PROBABILITY CALCULATIONS ON A DIGITAL COMPUTER.
Technical rept. for Feb 64-Apr 65,
MEASUREMENT ANALYSIS CORP LOS ANGELES CALIF
Pagination or Media Count:
The purpose of this report is to discuss computer implementations of tests for normality. Some of the underlying statistical considerations and random process theory are briefly reviewed to provide a background for subsequent material. Methods for digitally computing basic statistical parameters, such as the mean variance, skewness, kurtosis, etc., are given, along with procedures for computing the sample probability density function. Problems arising from the digital and discrete nature of the data are discussed, as well as sample sizes required for for the tests. The central part of the report consists of a detailed discussion of the computer implementation of the chi-square goodness-of-fit test as applied to testing for normality, followed by computer program flow charts, suitable for use in coding the procedure for a digital computer. Sufficient detail has been provided so that a programmer unfamiliar with the material should be able to write a program which will make maximum use of the time which is frequently lost during periods of input and output of data from the computer. The report concludes with an appendix which lists standard approximations for some of the mathematical functions required in the calculations. Author