Thomas, et al.[ 10] showed that in the two-group microarray study setting the t-statistic assumption of equal variances between the two groups could result in poor performance.
Additionally, because permutations are used to evaluate the significance level of testing statistic, our PLS-based MLAS do not depend on specific statistical assumption, for instance the normality assumption of target traits.
Logistic regression models were used to assess for interactions, to adjust for confounding, and to test for statistical assumptions using the Hosmer-Lemeshow goodness-of-fit statistic.
To pool several studies and estimate a summary statistic some assumptions are made.
Sample size was determined on the basis of the primary objective with a t statistic under the assumption of a one-sided t test of size 2.5%, a zero mean treatment difference, and a 1.3% SD for HbA1c.
Sample size was determined on basis of the primary objective with a t statistic under the assumption of a one-sided t test of size 2.5%, a zero mean treatment difference, and a 1.3% SD for A1C.
We will show in this paper that, given an appropriate statistic and distributional assumption, the hypothesis in the test of AUC can be equivalent to that in the LR or Wald tests and, for the particular statistic for the test of AUC we consider, the test of AUC actually can perform better than those statistical significance tests at small sample sizes.
Comparisons between continuous variables among multiple groups were performed by analysis of variance using the Brown Forsythe statistic when the assumption of equal variances did not hold while the proportions were compared by means of the χ test, using Fisher's exact test when necessary.
The SAM procedure described by Tusher et al. [ 3] based on the t-statistic (under the assumption of equal variances between the two groups) and the Welch statistic (under the assumption of unequal variances between the two groups) was modified as follows: 1.
The KS statistic makes no assumptions about the distribution of the data and is independent of scale changes.
The bootstrap is non parametric in the sense that it evaluates the performance of any test statistic without making assumptions about the form of the distribution.
