PROC MIXED AIC -- Differences between SAS Studio and SAS/STAT

[toxicatom5]

Hello,

My colleague and I are looking at model selection criteria. I'm using SAS/STAT and they are using SAS Studio. We're running the same PROC MIXED code, but are getting slightly different output values for -2 Res Log Likelihood and, subsequently, for AIC. Is anyone able to elaborate on why these differences may be occurring between SAS Studio and SAS/STAT?

Thanks in advance for any help!

Adam

[ballardw · Accepted solution]

Since you are running one different processors, with SAS Studio possibly using a completely different operating system. Consider if you use Windows with a local install and your compatriot runs on a system with SAS Studio connecting to a Unix server. There are a number of permutations of this.

If the values you see start at several places to the right of a decimal place then different processors/OS and rounding are likely and typical.

[toxicatom5]

Thank you -- this makes sense and I appreciate your insight.

[SteveDenham]

Be aware that the same issue that @ballardw outlined can happen even with similar operating systems and the same version of SAS/STAT.  Check to make sure that your CLASS and MODEL statements align with each other and with the sort in the dataset you are analyzing.  We just ran into this issue and a sort, followed by a CLASS statement with the by variables from the sort in the same order, and a MODEL statement that reflects the design solved our differences.

SteveDenham

[toxicatom5]

This is a very interesting addition, and makes complete sense. Thank you -- I will check that we have everything aligned and see if it resolves our differences.