1.3
The Cornu Criterion - A Valid Test for Normality of Small Sample Sizes
PAPER WITHDRAWN
Ron Lowther, Air Force Combat Climatology Center, Asheville, NC; and J. F. Griffiths and J. F. Griffiths
This paper revitalizes the Cornu Criterion as a simple, valid test for normality when working with small sample sizes. It is based on the relationship between the mean deviation to the standard deviation within a purely normal distribution. Statistics require that certain conditions are met by data before any statistical tests on the data are considered valid. One must always know the distribution of the data beforehand in order to determine if these conditions are met. Many meteorological variables are not normally distributed by nature, however, most existing tests for normality are difficult to use and not suitable for small sample sizes. Although the Cornu Criterion was only originally computed for a few selected sample sizes, simple linear interpolation techniques are used to enhance its value as a valid test for small samples of climate data.
Session 1, Data display and analysis
Tuesday, 9 May 2000, 10:50 AM-11:50 AM
Previous paper Next paper