Programming the statistical procedures from SAS

Include multiple independent variable

Reply
Occasional Contributor HM3
Occasional Contributor
Posts: 9

Include multiple independent variable

I want to do a logistic regression where there are many indepedent varables. How to

  1. write them all varaibles ?
  2. write a subset of all the variables?
  3. How to know which variables are imp for my calculations?
  4. Can VIF be done in logistic regression ?

Pl guide.

 

Regards

 

Grand Advisor
Posts: 16,880

Re: Include multiple independent variable

RE 1/2

To list multiple or select variables you can use variable lists.

Read the documentation that illustrates how you can reference multiple variables at once.

 

http://support.sas.com/documentation/cdl/en/lrcon/68089/HTML/default/viewer.htm#p0wphcpsfgx6o7n1sjtq...

 

RE 3

Look at the P-Value along with effect size to determine if a variable is significant

 

RE 4

I don't know if it does, but you can use PROC REG to calculate VIF first if required.

Grand Advisor
Posts: 9,451

Re: Include multiple independent variable

How to write them all varaibles ? Use variable list method : model y=FirstVariable -- LastVariable; write a subset of all the variables? Ditto. How to know which variables are imp for my calculations? Check ANOVA Table and ODDS Ratio. The biggest ODDS Ratio should be the most important variable. Can VIF be done in logistic regression ? No. You can't . But you can use SELECTION=STEPWISE to pick up the best sub-model.
Grand Advisor
Posts: 9,451

Re: Include multiple independent variable

How to write them all varaibles ?
 Use variable list method : model y=FirstVariable -- LastVariable; 

write a subset of all the variables? 
Ditto. 

How to know which variables are imp for my calculations?
 Check ANOVA Table and ODDS Ratio. 
The biggest ODDS Ratio should be the most important variable. 


Can VIF be done in logistic regression ? 
No. You can't . But you can use SELECTION=STEPWISE to pick up the best sub-model.

Ask a Question
Discussion stats
  • 3 replies
  • 120 views
  • 0 likes
  • 3 in conversation