Likelihood Analyses and Tests for Binary Data

Summary A number of examples involving binary responses with factorial and response surface designs are analysed to provide a comparison of three types of statistical inference: ordinary likelihood methods, conditional likelihood methods and exact tests of significance. The derivations of the statistical models and the procedures for applying the three inference methods are described. The numerical examples indicate that the likelihood and conditional likelihood functions may not differ greatly when both parameters of interest and nuisance parameters are of the same type (location parameters).