proc logistic 因变脸难道不应该是0 或者1吗？你这里好像是介于0和1之间？

In your data for the model, Number of unique profiles = 5. Variable Alcohol_Consumption takes DF of 4 and Intercept of 1 and thus total 5. 5-5=0. There is no DF left at all. That is, you are fitting a saturated model. Goodness-of-Fit, say Deviance, is to test the difference between the Full model and the model for Aggregated variables (it is the full model here). That is why the value of Deviance =0 as well.

You can see my blog in which I did some simple calculations.

JingJu