The range statistic and procedures to select the best systems

This paper investigates the hypothesis testing of equivalence of mean of multiple systems using the range statistic. The procedure allows unequal variances among systems. Equivalence testing has many applications in screening for new product development. First, the required sample sizes for equivalence testing at a specified confidence are derived. We then develop an indifference‐zone procedure to select only the best systems. For a specified confidence level P*, the selected subset is guaranteed to contain all the best systems and none of the systems that deviate more than a specified amount from the best systems. A table of critical constants is provided for implementing the procedure. An experimental performance evaluation demonstrates the validity and efficiency of the procedure.