Article
Mathematics
Shen-Ming Lee, Phuoc-Loc Tran, Truong-Nhat Le, Chin-Shang Li
Summary: In this study, we investigate the association between a sensitive characteristic and observed binary random variables using Warner's randomized response technique and a latent class model. We provide an EM algorithm to estimate the parameters and utilize the likelihood ratio test to identify the significant variables. We also use a latent class model to predict individuals' sensitive or non-sensitive group based on observed binary variables.
Article
Mathematical & Computational Biology
Ryota Ishii, Kazushi Maruo, Hisashi Noma, Masahiko Gosho
Summary: Accelerated failure time (AFT) models provide an intuitive estimator for survival data analysis, but the risk of model misspecification can lead to deviation of test size from the nominal level. This article introduces a robust test based on the Bartlett adjustment and nonparametric bootstrap method, which shows close-to-nominal test size in small samples with misspecified error distribution and/or mean structure in AFT models.
STATISTICS IN BIOPHARMACEUTICAL RESEARCH
(2021)
Article
Statistics & Probability
Habiba Khatun, Manas Ranjan Tripathy
Summary: One novel method to compare independent populations is to compare their quantiles. Previous studies have focused on comparing quantiles for normal and exponential models, but datasets may be modeled using other distributions, such as the logistic distribution. In this study, we compare quantiles for two logistic populations using likelihood ratio tests, and propose modifications and a computational approach. Simulation studies show that all proposed tests except the asymptotic likelihood ratio test achieve the nominal level. Additionally, we highlight the importance of our model problem using real-life situations.
COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
(2023)
Article
Mathematical & Computational Biology
Feifei Yan, Qing-Song Xu, Man-Lai Tang, Ziqi Chen
Summary: In this article, a profile likelihood ratio test (PLRT) based on the estimated error density for the multiple linear regression model is developed, which outperforms existing likelihood ratio tests in terms of type I error rates, powers, performance in the absence of error moments, and correct model selection in multiple testing problems. The proposed PLRT is illustrated with analysis of mammalian eye gene expression dataset and concrete compressive strength dataset.
STATISTICS IN MEDICINE
(2021)
Article
Mathematical & Computational Biology
Yuhan Zou, Zuoxiang Peng, Jerry Cornell, Peng Ye, Hua He
Summary: This article addresses the issue of censored data due to biomarker detection limits by proposing a new test method that compares observed censored data with expected results to test latent classes. Simulation studies show that this method performs well in practical applications.
STATISTICS IN MEDICINE
(2021)
Article
Multidisciplinary Sciences
Abdulla Mamun, Sudhir Paul
Summary: The problem of model selection in regression analysis has been well studied through methods like forward selection, backward elimination, and stepwise selection. This paper aims to examine the properties of these procedures in generalized linear regression models, with the normal linear regression model as a special case. The score test is used as the main tool, while other large sample tests such as the F test, likelihood ratio test, and Wald test, as well as AIC and BIC, are included for comparison. Simulation studies are conducted to evaluate the properties of these procedures for symmetric and asymmetric distributions, including normal, Poisson, and binomial regression models. Extensions for skewed distributions and over-dispersed Poisson and binomial regression models are also evaluated. Two health datasets are analyzed using these methods.
Article
Statistics & Probability
Moming Li, Guoqing Diao
Summary: Density-ratio models have gained attention due to their relationship with generalized linear models and applications in missing-data analyses. However, the standard density-ratio model has limitations as it does not account for heterogeneity within the population. To address these limitations, a new density-ratio model with a stratification procedure and dispersion parameters is proposed in this study. The model retains attractive properties of the standard model while allowing violation of the density-ratio assumption for certain covariates, and provides a validation tool using a Kolmogorov-Smirnov-type statistic to check the modeling assumption. Estimation of parameters is done simultaneously using an efficient nonparametric maximum likelihood approach, and the resulting estimators are consistent and asymptotically normal.
Article
Public, Environmental & Occupational Health
Miceline Mesidor, Caroline Sirois, Marc Simard, Denis Talbot
Summary: This paper introduces an approach that uses the bootstrap to validate and quantify the uncertainty in the number of groups in longitudinal finite mixture models. The method is examined through a simulation study and applied to data from the Quebec Integrated Chronic Disease Surveillance System. It allows for the investigation and evaluation of the statistical validity and uncertainty of the groups identified in the original data.
AMERICAN JOURNAL OF EPIDEMIOLOGY
(2023)
Article
Statistics & Probability
Weiwei Zhuang, Yadong Li, Guoxin Qiu
Summary: The research focuses on the application of stochastic dominance in economics and finance, proposing a new estimation method and establishing its statistical inference theory. Simulation results show that this estimation method has advantages in improving estimation efficiency and testability.
JOURNAL OF APPLIED STATISTICS
(2022)
Article
Multidisciplinary Sciences
Amrit Sudershan, Kanak Mahajan, Rakesh K. Panjaliya, Manoj K. Dhar, Parvinder Kumar
Summary: Sampling methods for studying population behavior are uncertain and require representative samples for generalization. Sample size greatly affects the detection of research effects, with smaller samples having lower statistical power and higher risk of missing underlying differences. This study provides a calculation method for determining sample availability during research.
SCIENTIFIC REPORTS
(2022)
Article
Statistics & Probability
Kai Peng, Cheng Peng
Summary: This article proposes new semiparametric and bootstrap tests for the ratio of two variances based on the density ratio model, and compares them with existing tests. It is shown that the ratio of two independent variances follows an asymptotic log-normal distribution. The article also provides theoretical comparisons and numerical comparisons of various tests, as well as examples to illustrate the implementation of the new tests.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2023)
Article
Economics
Prosper Dovonon, Abderrahim Taamouti, Julian Williams
Summary: In this paper, we derive the asymptotic distributions of likelihood-ratio-type test statistics for identifying the eigenvalue structure of both integrated and spot covariance matrices estimated with high-frequency data. Unlike existing approaches, our tests do not require large cross-section and we propose a bootstrap method to approximate the asymptotic distributions. Monte Carlo simulations show that the bootstrap-based test controls sample size and has good power.
JOURNAL OF ECONOMETRICS
(2022)
Article
Multidisciplinary Sciences
Joas S. Santos, Francisco Cribari-Neto
Summary: This paper focuses on the potential misleading results of hypothesis testing when the sample size is small. To address this issue, a Bartlett-corrected likelihood ratio test is proposed. Monte Carlo simulations are conducted to compare the performance of the corrected test with the standard likelihood ratio test in finite samples. The results indicate that the corrected test has excellent control of the type I error frequency. An empirical application is also presented and discussed.
ADVANCED THEORY AND SIMULATIONS
(2023)
Article
Health Care Sciences & Services
Joshua R. Nugent, Ken P. Kleinman
Summary: Using simulations with a random-intercept LMM structure, this study examines Type I error rates of LRT and Wald test with different degrees of freedom choices across varying combinations of cluster size, number of clusters, and ICC. Results indicate that the LRT may be anti-conservative in certain scenarios, while Wald tests with specific DF methods can maintain Type I error control.
BMC MEDICAL RESEARCH METHODOLOGY
(2021)
Article
Mathematical & Computational Biology
Guohai Zhou, Lang Wu
Summary: This article introduces multiparameter one-sided or constrained tests for NLME models with censored responses, such as viral dynamic models with viral loads subject to lower detection limits. Through simulations, it is demonstrated that the proposed tests are more powerful than the corresponding two-sided or unrestricted tests. The methods are applied to two AIDS datasets with new findings.
STATISTICS IN MEDICINE
(2021)
Article
Mathematics, Interdisciplinary Applications
Sunghoon Kim, Ashley Stadler Blank, Wayne S. DeSarbo, Jeroen K. Vermunt
Summary: The study introduces a hierarchical clustering model based on NFL fan survey data, which reveals the dispersion characteristics between teams and fans more effectively. The results show that NFL fans can be divided into three non-contiguous team segments, with differences in consumer behavior being observed.
JOURNAL OF CLASSIFICATION
(2022)
Article
Statistics & Probability
F. J. Clouth, S. Pauws, F. Mols, J. K. Vermunt
Summary: The bias-adjusted three-step LCA is extended to incorporate IPW, separating the estimation of the measurement model from the treatment effect estimation. This new approach solves conceptual issues and makes it easier for model selection and the use of multiple imputation. The implementation of this method in Latent GOLD is evaluated in a simulation study and illustrated with data of prostate cancer patients.
ADVANCES IN DATA ANALYSIS AND CLASSIFICATION
(2022)
Article
Mathematics, Interdisciplinary Applications
Leonie V. D. E. Vogelsmeier, Jeroen K. Vermunt, Anne Bulow, Kim De Roover
Summary: This paper introduces a simplified method for estimating the latent Markov factor analysis (LMFA) and facilitates the exploration of covariate effects. A real data example demonstrates the empirical value of this method.
MULTIVARIATE BEHAVIORAL RESEARCH
(2023)
Article
Mathematical & Computational Biology
Stan Altan, Dhammika Amaratunga, Javier Cabrera, Jeonifer Garren, Helena Geys, John Kolassa, David LeBlond, Dingzhou Li, Jason Liao, Jia Liu, Mariusz Lubomirski, Guillermo Miro-Quesada, Steven Novick, Martin Otava, John Peterson, Katharina Reckermann, Tim Schofield, Charles Tan, Kanaka Tatikola, Fetene Tekle, Jennifer Thomas, Kim Vukovinsky
STATISTICS IN BIOPHARMACEUTICAL RESEARCH
(2023)
Article
Psychology, Educational
E. Damiano D'Urso, Jesper Tijmstra, Jeroen K. Vermunt, Kim De Roover
Summary: Assessing the measurement model of self-report scales is crucial for obtaining valid measurements of individuals' latent psychological constructs. This study found that the acquiescence response style has an impact on the measurement results, especially in balanced scales. The use of informed rotation approaches can help resolve this issue.
EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT
(2023)
Article
Psychology, Educational
Martijn Schoenmakers, Jesper Tijmstra, Jeroen Vermunt, Maria Bolsinova
Summary: This article examines two frequently used item response theory (IRT) models, the multidimensional nominal response model (MNRM) and the IRTree model. The study reveals conceptual differences between these models, which result in different conclusions about the size and presence of differences in substantive trait between groups. A simulation study shows that the IRTree model and MNRM can drastically differ in their conclusions when groups differ in their average extreme response style (ERS). An empirical example is provided and implications for the future use of both models and the conceptualization of ERS are discussed.
EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT
(2023)
Article
Mathematics, Interdisciplinary Applications
F. J. Clouth, S. Pauws, J. K. Vermunt
Summary: This article presents an extension of the bias-adjusted three-step latent class analysis with inverse propensity weighting (IPW) to account for differential item function (DIF) caused by treatment or exposure variables. The proposed method includes treatment with its direct effect on the class indicators in the step-one model and incorporates IPW in the step-three model to adjust for classification errors that differ across treatment groups. DIF caused by confounders used to create the propensity scores is found to be less problematic. The newly proposed approach is demonstrated using synthetic and real-life data examples and implemented in Latent GOLD program.
STRUCTURAL EQUATION MODELING-A MULTIDISCIPLINARY JOURNAL
(2023)
Article
Cell Biology
Hua-Rong Lu, Manabu Seo, Mohamed Kreir, Tetsuya Tanaka, Rie Yamoto, Cristina Altrocchi, Karel van Ammel, Fetene Tekle, Ly Pham, Xiang Yao, Ard Teisman, David J. Gallacher
Summary: Drug-induced seizure risk is a significant safety concern in drug development, leading to increased costs and delays. This study investigated the use of fluorescent dyes to measure changes in calcium oscillations in hiPSC-derived neurons co-cultured with primary astrocytes in 2D and 3D forms as a potential indicator of seizure risk. The results showed high accuracy in identifying drugs with seizure risk using this approach in 2D cultures compared to 3D cultures.
Article
Psychology, Multidisciplinary
Brenda De Wit-De Visser, Madeleine Rijckmans, Jeroen K. Vermunt, Arno van Dam
Summary: ASPD and ASB have significant impacts on individuals, their environment, and society, with no evidence-based treatments currently available. Contradictory research findings on therapy effectiveness and underlying factors of ASB further fuel the debate on the accuracy of the conceptualization of ASPD and the homogeneity of this population.
FRONTIERS IN PSYCHOLOGY
(2023)
Article
Pharmacology & Pharmacy
C. Altrocchi, K. Van Ammel, M. Steemans, M. Kreir, F. Tekle, A. Teisman, D. J. Gallacher, H. R. Lu
Summary: This study developed an in vitro screening platform using human-induced pluripotent stem cell-derived cardiomyocytes to evaluate both acute and delayed electrophysiological and cytotoxic effects of reference compounds, which can contribute to the early assessment of drug-induced cardiotoxicity.
FRONTIERS IN PHARMACOLOGY
(2023)
Article
Psychology, Multidisciplinary
Mihai A. A. Constantin, Noemi K. Schuurman, Jeroen K. K. Vermunt
Summary: We present a general method for sample size computations in the context of cross-sectional network models. The method is an automated Monte Carlo algorithm that iteratively concentrates computations on relevant sample sizes to find the optimal size. It requires inputs of a hypothesized network structure or desired characteristics, an estimation performance measure and target value, and a statistic and target value to reach the performance measure. The method includes a Monte Carlo simulation, curve-fitting, and stratified bootstrapping steps to provide sample size recommendations with uncertainty quantification. Rating: 7/10
PSYCHOLOGICAL METHODS
(2023)
Article
Psychology, Mathematical
Leonie V. D. E. Vogelsmeier, Jeroen K. Vermunt, Kim De Roover
Summary: Intensive longitudinal data (ILD) have gained popularity in studying within-person dynamics in psychological constructs. Before exploring the dynamics, it is important to examine whether the measurement model (MM) is consistent across subjects and time, in order to ensure the constructs have the same meaning. Latent Markov factor analysis (LMFA) can be used to investigate MM differences in ILD, and the new user-friendly software package lmfa makes it easier for researchers to perform this analysis.
BEHAVIOR RESEARCH METHODS
(2023)
Article
Psychology, Mathematical
Jonas M. B. Haslbeck, Jeroen K. Vermunt, Lourens J. Waldorp
Summary: Gaussian mixture models (GMMs) are widely used for exploring heterogeneity in multivariate continuous data, but their performance in estimating GMMs for ordinal data is uncertain. In this study, we investigate this by simulating data from various GMMs, thresholding them in ordinal categories, and evaluating recovery performance. We find that the number of components can be reliably estimated with enough ordinal categories and variables, but the estimates of component model parameters are biased regardless of sample size.
BEHAVIOR RESEARCH METHODS
(2023)
Article
Social Sciences, Mathematical Methods
Laura Boeschoten, Sander Scholtus, Jacco Daalmans, Jeroen K. Vermunt, Ton de Waal
SURVEY METHODOLOGY
(2022)
Article
Psychology, Mathematical
E. Damiano D'Urso, Kim De Roover, Jeroen K. Vermunt, Jesper Tijmstra
Summary: The study compared the performance of scale- and item-level approaches based on multiple group categorical confirmatory factor analysis and multiple group item response theory in testing measurement invariance with ordinal data. Results showed that, in general, MG-CCFA-based approaches outperformed MG-IRT-based approaches at the scale level. The best performing approach at the item level depends on the tested parameter, with likelihood ratio test providing the best trade-off when testing loadings equivalence and chi(2) test outperforming others when testing thresholds equivalence.
BEHAVIOR RESEARCH METHODS
(2022)