Published on July 1, 2012 | http://doi.org/10.5867/medwave.2012.06.5437
Random error in clinical trials (II)
Rol del azar en los estudios clínicos II
Abstract
Random error in clinical trials can induce several types of flaws. One of them is to show a possible association that does not really exist when too many variables are linked (Type I error). Another type of failure is the opposite: the correlation does actually exist, but the study does not account for it because of insufficient sample size (Type II error). Another common mistake occurs in the unit of analysis in cluster randomized trials, when analyzing data from individual subjects as if they were independent.