Abstract
This paper deals with the detection procedure of outliers. The difference between the order statistics i.e., Irwin type test statistics, is employed. A new outlier detection procedure which avoids the danger of masking and swamping is proposed and the percentage points are given as α=0.01,0.05,0.10,n=3(1)20. The power of the proposed procedure and the sequential tests by Grubbs or Dixon with several outliers are examined through the Monte Carlo experiments. The results of this study show that the proposed procedure have a high power for the case where many outliers are present.
