Skip to main content

Table 7 Fitting nested deviance models. Fitting nested models to the data in order to get deviance scores. The difference in deviance between models is a better indicator of the significance of the associated effect (β1) when the logistic regression fits are near the boundary of the space, giving proportions close to zero.

From: Overdispersed logistic regression for SAGE: Modelling multiple groups and covariates

Model 1: Full Model Deviance = 5.0742  
Coefficients Estimate (s.e) t-value p-value
β 0 -11.494 13.518 -0.06 0.9519
β 1 5.987 13.524 -0.03 0.9750
Model 2: Null Model Deviance = 8.7541  
Coefficients Estimate (s.e) t-value p-value
β 0 -5.794 0.392 -14.772 6.05e-06