Fortunately, both the probit and logit models are not only the most popular statistical methods developed to analyse dichotomous responses of dependent variables in regression models but also yield more efficient and consistence parameter estimations (Demaris 1992; Long 1997).
The statistical analyses used binomial (logit link function), Poisson (log link function), and gamma (log link function) error terms for the outcome of encounters (oviposition vs. rejection), frequencies of behaviours, and handling times, respectively.
The Wald statistic was used to indicate the statistical significance of each fitted logit coefficient (different from zero) corresponding to each independent variable.
The ability parameters (EAP/PV-estimators) range from −.952 to.900 logits for the conceptual dimension and from −1.664 to 1.547 logits for the statistical dimension.
The statistical theory of log-linear models and logit analysis for the evaluation of multi-way contingency tables was used [ 25], including the theory of graphical models [ 26].
Both the probit and logit links were assessed.
Our method is elegant, easy to implement, and produces a unified Gibbs sampler framework useful for both the logit and the probit link.
Implementation of the logit-t method in the statistical language R [40] applying two-way ANOVA instead of t-test, was used to rank genes according to their statistical significances of differential expression.
To test for the statistical significance of unconditional probabilities, we used logit-loglinear models [ 35, 36].
Finally, to ensure that the most appropriate model was used, we repeated the statistical analysis for the main effects model using a random-effects logit model (Model 6), a fixed-effects logit model (Model 7), and a random-effects probit model (Model 8).
