A procedure to facilitate testing of a two-sided composite null hypothesis about the mean of a normally distributed random variable

Abstract

A procedure was developed to aid in the testing of a two-sided composite null hypothesis about the mean of a normally distributed random variable for situations where either the population variance is known or unknown. The procedure was designed to eliminate the requirement for iterative type solution techniques normally used in determining the acceptance or rejection region of the subject hypothesis. This thesis provides guidelines, curves, and tables which will aid in testing a two-sided corporate null hypothesis. Provisions were also incorporated into the procedure to permit testing of hypotheses about the relative displacement of the coefficient of variation from zero.