Date of this Version
Journal of the American Statistical Association, Vol. 45, No. 250 (Jun., 1950), pp. 257-260
One frequently encounters the need of a rational selection of sample size when it is desired to estimate the standard deviation. Changing the permissible error of the estimate from an absolute to a relative one is acceptable in many cases and permits an exact, a priori solution to the problem of sample size without involving any previous estimates.