Tests for Multiple Outliers in an Exponential Sample
In this article we apply the prediction-based statistic of Balasooriya (1989) to test for multiple upper outliers in an exponential sample. The prediction-based statistics are related to the (randomly) normalized spacing statistics of Sweeting (1983); and are thus mutually independent. Rosner's (1975) inside-out sequential procedure for testing up to k outliers is adopted. For an assigned overall significance level the necessary critical values at each stage can be easily obtained. An example is provided to illustrate the use of the proposed test. Results of two Monte Carlo experiments for comparing the power of the prediction-based test and Kimber's (1982) test are also reported. The experiments are designed to mimic the unconditional slippage model of Karlin and Truax (1960) and a conditional slippage model in which the outliers are the extreme order statistics
TSE, Yiu Kuen and Balasooriya, U..
Tests for Multiple Outliers in an Exponential Sample. (1991). Sankhya B. 53, (1), 56-63. Research Collection School Of Economics.
Available at: http://ink.library.smu.edu.sg/soe_research/331