China
Africa
Explore chapters and articles related to this topic
Introduction and Motivation
Published in Corain Livio, Arboretti Rosa, Bonnini Stefano, Ranking of Multivariate Populations, 2017
Corain Livio, Arboretti Rosa, Bonnini Stefano
Multiple Comparison Procedures (MCPs) have been proposed just to determine which populations differ after obtaining a significant omnibus test result, such as the ANOVA F-test. However, when MCPs are applied with the goal of ranking populations they are at best indirect and less efficient, because they lack protection in terms of a guaranteed probability against picking out the ‘worse’ population. This drawback motivated the foundation of the so-called ranking and selection methods (Gupta and Panchapakesan, 2002), the formulations of which provide more realistic goals with respect to the need to rank or select the ‘best’ populations. A further class of procedures with some connection with the ranking problem are the constrained - or order restricted - inference methods (Robertson et al., 1988; Silvapulle and Sen, 2005). Finally, the ranking problem has been addressed in the literature from the point of view of investigating and modelling the variability of sampling statistics used to rank populations, that is, the empirical estimators whose rank transformation provides the estimated ranking of the populations of interest (Hall and Miller, 2009, 2010; Hall and Schimek, 2012).
A survey of four Scottish proposed wind farms
Published in Giuseppe Pellegrini-Masini, Wind Power and Public Engagement, 2020
The “omnibus test of model coefficients” tells if the new model has the –2 Log likelihoods statistic significantly reduced compared to the baseline model, which would mean that the new model is capable of explaining more of the variance in the outcome. As can be seen (Table 5.81), in this case the chi-square value is significant at the 0.0005 level.
Perception of rhythmic agency for conversational labeling
Published in Human–Computer Interaction, 2023
Christine Guo Yu, Alan F. Blackwell, Ian Cross
Experiments 1a and 1b both compare measures in the four conditions {CA, CR, UR, UC}. In the following analyses, we first report an omnibus test to determine whether there is any statistically significant difference between the four conditions. If a significant difference is found, we then carry out contrast analysis (Haans, 2018; Rosenthal et al., 1985) to test whether the ordering of the measured values in the four conditions supports hypotheses in relation to sense of agency. Bonferroni adjustment is applied to these significance tests (i.e., where k tests are carried out, α = 0.05/k). Not all measures follow a normal distribution – we report this for each measure (based on Shapiro-Wilk test for normality), and use a non-parametric test whenever the data is not normally distributed.
Identification of the behavioural factors in the decision-making processes of the energy efficiency renovations: Dutch homeowners
Published in Building Research & Information, 2022
Shima Ebrahimigharehbaghi, Queena K. Qian, Gerdien de Vries, Henk J. Visscher
Binary logistic regression model is used to describe the relation between the dependent and independent variables: where P is the probability of events, and X represents independent variables. After estimation, the omnibus tests of the model coefficients and the Hosmer and Lemeshow test were applied to validate the models (Table 8). The omnibus test checks whether the model estimates the outcome with the explanatory variables better than without (Brant, 1990). The omnibus tests were statistically significant, and the models were better with explanatory variables than without. The Hosmer and Lemeshow test illustrated the goodness of fit, which is an insignificant factor for a good model.
Exploring the potential utility of a wearable accelerometer for estimating impact forces in ballet dancers
Published in Journal of Sports Sciences, 2020
Thomas Gus Almonroeder, Lauren Benson, Alexandra Madigan, Drake Everson, Cameron Buzzard, Madison Cook, Brian Henriksen
Spearman’s rho correlations were performed to quantify the strength of the relationship between the impact accelerations and the peak vertical GRFs and loading rates. Separate correlation analyses were performed for each time point. The strength of the correlation coefficients (r) were assessed using the following criteria: 0.10 to 0.29, weak relationship; 0.30 to 0.49, moderate relationship; 0.50 to 0.69 strong relationship; > 0.70 very strong relationship (Hopkins, Marshall, Batterham, & Hanin, 2009). The statistical significance of the correlation coefficients was also evaluated. We used the non-parametric Spearman’s rho correlation approach because not all of the distributions satisfied the assumption of normality required for parametric testing. To explore the influence of fatigue, repeated measures ANOVA tests were performed to assess the impact accelerations, vertical GRFs, and loading rates across the time points (baseline, 25%, 50%, 75%, 100%). Fisher’s least significant difference post-hoc tests were conducted in the case of a significant omnibus test. Partial eta squared (η2) effect sizes were reported for the ANOVA tests. Partial eta squared values of 0.01, 0.06, and 0.14 were considered small, medium, and large effect sizes (Cohen, 1988). In addition, we also compared the changes in the impact accelerations to the minimal detectable change (MDC) associated with the measure. The MDC reflects the smallest difference that exceeds potential measurement error; for this study “error” likely reflects natural variability among landings. The consistency of the impact accelerations for the baseline landings was captured using Cronbach’s alpha (α). This coefficient, along with the standard deviation for the baseline trials (SDbaseline), was used to calculate the standard error of the measurement (SEM) using Equation (1). The MDC was calculated for a 95% confidence interval using Equation (2).