Dear Friends,
I have the following question on comparing calibration (using Hosmer-Lemeshow test) of two logistic regression models based on reported p-value.
Can we use p-values for comparing calibration for two different logistic regression models? If one model has p value .2 and another model has .8, is it justified to conclude that second model is better calibrated compare to the first model ? [My intuition says NO, because p- value only measure of consistency between the observed data and the null hypothesis. And, here the null hypothesis is that the model is well calibrated. p-value does not tell about how strong the calibration is.]
Thank you for your response.
Best regards,
Suman Kundu
PhD student.
Erasmus MC, Rotterdam
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|