Can statistics cause brain damage?

Abstract

It is hoped that the comments made above will help contributors to the Journal of Cerebral Blood Flow and Metabolism identify weaknesses in their use of statistics. A little extra care in the presentation of results and adherence to the following guidelines should result in major improvement: (1) Give an explicit description of the type of statistical analysis used. (2) Increase the use of confidence intervals. (3) Use the Bonferroni correction (or other appropriate procedures) in multigroup comparisons. (4) Be aware of the assumptions underlying statistical methods and of weakness in experimental design. However, real benefits will be gained when statistics is viewed as being as important to the experiment as the choice of method for assessing cerebral blood flow itself.