Article
Neurosciences
Alexander Bowring, Fabian J. E. Telschow, Armin Schwartzman, Thomas E. Nichols
Summary: Current statistical inference methods for task-fMRI have limitations in solely focusing on detection of non-zero signal or signal change, and in only indicating regions where null hypothesis can be rejected without providing spatial uncertainty about activation. By developing spatial Confidence Sets (CSs), this work addresses these issues and allows determination of brain regions with reliable Cohen's d response under a specific threshold.
Article
Public, Environmental & Occupational Health
Mohammad Ali Mansournia, Maryam Nazemipour, Ashley Naimi, Gary S. Collins, Michael J. Campbell
Summary: Statistical estimates have uncertainty due to sampling variability, and robust standard errors can address violations of model assumptions, often used in clinical papers.
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
(2021)
Article
Automation & Control Systems
Ibrahim Merad, Stephane Gaiffas
Summary: In this paper, statistically robust and computationally efficient linear learning methods are proposed for high-dimensional batch settings. Two algorithms are employed based on the type of loss function. The framework is instantiated on several applications, showing efficient and robust learning algorithms that achieve near-optimal estimation rates under heavy-tailed distributions and outliers. Experimental results confirm the theoretical findings and provide a comparison with other recent approaches.
JOURNAL OF MACHINE LEARNING RESEARCH
(2023)
Article
Geochemistry & Geophysics
Bangjie Zhang, Gang Xu, Hanwen Yu, Hui Wang, Hao Pei, Wei Hong
Summary: This article proposes a novel robust gridless compressed sensing (RGLCS) algorithm for high-resolution 3-D imaging. The algorithm uses atomic norm minimization to model the joint-sparsity pattern on elevation distribution between adjacent pixels, and models outliers and disturbances as sparsely distributed spike noise in the image domain. Experimental results validate the effectiveness of the proposed algorithm.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Social Sciences, Mathematical Methods
Rand Wilcox
Summary: This paper discusses alternative methods of comparing groups using robust measures of effect size instead of measures of location, highlighting the deeper understanding of group comparisons that can be achieved through comparing effect sizes.
METHODOLOGY-EUROPEAN JOURNAL OF RESEARCH METHODS FOR THE BEHAVIORAL AND SOCIAL SCIENCES
(2022)
Article
Psychology, Multidisciplinary
Xiaofeng Steven Liu
Summary: Cohen's d has a positive bias and the traditional bias correction method based on strict distribution assumption may not always work for small studies with limited data. Non-parametric bootstrapping, which is not limited by distribution assumption, can be used to remove the bias in Cohen's d. A real example is provided to demonstrate the implementation of bootstrap bias estimation and the removal of sizable bias in Cohen's d.
JOURNAL OF GENERAL PSYCHOLOGY
(2023)
Article
Engineering, Multidisciplinary
A. Audu, O. O. Ishaq, R. V. K. Singh, A. Danbaba, F. Manu
Summary: This study aimed to improve the efficiency of Zaman estimators using exponential transformation technique and proposed a new class of estimators. The empirical study through simulations found that these proposed estimators were more efficient under different distributions and robust regression methods.
QUALITY AND RELIABILITY ENGINEERING INTERNATIONAL
(2023)
Article
Multidisciplinary Sciences
Vishal Midya, Jiangang Liao, Chris Gennings, Elena Colicino, Susan L. Teitelbaum, Robert O. Wright, Damaskini Valvi
Summary: The article discusses common issues in epidemiological research, introduces the new method delta-score and the concept of sufficient sample size, and demonstrates how to reduce the asymmetry between precision and utility through practical data.
Article
Economics
Benedikt M. Poetscher, David Preinerstorfer
Summary: We have developed theoretical finite-sample results on the size of wild bootstrap-based heteroskedasticity robust tests in linear regression models. These results provide an efficient diagnostic check to assess the reliability of a wide range of wild bootstrap-based tests.
ECONOMETRIC THEORY
(2022)
Article
Automation & Control Systems
Ayseguel Yabaci Tak, Ilker Ercan
Summary: This study proposes the use of meta fuzzy effect size functions (MFESF) to overcome the limitations of standard assumptions and evaluates its performance on different datasets. The results show that MFESF performs better than individual effect size methods on all datasets.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Physics, Multidisciplinary
Byungsoo Kim, Sangyeol Lee, Dongwon Kim
Summary: In this study, a robust estimation method for bivariate Poisson INGARCH models using the minimum density power divergence estimator was discussed, with demonstrations of consistency and asymptotic normality under certain regularity conditions. Monte Carlo simulations were used to evaluate the performance in the presence of outliers, and real data analysis and an artificial example were provided for illustration.
Article
Psychiatry
Sareh Panjeh, Anders Nordahl-Hansen, Hugo Cogo-Moreira
Summary: Objective-Cohen's d conventional effect size cutoffs might not accurately represent the distribution of effect sizes in different health areas. The cutoffs for effect size can vary depending on the research area, intervention type, and population. Therefore, we propose new cutoffs based on percentiles, which result in smaller values compared to Cohen's thresholds. Applying Cohen's effect size thresholds for improving sleep quality may overestimate the effect sizes in real-world contexts.
INTERNATIONAL JOURNAL OF METHODS IN PSYCHIATRIC RESEARCH
(2023)
Article
Statistics & Probability
Yahia S. S. El-Horbaty, Eman M. M. Hanafy
Summary: Permutation methods are commonly used for inferring zero variance components in linear mixed models, but they may not be optimal when data has heavy-tailed, heavy-skewed distributions, or outliers. This article proposes the use of robust rank-based estimation as an alternative, which approximates the finite sample distribution of the test statistic by permuting cluster indices. Empirical results show that the new test maintains acceptable Type I error rates for heavy-tailed and heavy-skewed data, while being robust against outliers. It also demonstrates superior power compared to existing tests in most cases.
STATISTICAL PAPERS
(2023)
Article
Multidisciplinary Sciences
Josefina Weinerova, Denes Szucs, John P. A. Ioannidis
Summary: The study evaluates manually collected and computer-extracted correlation effect sizes, presents statistical results of different samples, and suggests that large sample sizes may lead to reporting smaller correlations.
ROYAL SOCIETY OPEN SCIENCE
(2022)
Article
Engineering, Electrical & Electronic
Guangxiao Zhang, Xiaoyang Tong, Qiteng Hong, Campbell D. Booth
Summary: This paper proposes a robust pilot protection scheme based on waveform similarity. The Kendall's Tau coefficient (KTC) algorithm is used to distinguish different datasets of a protected transmission line. The proposed scheme achieves reliable fault identification under various fault conditions.
IEEE TRANSACTIONS ON POWER DELIVERY
(2022)