MONTE CARLO INVESTIGATION OF THE PROBABILITY DISTRIBUTIONS OF DIXON'S CRITERIA FOR TESTING OUTLYING OBSERVATIONS.
BALLISTIC RESEARCH LABS ABERDEEN PROVING GROUND MD
Pagination or Media Count:
An empirical or Monte Carlo method for determining the distribution of Dixon-type sample statistics for testing outlying observations is presented. Results are tabulated for samples generated from a normal distribution and for samples generated from a uniform distribution. The method employed was to select random samples of sizes n 5, 10, 15 and 20 from each of the aforementioned distributions. After ordering the sample values such that X sub 1 X sub 2 . . . . X sub n, six different statistics for each sample size for each distribution were computed. A sampling distribution was therefore obtained empirically for each sample size for each distribution after 500 such sample trials. Finally, the upper alpha probability levels or percentage points are tabulated for each sample size for each distribution. Author
- Operations Research