KOLMOGOROV-TYPE TESTS FOR EXPONENTIALITY WHEN THE SCALE PARAMETER IS UNKNOWN
STANFORD UNIV CA DEPT OF STATISTICS
Pagination or Media Count:
The paper shows how four statistics Kolmogorov-Smirnov, Cramer-von Mises, and the Kuiper and Watson extensions may be used to test whether a given sample comes from the exponential distribution with unknown parameter. Simple modifications of the basic definitions make it possible to use each statistic with only one line of percentage points in turn, these may be reduced to chi- square points. The tests are more powerful than the usual Pearson chi-square test, and are very well adapted for use with a computer.
- Statistics and Probability