Dataset Information

Correlation among baseline variables yields non-uniformity of p-values.

ABSTRACT: A recent paper in Neurology used statistical techniques to investigate the integrity of the randomization in 33 clinical trials conducted by a group of investigators. Without justification, the approach assumed that there would be no impact of correlation among baseline variables. We investigated the impact of correlation on the conclusions of the approach in several large-scale simulation studies that replicated the sample sizes and baseline variables of the clinical trials in question and utilized proper randomization. Additionally, we considered scenarios with larger numbers of baseline variables. We found that, with even moderate correlation, there can be substantial inflation of the type I error of statistical tests of randomization integrity. This is also the case under no correlation, in the presence of some discrete baseline variables, with a large number of variables. Thus, statistical techniques for assessing randomization integrity should be applied with extreme caution given that very low p-values, which are taken as evidence against valid randomization, can arise even in the case of valid randomization, in the presence of correlation. More generally, the use of tests of goodness of fit to uniformity for the purpose of testing a global null hypothesis is not advisable in the presence of correlation.

SUBMITTER: Betensky RA

PROVIDER: S-EPMC5590981 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BACKGROUND:In livestock, residual variance has been studied because of the interest to improve uniformity of production. Several studies have provided evidence that residual variance is partially under genetic control; however, few investigations have elucidated genes that control it. The aim of this study was to identify genomic regions associated with within-family residual variance of yearling weight (YW; N?=?423) in Nellore bulls with high density SNP data, using different response variables. For this, solutions from double hierarchical generalized linear models (DHGLM) were used to provide the response variables, as follows: a DGHLM assuming non-null genetic correlation between mean and residual variance (rmv???0) to obtain deregressed EBV for mean (dEBVm) and residual variance (dEBVv); and a DHGLM assuming rmv?=?0 to obtain two alternative response variables for residual variance, dEBVv_r0 and log-transformed variance of estimated residuals (ln_[Formula: see text]). RESULTS:The dEBVm and dEBVv were highly correlated, resulting in common regions associated with mean and residual variance of YW. However, higher effects on variance than the mean showed that these regions had effects on the variance beyond scale effects. More independent association results between mean and residual variance were obtained when null rmv was assumed. While 13 and 4 single nucleotide polymorphisms (SNPs) showed a strong association (Bayes Factor?>?20) with dEBVv and ln_[Formula: see text], respectively, only suggestive signals were found for dEBVv_r0. All overlapping 1-Mb windows among top 20 between dEBVm and dEBVv were previously associated with growth traits. The potential candidate genes for uniformity are involved in metabolism, stress, inflammatory and immune responses, mineralization, neuronal activity and bone formation. CONCLUSIONS:It is necessary to use a strategy like assuming null rmv to obtain genomic regions associated with uniformity that are not associated with the mean. Genes involved not only in metabolism, but also stress, inflammatory and immune responses, mineralization, neuronal activity and bone formation were the most promising biological candidates for uniformity of YW. Although no clear evidence of using a specific response variable was found, we recommend consider different response variables to study uniformity to increase evidence on candidate regions and biological mechanisms behind it.

Dataset Information

Correlation among baseline variables yields non-uniformity of p-values.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets