Search

Publication bias undermines the results of meta‐analyses and other systematic reviews. Applying a formal test for publication bias is therefore a part of many protocols for conducting a meta‐analysis. To be effective in this role, the tests must have good power to detect the presence of publication bias. We derive the local limiting power functions and study the Pitman relative efficiency for two popular tests, Begg's and Egger's tests. The analysis suggests that the differences between the two arise out of built‐in robustness properties of Begg's test. These work to its detriment in the commonly used Gaussian model, but leave it with better efficiency when study effects are heavy‐tailed or skew. [ABSTRACT FROM AUTHOR]

We consider a recurrent event wherein the inter-event times are independent and identically distributed with a common absolutely continuous distribution function F. In this article, interest is in the problem of testing the null hypothesis that F belongs to some parametric family where the q-dimensional parameter is unknown. We propose a general Chi-squared test in which cell boundaries are data dependent. An estimator of the parameter obtained by minimizing a quadratic form resulting from a properly scaled vector of differences between Observed and Expected frequencies is used to construct the test. This estimator is known as the minimum chi-square estimator. Large sample properties of the proposed test statistic are established using empirical processes tools. A simulation study is conducted to assess the performance of the test under parameter misspecification, and our procedures are applied to a fleet of Boeing 720 jet planes' air conditioning system failures. [ABSTRACT FROM AUTHOR]

This article introduces the exchangeable Kendall's tau as a nonparametric intraclass association measure in a clustered data frame and studies the sampling properties of an estimator of this quantity. Its asymptotic distribution is derived under a multivariate exchangeable model. Two applications of the proposed statistic are considered. The first is an estimator of the intraclass correlation coefficient for data drawn from an elliptical distribution. The asymptotic properties of this estimator are investigated under an extended random effects ANOVA model. The second is a semiparametric intraclass independence test based on the exchangeable Kendall's tau whose performance is assessed using Pitman efficiencies and a Monte Carlo study. The Canadian Journal of Statistics 42: 384-403; 2014 © 2014 Statistical Society of Canada [ABSTRACT FROM AUTHOR]

In missing data problems, it is often the case that there is a natural test statistic for testing a statistical hypothesis had all the data been observed. A fuzzy p-value approach to hypothesis testing has recently been proposed which is implemented by imputing the missing values in the “complete data” test statistic by values simulated from the conditional null distribution given the observed data. We argue that imputing data in this way will inevitably lead to loss in power. For the case of scalar parameter, we show that the asymptotic efficiency of the score test based on the imputed “complete data” relative to the score test based on the observed data is given by the ratio of the observed data information to the complete data information. Three examples involving probit regression, normal random effects model, and unidentified paired data are used for illustration. For testing linkage disequilibrium based on pooled genotype data, simulation results show that the imputed Neyman Pearson and Fisher exact tests are less powerful than a Wald-type test based on the observed data maximum likelihood estimator. In conclusion, we caution against the routine use of the fuzzy p-value approach in latent variable or missing data problems and suggest some viable alternatives. [ABSTRACT FROM AUTHOR]

Blest (2000, Aust. N. Z. J. Stat. 42, 101–111) proposed a new measure of rank correlation that is sensitive to discrepancies in the small ranks. This paper investigates the efficiency properties of non-parametric tests for independence based on Blest's correlation coefficient and its modifications. Pitman efficiency comparisons are made with analogous tests existing in the literature. Conditions for Pitman optimality of the Blest-type tests are established. [ABSTRACT FROM AUTHOR]

This paper develops a non-parametric multi-sample inference for simple-tree alternatives for ranked-set samples. The multi-sample inference provides simultaneous one-sample sign confidence intervals for the population medians. The decision rule compares these intervals to achieve the desired type I error. For the specified upper bounds on the experiment-wise error rates, corresponding individual confidence coefficients are presented. It is shown that the testing procedure is distribution-free.To achieve the desired confidence coefficients for multi-sample inference, a nonparametric confidence interval is constructed by interpolating the adjacent order statistics. Interpolation coefficients and coverage probabilities are provided, along with the nominal levels. [ABSTRACT FROM AUTHOR]