Quartz | Level 8

## How to use proc logistic to estimate probability difference -- or common risk difference

I try to use proc logistic regression to estimate the probability difference between placebo and treatment group. It gives me odds ratio difference, how can I change to risk difference?

Also any difference between "risk difference" or "proportional difference"  or "probability difference" between proc freq and proc logistic?

Thanks,

1 ACCEPTED SOLUTION

Accepted Solutions
Super User

## Re: How to use proc logistic to estimate probability difference -- or common risk difference

Better post it at Statistical Forum ,since it is about stat . @StatDave  @Rick_SAS  is there .

If your data is contingency table , like

placebo   treatment

good    12           23

you could try

PROC FREQ;

run;

"risk difference" is for   12/(23+12)  V.S.  2/(2+3)   or  23/(23+12)  V.S.  3/(2+3)

"proportional difference"(a.k.a  odds ratio) is for 12/23  V.S.  2/3

2 REPLIES 2
Super User

## Re: How to use proc logistic to estimate probability difference -- or common risk difference

Better post it at Statistical Forum ,since it is about stat . @StatDave  @Rick_SAS  is there .

If your data is contingency table , like

placebo   treatment

good    12           23

you could try

PROC FREQ;

run;

"risk difference" is for   12/(23+12)  V.S.  2/(2+3)   or  23/(23+12)  V.S.  3/(2+3)

"proportional difference"(a.k.a  odds ratio) is for 12/23  V.S.  2/3

Quartz | Level 8

## Re: How to use proc logistic to estimate probability difference -- or common risk difference

Thank you!
Actually I find a very useful link. it explained everything:
http://support.sas.com/kb/37/228.html
Discussion stats
• 2 replies
• 2597 views
• 3 likes
• 2 in conversation