China 
Africa 
Explore chapters and articles related to this topic
Let’s Find Out
Published in S. Kanimozhi Suguna, M. Dhivya, Sara Paiva, Artificial Intelligence (AI), 2021
Jayden Khakurel, Indu Manimaran, Jari Porras
Quantitative data collected from the two sessions were analyzed using the statistical data analysis language R and the descriptive statistical analysis functions available in R core (R Core Team 2017) and the psych library (Revelle 2017). We first used the Mann–Whitney U test (Wohlin et al. 2012) to analyze the difference in distributions between the data sets. A continuity correction was enabled to compensate for non-continuous variables (Bergmann and Ludbrook 2000). The Bonferroni correction was used to adjust the p-value to compensate for the family-wise error rate in multiple comparisons (Abdi 2007). We calculated the effect size r using the guidelines by Tofan et al. (2016) for the Mann–Whitney U test. We evaluated the effect size as proposed by Cohen (1994): in r, a large effect is 0.5, a medium effect is 0.3, and a small effect is 0.1.
Applied Univariate Statistics
Published in Nick Zacharov, Sensory Evaluation of Sound, 2018
Per Bruun Brockhoff, Federica Belmonte
The Bonferroni technique is, due to its simplicity and broad applicability, worth recapping, as discussed in more detail in Brockhoff et al. (2015a). The Bonferroni-correction amounts to using α/M as the level in each individual test or CI instead of α, when M multiple tests are carried out. Imagine that we performed an ANOVA in a situation with k = 15 groups. And then we do all the M = 15 · 14/2 = 105 possible pairwise hypothesis tests. Assume for a moment that the overall null hypothesis is true, that is, there really are no mean differences between any of the 15 groups. Now consider what would happen if we still performed all the 105 tests with α = 0.05! How many significant results would we expect among the 105 hypothesis tests? The answer is that we expect α · 105 = 0.05 · 105 = 5.25, that is, approximately 5 significant tests are expected. And what would the probability be of getting at least one significant test out of the 105? The answer to this question can be found using the binomial distribution: () P(At least one significant result out of 105 independent tests)=1−0.95105=0.9954
The Need of External Validation for Metabolomics Predictive Models
Published in Raquel Cumeras, Xavier Correig, Volatile organic compound analysis in biomedical diagnosis applications, 2018
Raquel Rodríguez-Pérez, Marta Padilla, Santiago Marco
Applying the same statistical test to each of the metabolites in a data set would result in a high number of false positives (also known as false discoveries or type I errors). Therefore, to reduce the probability of such error type, methods for multiple testing have been developed. These methods apply hypothesis testing considering the whole set (or a subset) of the measured metabolites, i.e., the whole set of tests or hypotheses. For this, several strategies have been followed, such as controlling the familywise error rate (FWER) or the false discovery rate (FDR) (Benjamini and Hochberg, 1995). FWER is the probability of having at least one false positive among all the tests, whereas FDR is the expected proportion of false positives among all the significant tests. The latter has been especially conceived for data sets with large number of tests and small sample size, as it is the case in genomics. Bonferroni correction (Bland and Altman, 1995) is a method that follows the FWER strategy, while the p-value step-up method proposed by Benjamini and Hochberg (Benjamini and Hochberg, 1995) controls FDR. Moreover, other quantities derived from the mentioned strategies, such as the q-value (Storey, 2003) or the local FDR (LFDR) (Efron et al., 2001), can be used analogously to p-value for multiple comparisons, even for data sets with small number of tests (Bickel, 2013; Padilla and Bickel, 2012).
Effects of tumor necrosis factor (TNF) gene polymorphisms on the association between smoking and lung function among workers in swine operations
Published in Journal of Toxicology and Environmental Health, Part A, 2021
Zhiwei Gao, James A. Dosman, Donna C. Rennie, David A. Schwartz, Ivana V. Yang, Jeremy Beach, Ambikaipakan Senthilselvan
In order to examine modification effects of the polymorphisms in the TNF gene on the association between smoking status and lung function, an interaction term between smoking status and a SNP was introduced into a multiple regression model after controlling for potential confounders including age, gender, height, weight, and atopy. Respiratory symptoms were not analyzed in this study. In order to control false positives due to multiple comparisons, the Bonferroni correction was applied in the statistical analysis p < .017(=0.05/3). In addition, the modification effects of several polymorphisms in the TLR2 and NOS3 genes on the correlation between smoking and lung function were examined since these polymorphisms were significantly associated with lung function among workers in swine operations in our previous studies (Gao et al. 2013, 2014) All statistical analyses were carried out using SAS software, copyright © 2002–2012 SAS Institute Inc., Cary, NC, USA.
Myoelectric manifestation of muscle fatigue in repetitive work detected by means of miniaturized sEMG sensors
Published in International Journal of Occupational Safety and Ergonomics, 2018
Alberto Ranavolo, Giorgia Chini, Alessio Silvetti, Silvia Mari, Mariano Serrao, Francesco Draicchio
Statistical analysis was performed using PASW version 17 (PASW Statistic, formerly SPSS, USA). The Shapiro–Wilk test was applied to verify the null hypothesis that the acquired sample (in relation to the parameters calculated) came from a normally distributed population. A two-way analysis of variance (repeated-measures ANOVA) was performed to investigate the effects of the factors as well as their interactions on variables. We considered the risk class (RC; four levels: AC, VL, L, AV) and sessions (three levels: pre, post, post30) as within-subject factors and gender (two levels: male, female) as a between-subjects factor. Parametric paired t tests were performed to detect any significant differences between conditions. p < 0.05 was considered statistically significant. Bonferroni correction was used when multiple comparisons were performed.
Scaling-up process characterization
Published in Quality Engineering, 2018
Obviously with many tests being calculated and sorted, the p-value is no longer meaningful, because it suffers from selection bias. The selection bias needs to be corrected. There is a rich literature of multiple-test adjustments. The Bonferroni correction is appealing for its simplicity, dividing the significance criterion for the p-value by the number of tests, but if there are thousands or millions of tests, the Bonferroni correction is too strict, and nothing may qualify as significant. Rather than control for experimentwise error rate, it is more practical to only control for the false-discovery rate (FDR), the expected rate of declaring something significant when the actual difference is zero. The FDR adjustment has become very widespread in genomics, where there may be hundreds of thousands or millions of gene expressions to screen. The classic FDR adjustment is the Benjamini-Hochberg adjustment (Benjamini and Hochberg 1995), which involves ranking the p-values and adjusting from the largest to the smallest, such that the adjustment for the largest is equivalent to the unadjusted p-value, and the adjustment for the smallest is equivalent to the Bonferroni p-value.