Binary logistic regression's backward stepwise method with a
likelihood ratio test was applied on the reduced variables to identify
the predictors that will best explain WSSV occurrence. The backward
stepwise logistic regression eliminates at each consecutive step the
variable that contributes the least to explain the model until the
smallest possible log-likelihood ratio, indicator for the statistical fit of
the model, is attained. The software lists all steps and for each step it
gives indications for the model fit. The model giving the best
combination of low negative log-likelihood value, R2 value close to
one, and high classification ratio (CR), in predicting WSSV incidence
for each group was retained.