Does the intercept p value matter?

#1
Hi everyone,

I work on a model with logistic regression
as a result I have intercept value = 0.06 , with a p-value = 0.86.

Does this model seem good? Does the fact that the intercept has a p value > 0.05 mean the model is not significant?

thank you.
 
#2
No, the p-value for the intercept, or the intercept estimate itself, does not matter.

But the p-value for the parameter for the explanatory variable matters. And the corresponding parameter matters even more.
 

noetsi

Fortran must die
#4
You should look at the global measures of the model to determine if the model is significant. Not test of the slope. Usually these involve chi square test of some type. The Hosmer-Lemeshow test of model fit is also commonly used to assess model value.
 

hlsmith

Omega Contributor
#5
Side note is that the intercept does mean something, but not likely what you are looking for. if X1 and X2 were binary variables, the intercept could be used to calculate the predicted probability of the outcome when both X1 and X2 were set to their reference value. So regardless of your predictors the predicted probability of the outcome in your sample.