Cramer-von Mises variance estimators for simulations
David M. Goldsman, Keebom Kang, Andrew F. Seila · 2002
The authors study estimators for the variance parameter sigma /sup 2/ of a stationary process. The estimators are based on weighted Cramer-von Mises statistics formed from the standardized time series of the process. Certain weightings yield estimators which are first-order unbiased for sigma /sup 2/ and which have low variance. It is also shown how the Cramer-von Mises estimators are related to the standardized time series area estimator; this relationship is used to establish additional estimators for sigma /sup 2/.>