BookmarkSubscribeRSS Feed
jiltao
SAS Super FREQ

The SCALE=P option in PROC GENMOD is one way to address overdispersion, as explained in the documentation below --

https://go.documentation.sas.com/doc/en/pgmsascdc/v_021/statug/statug_genmod_syntax22.htm#statug.gen...

PROC GLIMMIX does not have this option, it uses a separate statement to address overdispersion -

random _residual_;

This is explained in the link below --

https://go.documentation.sas.com/doc/en/pgmsascdc/v_021/statug/statug_glimmix_details05.htm

 

Ksharp
Super User
You could use PROC LOGISTIC + MODEL 's option SCALE= to get it.
You also could use PROC LOGISTIC +EFFECT to get cubic splines effect.
You also could use PROC LOGISTIC +/param=glm + lsmean trt/ ilink diff + %nlmeans to get risk difference or risk ratio.
Check the second example:
https://support.sas.com/kb/23/003.html
For Risk Ratio or Relative Risk

https://support.sas.com/kb/37/228.html
For Risk Difference
BayzidurRahman
Obsidian | Level 7

Hi, My outcome is count with an offset.

Ksharp
Super User
You could change it into Binomial Distribution, and feed it into PROC LOGISTIC .


n    c  age
10   2  1
4    1  2

---->
y  age
1  1
1  1
0  1
0  1
0  1
0  1
0  1
0  1
0  1
0  1

1  2
0  2
0  2
0  2

And using:
model Y(event='1') = age;

Or just usa another style of MODEL statement:
model c/n = age;


Note: This assuming Binomial Distribution , Not Poisson Distribution, 
But I think they are almost the same thing, since Poisson is modeling for more discrete data,
as long as your data is NOT so discrete ,Binomial is more than suffice .
BayzidurRahman
Obsidian | Level 7

Thanks for the information. My data is very discrete with overdispersion. I need to either adopt scaled variance with Poisson distribution or negative binomial distribution.

Ksharp
Super User
You could try my code firstly.
As long as there is no WARNING or ERROR info in LOG ,that would be OK.

hackathon24-white-horiz.png

The 2025 SAS Hackathon has begun!

It's finally time to hack! Remember to visit the SAS Hacker's Hub regularly for news and updates.

Latest Updates

What is ANOVA?

ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 20 replies
  • 4289 views
  • 10 likes
  • 5 in conversation