Statistical Issues Research Articles

Review of Statistical and Methodological Issues in the Forensic Prediction of Malingering from Validity Tests: Part II-Methodological Issues.

Forensic neuropsychological examinations to detect malingering in patients with neurocognitive, physical, and psychological dysfunction have tremendous social, legal, and economic importance. Thousands of studies have been published to develop and validate methods to forensically detect malingering based largely on approximately 50 validity tests, including embedded and stand-alone performance and symptom validity tests. This is Part II of a two-part review of statistical and methodological issues in the forensic prediction of malingering based on validity tests. The Part I companion paper explored key statistical issues. Part II examines related methodological issues through conceptual analysis, statistical simulations, and reanalysis of findings from prior validity test validation studies. Methodological issues examined include the distinction between analog simulation and forensic studies, the effect of excluding too-close-to-call (TCTC) cases from analyses, the distinction between criterion-related and construct validation studies, and the application of the Revised Quality Assessment of Diagnostic Accuracy Studies tool (QUADAS-2) in all Test of Memory Malingering (TOMM) validation studies published within approximately the first 20years following its initial publication to assess risk of bias. Findings include that analog studies are commonly confused for forensic validation studies, and that construct validation studies are routinely presented as if they were criterion-reference validation studies. After accounting for the exclusion of TCTC cases, actual classification accuracy was found to be well below claimed levels. QUADAS-2 results revealed that extant TOMM validation studies all had a high risk of bias, with not a single TOMM validation study with low risk of bias. Recommendations include adoption of well-established guidelines from the biomedical diagnostics literature for good quality criterion-referenced validation studies and examination of implications for malingering determination practices. Design of future studies may hinge on the availability of an incontrovertible reference standard of the malingering status of examinees.

Bayesian and non-Bayesian estimation of dynamic cumulative residual Tsallis entropy for moment exponential distribution under progressive censored type II

Abstract The dynamic cumulative residual (DCR) entropy is a helpful randomness metric that may be used in survival analysis. A challenging issue in statistics and machine learning is the estimation of entropy measures. This article uses progressive censored type II (PCT-II) samples to estimate the DCR Tsallis entropy (DCRTE) for the moment exponential distribution. The non-Bayesian and Bayesian approaches are the recommended estimating strategies. We obtain the DCRTE Bayesian estimator using the gamma and uniform priors via symmetric and asymmetric (linear exponential and general entropy) loss functions (LoFs). The Metropolis–Hastings algorithm is employed to generate Markov chain Monte Carlo samples from the posterior distribution. The accuracy of different estimates for various sample sizes, is implemented via Monte Carlo simulations. Generally, we note based on the simulation study that, in the majority of cases, the DCRTE Bayesian estimates under general entropy followed by linear exponential LoFs are preferable to the others. The accuracy measure of DCRTE Bayesian estimates using a gamma prior has smaller values than the others using a uniform prior. As sample sizes grow, the Bayesian estimates of the DCRTE are closer to the true value. Finally, analysis of the leukemia data confirmed the proposed estimators.

Framing the statistical criteria for accelerated approval of oncology drugs: A pathway for front runners

FDA has recently proposed a one-trial approach in Project FrontRunner for both accelerated approval and regular approval in the same trial. The goal of this short communication is to make the regulatory criteria for decision-making on accelerated approval based on the interim analysis more explicit. Two related high-level statistical issues are addressed under simplified yet representative scenarios. The first is at what Type I error level should the interim analysis be tested? To conform with the conventional two-trial approach, it may be tested at the 1.0% level without paying penalty for regular approval. This criterion is more relaxed than current practice. The second is what may be considered as adequate totality of evidence for accelerated approval. It turns out that, with the first criterion met, the observed treatment effect of the clinical endpoint intended for regular approval needs to be close to the target effect for the study. The criterion helps make the regulatory decision more transparent and objective.

14th Annual University of Pennsylvania Conference on statistical issues in clinical trials/subgroup analysis in clinical trials: Opportunities and challenges (morning panel discussion).

14th Annual University of Pennsylvania Conference on statistical issues in clinical trials/subgroup analysis in clinical trials: Opportunities and challenges (morning panel discussion).

Launching “methods and statistics tutorials”: A collection of resources for systematic reviewers

Robust and innovative methods are the lynchpin of Cochrane and remain the basis of its reputation as the home of high-quality systematic reviews. As methodology evolves to respond to the evidence needs of our stakeholders, it is crucial for Cochrane to be able to offer timely and clear advice to those working on our reviews. Cochrane's Methods Support Unit (MSU) was established in 2019 [1] to provide methodological and statistical support to authors and editors working on Cochrane protocols and reviews. A key challenge in the first 2 years of the Unit was the introduction of our new tool [2] to assess the risk of bias in randomized controlled trials (RCTs), but we have dealt with a multiplicity of different issues since being established. By March 2023, the Unit has dealt with over 1200 referrals relating to protocols, systematic reviews, updates of reviews, and statistical and methodological queries relating to reviews. Our work has ranged from full methodological and statistical reviews of research incorporating network meta-analyses to individual queries about the use of an intracluster correlation coefficient to adjust data from an individual study. It quickly became apparent that similar queries were often raised and we noticed common problems when working on reviews. Our monthly web clinic series was set up to allow authors and editors to raise questions and then later to provide guidance on a specific topic in addition [3]. However, we also felt that there is a need for accessible help with these issues which would be instantly available. Professor Doug Altman was a leader and pioneer in Cochrane and in the field of medical statistics [4]. One of Professor Altman's legacies was the Statistics Notes series in The BMJ, which provided clear advice on statistical concepts to medical researchers. Inspired by this work, former and current MSU managers (Dr Kerry Dwan and Rachel Richardson) are launching this series to provide easily accessible advice to the evidence synthesis community on the common methodological and statistical issues that we have observed within Cochrane. The hope is to make this as interactive and easy to understand as possible. In addition, the series will collaborate with Cochrane Training [5] to complement the articles in the series with visual learning, such as short videos or quick e-learning checks, to help those who learn better by doing or watching rather than reading. There are various sources of advice on methodological and statistical issues already available to reviewers, including the Cochrane Handbook for Systematic Reviews of Intervention [6] and resources from specialized Cochrane Methods Groups. This series will complement existing sources by responding to the day-to-day difficulties encountered by reviewers when implementing systematic review methods. The MSU is in a unique position to provide advice; having dealt with thousands of queries and requests for peer review since our inception, we know what reviewers find difficult and also why these issues cause confusion. We also believe that linking the articles with resources from Cochrane Training will make this series uniquely useful. dealing with data from crossover RCTs; the correct way to undertake and interpret subgroup analyses; inclusion of multiarm trials in meta-analysis; heterogeneity; the use of fixed versus random effects models in meta-analysis. Early articles in the series will focus on the use of effect measures in systematic reviews, specifically the standardized mean difference and the use of risk ratios versus odds ratios. These two topics often cause difficulties for authors and editors. Examples will be given in the articles of when to use these measures, how to calculate them and their interpretation. We will also cover cluster RCTs early in the series. We welcome proposals for this series from members across the evidence synthesis community on areas they see as methodological and statistical challenges for authors. Articles should be short tutorials, use the Commentary article type, and, in general, be less than 1000 words. Cochrane's mission is to produce trusted synthesized evidence, make it accessible to all, and advocate for its use. Ensuring that evidence synthesis methods are implemented wisely and accurately is foundational to us being able to achieve this mission. With the Methods and Statistics Tutorials series, we hope to support and onboard a wider and more diverse community of authors to produce the next generation of high-quality Cochrane Reviews and other evidence syntheses. Kerry Dwan: Conceptualization; writing—original draft; writing—review and editing. Rachel Richardson: Conceptualization; writing—review and editing. Kerry Dwan and Rachel Richardson are former and current employees, respectively, of Cochrane. None The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1002/cesm.12017. Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Improve steganography system using agents software based on statistical and classification technique

In digital communications, information security is a paramount necessity. In the hiding algorithm, there are three basic parameters: security, capacity, and imperceptibility. Therefore, there are many ways to design the steganography algorithm, such as least significant bit (LSB), discrete wave transformation (DWT), and discrete cosine transform (DCT). The aim of this paper is to improve agent software design based on a steganography system. It proposed an agent system based on a support vector machine (SVM) classifier to hide a secret message in a certain cover image. The common dataset for steganography uses 80% training and 20% testing to get accurate results. Developing an agent system depends on six statistical parameters such as energy, standard deviation, histogram, variance, mean, and entropy. This resulted in features classified by the SVM classifier to predict the best cover image to be nominated for embedding. Worthy results were obtained in terms of imperceptibility, attack, and cover image prediction by statistical issues.

Construction of a Nearly Unbiased Statistical Estimator of Sinogram to Address CT Number Bias Issues in Low-Dose Photon Counting CT.

Photon counting detector (PCD)-CT has demonstrated promise to reduce ionizing radiation exposure further and improve spatial resolution. However, when the radiation exposure or the detector pixel size is reduced, image noise is elevated, and the CT number becomes more inaccurate. This exposure level-dependent CT number inaccuracy is referred to as statistical bias. The issue of CT number statistical bias is rooted in the stochastic nature of the detected photon number, N, and a log transformation used to generate the sinogram projection data. Due to the nonlinear nature of the log transform, the statistical mean of the log-transformed data is different from the desired sinogram, the log transform of the statistical mean of N. Consequently, when a single instance of N is measured, as in clinical imaging, the log-transform leads to an inaccurate sinogram and statistically biased CT numbers after reconstruction. This work presents a nearly unbiased and closed-form statistical estimator of sinogram as a simple yet highly effective method to address the statistical bias issue in PCD-CT. Experimental results validated that the proposed approach addresses the CT number bias problem and improves the quantification accuracy of both non-spectral and spectral PCD-CT images. Additionally, the process can slightly reduce noise without adaptive filtering or iterative reconstruction.

Time-series forecasting of mortality rates using transformer

Predicting mortality rates is a crucial issue in life insurance pricing and demographic statistics. Traditional approaches, such as the Lee-Carter model and its variants, predict the trends of mortality rates using factor models, which explain the variations of mortality rates from the perspective of ages, gender, regions, and other factors. Recently, deep learning techniques have achieved great success in various tasks and shown strong potential for time-series forecasting. In this paper, we propose a modified Transformer architecture for predicting mortality rates in major countries around the world. Through the multi-head attention mechanism and positional encoding, the proposed Transformer model extracts key features effectively and thus achieves better performance in time-series forecasting. By using empirical data from the Human Mortality Database, we demonstrate that our Transformer model has higher prediction accuracy of mortality rates than the Lee-Carter model and other classic neural networks. Our model provides a powerful forecasting tool for insurance companies and policy makers.

The Social Brain Hypothesis Thirty Years On: Some Philosophical Pitfalls of Deconstructing Dunbar’s Number

The social brain hypothesis was proposed 30 years ago as an explanation for the fact that primates have much larger brains than all other animals. The claim was that primates live in unusually complex societies, and hence need a large ‘computer’ to manage the relationships involved. The core evidence subsequently provided in support of this claim was a simple statistical relationship between the social group size characteristic of a species and the size of its brain, with humans fitting into this pattern. However, testing evolutionary hypotheses raises some challenging philosophical and statistical issues that are often overlooked, and great care is needed to ensure that we test the hypothesis we think we are testing. Here, I examine some of these challenges and illustrate the traps they can create for the unwary.

Student’s creative model in solving mathematics controversial problems

For students to compete with the rapid advancement in science, technology, and the arts, creativity must be more than just a necessary skill. This study of levels of creativity performed when addressing statistical issues is a follow-up to earlier studies. To create a distinctive model, a controversial aspect was used. The study revealed that there were five levels of creative models, in addition to the three levels of the earlier research: pre-imitation, imitation, modification, combination, and construction. The pre-imitation stage is defined by the subject's limited capacity for imitation. The level of imitation is determined by the act of copying methods even when one does not actually understand them. The modification level is essentially defined by the process of altering a procedure so that it can be applied to solve an issue. The process for merging several settings or problem-solving strategies also serves to define the level of combination. The construction level is determined by the process of developing new methods to handle problems.

Proceedings of the University of Pennsylvania 14th annual conference on statistical issues in clinical trials: Subgroup analysis in randomized clinical trials-Challenges and opportunities.

Proceedings of the University of Pennsylvania 14th annual conference on statistical issues in clinical trials: Subgroup analysis in randomized clinical trials-Challenges and opportunities.

14th Annual University of Pennsylvania Conference on statistical issues in clinical trials/subgroup analysis in clinical trials: Opportunities and challenges (afternoon panel discussion).

14th Annual University of Pennsylvania Conference on statistical issues in clinical trials/subgroup analysis in clinical trials: Opportunities and challenges (afternoon panel discussion).

Perspective: Is the Response of Human Energy Expenditure to Increased Physical Activity Additive or Constrained?

The idea that increasing physical activity directly adds to TEE in humans (additive model) has been challenged by the energy constrained hypothesis (constrained model). This model proposes that increased physical activity decreases other components of metabolism to constrain TEE. There is a logical evolutionary argument for trade-offs in metabolism, but, to date, evidence supporting constraint is subject to several limitations, including cross-sectional and correlational studies with potential methodological issues from extreme differences in body size/composition and lifestyle, potential statistical issues such as regression dilution and spurious correlations, and conclusions drawn from deductive inference rather than direct observation of compensation. Addressing these limitations in future studies, ideally, randomized controlled trials should improve the accuracy of models of human energy expenditure. The available evidence indicates that in many scenarios, the effect of increasing physical activity on TEE will be mostly additive although some energy appears to "go missing" and is currently unaccounted for. The degree of energy balance could moderate this effect even further.

Benchmarking Sentinel-2-derived predictors for long-term burn severity modelling: the 2016–17 Chilean firestorm

ABSTRACT Remotely sensed burn severity modelling is commonly addressed by regressing fire-induced forest biophysical damage measured in the field, using the Composite Burn Index (CBI) method on a pre- to post-fire difference of the normalized burn ratio (∆NBR) representing early fire effects. Although less often addressed, it may also be modelled on a wider post-fire temporal range to assess longer-lasting ecological effects. However, this approach may involve a larger set of potential predictors, posing a high-dimensional statistical issue. This study examines the performance of different spectrotemporal combinations of remotely sensed predictors for long-term burn severity modelling of the two largest wildfires during the central Chilean firestorm in the summer of 2017, in order to assess their predictive contribution and to support their optimal selection. Sentinel-2 images acquired annually in 2016–2020 were used to create different spectrotemporal combinations of predictor sets, encompassing ∆NBR, differenced normalized difference water index (∆NDWI) and differenced normalized difference vegetation index (∆NDVI), as well as differenced bands. Thereafter, they were modelled by CBI data collected in the field, using conventional (least-squares-based) and ridge-regularized linear regressions, and assessed by spatial cross-validation. Among the multiple-variable models, ridge regression outperformed multiple linear regression, the highest accuracies being achieved by the multitemporal sets comprising any of the differenced index types. Among these, ∆NBRs reached the lowest average root mean square error (RMSEavg = 0.591). Results confirm the importance of ∆NBR as burn severity predictor in a long-term assessment context, top-performing for both, single- and multiple-variable models. They also highlight the value of tracking the temporal trajectory of this differenced index when enlarging the number of post-fire image dates involved, and hence, no a-priori knowledge about what predictor to select exists.

High-Dimensional Mediation Analysis: A New Method Applied to Maternal Smoking, Placental DNA Methylation, and Birth Outcomes.

High-dimensional mediation analysis is an extension of unidimensional mediation analysis that includes multiple mediators, and increasingly it is being used to evaluate the indirect omics-layer effects of environmental exposures on health outcomes. Analyses involving high-dimensional mediators raise several statistical issues. Although many methods have recently been developed, no consensus has been reached about the optimal combination of approaches to high-dimensional mediation analyses. We developed and validated a method for high-dimensional mediation analysis (HDMAX2) and applied it to evaluate the causal role of placental DNA methylation in the pathway between exposure to maternal smoking (MS) during pregnancy and gestational age (GA) and birth weight of the baby at birth. HDMAX2 combines latent factor regression models for epigenome-wide association studies with tests for mediation and considers CpGs and aggregated mediator regions (AMRs). HDMAX2 was carefully evaluated using simulated data and compared to state-of-the-art multidimensional epigenetic mediation methods. Then, HDMAX2 was applied to data from 470 women of the Etude des Déterminants pré et postnatals du développement de la santé de l'Enfant (EDEN) cohort. HDMAX2 demonstrated increased power in comparison with state-of-the-art multidimensional mediation methods and identified several AMRs not identified in previous mediation analyses of exposure to MS on birth weight and GA. The results provided evidence for a polygenic architecture of the mediation pathway with a posterior estimate of the overall indirect effect of CpGs and AMRs equal to lower birth weight representing 32.1% of the total effect [standard deviation ]. HDMAX2 also identified AMRs having simultaneous effects both on GA and on birth weight. Among the top hits of both GA and birth weight analyses, regions located in COASY, BLCAP, and ESRP2 also mediated the relationship between GA and birth weight, suggesting reverse causality in the relationship between GA and the methylome. HDMAX2 outperformed existing approaches and revealed an unsuspected complexity of the potential causal relationships between exposure to MS and birth weight at the epigenome-wide level. HDMAX2 is applicable to a wide range of tissues and omic layers. https://doi.org/10.1289/EHP11559.

Statistical framework for fully register based population counts.

The increasing availability of registers or administrative archives has been a strong push towards moving from traditional censuses to combined censuses or even completely register based censuses. In this context, a statistical framework needs to be designed in order to delineate all the statistical issues of the new estimation process. To this aim, a population frame needs to be defined for both surveying and estimation phases. Sampling surveys should be designed for quality assessment and for improving the quality of the register based estimation process. Drawing on similar experiences, a formalisation of the population size estimation process fully based on administrative data is presented. An application to Italian estimation process is reported.

Factor Analysis: Dealing with Response Bias

This paper proposes an innovative method for factor analyzing data that potentially contains individual response bias. Past methods include the use of “ipsative” data, or, related to that, “ipsatized” data. Unfortunately, factor analysis as the main method used for analyzing the dimensionality of data, cannot be applied to ipsative data. In contrast, normalization of data as an alternative method to filter out response bias, is not hampered by the technical statistical issues inherent to applying multivariate techniques to ipsative data. Using high-quality data from a survey in Nepal that makes use of – among others – the High-Performance Organizations (HPO) framework, this paper shows that the traditional approach of directly applying Confirmatory Factor Analysis (CFA) starting from an existing model or theory, is inferior to our approach. Even applying Exploratory Factor Analysis (EFA) to the raw (non-normalized) data before using CFA, is unable to detect the optimal dimensionality, or structure, in the data. A better structure can be obtained by performing EFA on normalized data that corrects for response bias in the raw data. This paper convincingly shows that the newly identified structure is superior to the original structure suggested by the HPO framework. Applying a CFA using the newly detected structure on the raw data, gives excellent goodness-of-fit statistics, with more items retained, and no need of forced methods to improve the model fit. The findings suggest that existing models and questionnaires based on these models, are not necessarily as valid and reliable as empirical studies that make use of traditional analyses seem to suggest. When adopting existing instruments, researchers are advised to critically check the validity and reliability of these instruments – especially those vulnerable to response bias - and to apply the procedures laid out in this paper, in order to enhance the quality of their research, and to inform future researchers who consider using the same instruments or to warn them about the potential shortcomings of these instruments.

A Systematic Review of Joint Spatial and Spatiotemporal Models in Health Research.

With the advancement of spatial analysis approaches, methodological research addressing the technical and statistical issues related to joint spatial and spatiotemporal models has increased. Despite the benefits of spatial modelling of several interrelated outcomes simultaneously, there has been no published systematic review on this topic, specifically when such models would be useful. This systematic review therefore aimed at reviewing health research published using joint spatial and spatiotemporal models. A systematic search of published studies that applied joint spatial and spatiotemporal models was performed using six electronic databases without geographic restriction. A search with the developed search terms yielded 4077 studies, from which 43 studies were included for the systematic review, including 15 studies focused on infectious diseases and 11 on cancer. Most of the studies (81.40%) were performed based on the Bayesian framework. Different joint spatial and spatiotemporal models were applied based on the nature of the data, population size, the incidence of outcomes, and assumptions. This review found that when the outcome is rare or the population is small, joint spatial and spatiotemporal models provide better performance by borrowing strength from related health outcomes which have a higher prevalence. A framework for the design, analysis, and reporting of such studies is also needed.

The Hazards of Daily Stressors: Comparing the Experiences of Sexual and Gender Minority Young Adults to Cisgender Heterosexual Young Adults During the COVID-19 Pandemic

Some individuals may be at greater risk for encountering stressors in daily life than others, especially those with minority identities. Initial evidence shows that the disparities between cisgender heterosexual (CH) individuals and sexual and gender minority (SGM) individuals on stress-related experiences may be exacerbated by the COVID-19 pandemic. We examined the daily stressors experienced by undergraduate students during the COVID-19 pandemic (stressor exposure), the association between the experience of daily stress and same-day negative mood (stressor reactivity), and whether these varied between undergraduate students with SGM identities and their CH counterparts using a 14-day daily diary design. We did not find significant differences between SGM and CH groups on stressor exposure or stressor reactivity. One common feature of daily diary data is right censoring, which is when some individuals do not experience specific events during the study duration. We used multilevel survival analysis, which accounts for right censored data, to examine group differences in the risks of stressor exposure. We discuss the statistical issues involved when right-censored cases are not taken into consideration in studies of stressor exposure and propose multilevel survival analysis as one solution to move the field towards more accurately understanding whether, when, and why SGM individuals are at greater risk for stressors.

Study of Bayesian variable selection method on mixed linear regression models.

Variable selection has always been an important issue in statistics. When a linear regression model is used to fit data, selecting appropriate explanatory variables that strongly impact the response variables has a significant effect on the model prediction accuracy and interpretation effect. redThis study introduces the Bayesian adaptive group Lasso method to solve the variable selection problem under a mixed linear regression model with a hidden state and explanatory variables with a grouping structure. First, the definition of the implicit state mixed linear regression model is presented. Thereafter, the Bayesian adaptive group Lasso method is used to determine the penalty function and parameters, after which each parameter's specific form of the fully conditional posterior distribution is calculated. Moreover, the Gibbs algorithm design is outlined. Simulation experiments are conducted to compare the variable selection and parameter estimation effects in different states. Finally, a dataset of Alzheimer's Disease is used for application analysis. The results demonstrate that the proposed method can identify the observation from different hidden states, but the results of the variable selection in different states are obviously different.