A monte carlo comparison of several methods for analyzing small sample binomial data in a two-factor experiment without replication