China
Africa
Explore chapters and articles related to this topic
Determinants of health complaints of Bodetabek commuter workers using Bayesian multilevel logistic regression
Published in Yuli Rahmawati, Peter Charles Taylor, Empowering Science and Mathematics for Global Competitiveness, 2019
We can calculate DIC by adding the average of the deviance from all iterations (D¯) with the effective number of parameters (pD). The smaller the DIC of a model, the more fit the model has. This study also calculates the Intraclass Correlation Coefficient (ICC) from the two-level null model. The ICC indicates the proportion of variance explained by the existence of a hierarchical structure. Many researchers use the uninformative prior, because this prior will not affect the posterior distribution formed (Hox, 2010). In this study, we use 5 percent significance level. When the p-value is less than 0.05, it is considered statistically significant.
Detection of Changes
Published in Zbigniew W. Kundzewicz, Changes in Flood Risk in Europe, 2019
Sheng Yue, Zbigniew W. Kundzewicz, Linghui Wang
In statistics, a result is called statistically significant if it is unlikely to have occurred by chance. The amount of evidence required to accept that an event is unlikely to have arisen by chance is known as the significance level, i.e. the measure of the probability of rejecting H0 when it is true. The significance level can also be called the Type I error. In contrast, the Type II error is the probability of accepting the null hypothesis when it is false. The power of a test is the probability of correctly rejecting the null hypothesis when it is false. Table 2 indicates the relationship between significance level, Type I and Type II errors and the power of a test. Yue et al. (2002a) and Yue & Pilon (2004) indicated that the power of a test for detecting trend depends on the pre-assigned significance level, magnitude of trend, sample size, distribution type, variation and skewness of the tested time series.
Breathomics and its Application for Disease Diagnosis: A Review of Analytical Techniques and Approaches
Published in Raquel Cumeras, Xavier Correig, Volatile organic compound analysis in biomedical diagnosis applications, 2018
David J. Beale, Oliver A. H. Jones, Avinash V. Karpe, Ding Y. Oh, Iain R. White, Konstantinos A. Kouremenos, Enzo A. Palombo
However, a major pitfall in metabolomics-based research, as is the case in biological research in general, is the inability to reproduce results of published studies. This was discussed in detail in two well-publicized works by Ionnidis (2005) and Buttn et al. (2013), and it concluded that poor reproducibility occurred when sample power was low (i.e., experiments that are designed with a small sample size) and an observed smaller effect size between the compared groups were overvalued (i.e., the statistical significance that defines the difference between groups is close to the 0.05 p-value cut-off). Thus, in order to obtain statistically significant results that are reproducible, studies should ensure that a sample size is sufficiently large to reduce the variation between subjects within a group and to focus on findings that statistically show a substantial difference between two or more compared groups.
The Role of Platform Brand in the Association Between Social Media Use, Stress and Educational Attainment
Published in International Journal of Human–Computer Interaction, 2023
Marc S. Tibber, Minglei Wang, Chan Zhang
Some have suggested that such documented associations, whilst statistically significant, are too small to be of practical or clinical significance (Orben & Przybylski, 2019). Nonetheless, it is possible that more pronounced associations exist between SM engagement and wellbeing/mental health, but that these are dependent on moderating variables that are not consistently modelled across studies. For example, there is a relative dearth of research exploring the role of inter-individual differences in risks and resiliencies, including the potential role of even basic socioeconomic and demographic variables such as age, gender, ethnicity and socioeconomic status (Orben, 2020b). Critically for this study, the majority of research in the field to date has also drawn upon single-platform data (39% according to one scoping review), or else cross-platform data (43%), neither of which facilitate identification of potentially differential effects of platform brand (Schønning et al., 2020).
Influences on U.S. undergraduate engineering students’ perceptions of ethics and social responsibility: findings from a longitudinal study
Published in Australasian Journal of Engineering Education, 2022
Shiloh James Howland, Stephanie Claussen, Brent K. Jesiek, Carla B. Zoltowski
Our results reflect interesting contrasts when comparing groups of participants and non-participants in various activities over time. Performance on the Fundamentals of Engineering/Situational Judgement measure (FESJ) among honours program participants did not change, as they already scored relatively high, but non-participants’ scores improved over time. Conversely, fraternity/sorority participants maintained their relatively lower FESJ scores, but non-participants improved over time. Fraternity/sorority participants also showed a decrease in Justice for Self subscale scores, while non-participants saw no changes over time. Finally, students who participated in service-learning decreased their scores on the Moral Disengagement measure but non-participants’ scores were unchanged over time. However, in all cases, the effect sizes were small, indicating the reported changes may be statistically significant but not practically significant. We saw no interaction effects for the Political and Social Involvement Scale (PSIS) or the other ten experiences.
Multi-objective mobile robot path planning problem through learnable evolution model
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2019
The simulation for each scenario is repeated 31 times and the statistically significant differences (SSD) of the obtained results are determined through the statistical analysis. In principle, a statistically significant result refers to a result that is not attributed to chance. It is noted that, in the experiments, here the confidence level in the statistical test is 95% (p-value less than 0.05) which means the differences are unlikely to have happened by chance with a probability of 95%. For a detailed explanation of the tests, refer to (Sheskin, 2011).