. Participants' age range was 44-81 years (M = 63.13, SD = 7.48). A general factor of intelligence (g) was derived from four varied cognitive tests, accounting for one third of the variance in the cognitive test scores. The association between (age- and sex- corrected) total brain volume and a latent factor of general intelligence is r = 0.276, 95% C.I. = [0.252, 0.300]. A model that incorporated multiple global measures of grey and white matter macro- and microstructure accounted for more than double the g variance in older participants compared to those in middle-age (13.6% and 5. 4%, respectively). There were no sex differences in the magnitude of associations between g and total brain volume or other global aspects of brain structure. The largest brain regional correlates of g were volumes of the insula, frontal, anterior/superior and medial temporal, posterior and paracingulate, lateral occipital cortices, thalamic volume, and the white matter microstructure of thalamic and association fibres, and of the forceps minor. Many of these regions exhibited unique contributions to intelligence, and showed highly stable out of sample prediction.