0
$\begingroup$

I am running a mixed model in SAS using PROC glimmix, to analyze longitudinal count data (Poisson distribution, autoregressive covariance structure), and would like to produce fixed effect estimates for the continuous predictors in my model. I am working with the following code:

proc glimmix data = activedata;

class id time;

model DV = time var1 var2 var3/ dist=POISSON;

random _residual_;

random intercept var1 var2  / subject=id type=un;

random time / subject=id type=ar(1)  residual;

run;

This code runs without error and produces the expected "Type III tests of Fixed Effects" table (which gives F- and p-values), but I also need to estimate coefficients, standard errors, t- (or Wald), and p-values, as usually produced in the "Solutions for Fixed Effects" table. Any help for code that would help to produce these estimates (or why I will/should not be able to produce them) would be greatly appreciated!

$\endgroup$
1
  • $\begingroup$ Have you tried communities.sas.com ? That might be a better place to get SAS-specific help. $\endgroup$
    – JimB
    May 17, 2018 at 19:28

1 Answer 1

0
$\begingroup$

Use the "solution" option in the model statement. e.g.:

proc glimmix data = activedata;

class id time;

model DV = time var1 var2 var3/ solution dist=POISSON;

random _residual_;

random intercept var1 var2  / subject=id type=un;

random time / subject=id type=ar(1)  residual;

run;
$\endgroup$
5
  • $\begingroup$ Thank you @Ryan Simmons. I’ve tried adding the “solution” option before but in a slightly different spot (after “dist = Poisson”), so I will try it this way and post if that works. $\endgroup$
    – Rebecca
    May 18, 2018 at 11:15
  • $\begingroup$ @Rebecca the order in which you put the options after the "/" doesn't matter. SAS will process the option regardless. $\endgroup$ May 18, 2018 at 16:59
  • $\begingroup$ You can also put the line "ods output ParameterEstimates = datasetname", which will create a dataset (called datasetname) that includes all the parameter estimates for the fixed effects, with associated standard errors, etc. $\endgroup$ May 18, 2018 at 17:00
  • $\begingroup$ You can put that line either before the PROC GLIMMIX line, or anywhere during the PROC call. I usually prefer to put it at the very end of the PROC call, right before the "run" statement, but that's personal preference, it will work equally well elsewhere. $\endgroup$ May 18, 2018 at 17:01
  • $\begingroup$ Ryan, the “solution” option is working as you suggested now - I’m really not sure why it wasn’t working before. I will also try the ods output line you suggested when I have access to SAS again next Tuesday. Many thanks for all the answers! $\endgroup$
    – Rebecca
    May 19, 2018 at 2:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.