Abstract
Goodman's simplified runs test is examined as a test for serial correlation and is found to be both much quicker to apply than other available tests and to be appropriate for use with non-stationary data provided that the series has no trend in mean. Approximations to the power of the test are discussed and it is found that the power is acceptable for samples of size 100 or more. Particular consideration is given to the application of the test to economic data.

This publication has 0 references indexed in Scilit: