|Multiple Probit Analysis with a Nonzero Background.
Hasselblad, Victor ;
Stead, Andrew G. ;
Creason, John P. ;
|Health Effects Research Lab., Research Triangle Park, NC.
Probability theory ;
Probit analysis ;
Maximum likelihood estimation
|Some EPA libraries have a fiche copy filed under the call number shown.
The 'EM' (Expectation-Maximization) algorithm is applied to probit analysis with multiple independent variables and a nonzero response rate. The equations for the maximum likelihood estimators are relatively simple, and converge in all the cases so far examined. An animal bioassay example is used to demonstrate the method. Likelihood ratio tests are used for significance tests in the analysis.