Binary Regression with Unreplicated Data
- 1 December 1976
- journal article
- research article
- Published by JSTOR in Biometrics
- Vol. 32 (4) , 935-938
- https://doi.org/10.2307/2529276
Abstract
The results of a biological simulation study comparing the method of maximum likelihood for binary regression with unreplicated data and 2 approximate methods are presented and discussed. The 2 approximate methods are that of Cox and unweighted least squares with the 0''s replaced by -3 and the 1''s by +3. With 20 or more data values the maximum likelihood methods worked quite well but with fewer than 20 the approximate methods are to be preferred.This publication has 1 reference indexed in Scilit:
- A General Maximum Likelihood DiscriminantPublished by JSTOR ,1967