ASYMPTOTICALLY EFFICIENT SIMULTANEOUS ESTIMATION OF SEVERAL PARAMETERS.
COLUMBIA UNIV NEW YORK PSYCHOPHYSICS LAB
Pagination or Media Count:
In a chi - square test of goodness of fit of data to a theoretical model, the estimators of the parameters of the model must be asymptotically efficient. A method described by W. H. Kruskal of converting a consistent but inefficient estimator into one that is asymptotically efficient is extended to the case of simultaneous estimation of several parameters. The method is exhibited in a single formula. However, to render the formula intelligible and to motivate and prove the method necessitates derivation in detail of certain characteristics of RBAN estimators of parameters that underlie multinomial distributions. Author
- Statistics and Probability