A statistical test criterion in comparing two samples of data believed to be normally distributed.