topic Re: Degrees of Freedom in proc logistic output in Statistical Procedures

Degrees of Freedom in proc logistic output

KafeelBasha — Wed, 23 Aug 2017 05:42:58 GMT

Hello

In proc logistic output how the 'DF' column has 1 degrees of freedom for each independent variable.

Is there any formula?. Please help.

Re: Degrees of Freedom in proc logistic output

PeterClemmensen — Wed, 23 Aug 2017 06:31:32 GMT

Each ML estimate in Proc Logistic is free to vary, therefore it has 1 degree of freedom (itself).

If you look in the "Deviance and Pearson Goodness-of-Fit Statistics" you will see that the value in the DF column is the sum of the degrees of freedom in the individual estimates.

Re: Degrees of Freedom in proc logistic output

KafeelBasha — Wed, 23 Aug 2017 09:35:48 GMT

Can you please suggest some reference book, I didn't get it exactly.

Re: Degrees of Freedom in proc logistic output

PeterClemmensen — Wed, 23 Aug 2017 09:49:47 GMT

A reference book on PROC LOGISTIC or to Basic Statistics?

In regards to PROC LOGISTIC, I recommend Proc Logistic Using SAS by Paul D Allison

A good introduction to statistics in SAS is SAS Statistics by Example by Ron Cody

Re: Degrees of Freedom in proc logistic output

Rick_SAS — Wed, 23 Aug 2017 12:42:39 GMT

Each PARAMETER requires 1 DF, but a classification variable that has k levels requires k-1 DF. For example, if a classification variable has the values 'A', 'B', and 'C', that variable uses two DF.