Effect of the unreliability of measurement on the difference between groups.

Abstract

The ordinary formula, [diff.]/[standard deviation of diff.] (I), which denotes the significance of a difference between the means of 2 groups, is not lessened by presence of errors of measurement. On the contrary, a formula I t is presented, which is a correction of I for attenuation and indicates that an obtained difference is theoretically more significant the grosser the chance errors.