SAS Support Communities

Hi Sunny, The closest we have to this in SAS is utilizing item stores for models estimated by certain models.  Here is an example. %let nObs = 5000; %let nVars = 100; data SimuData;   array x{&nVars};   do obsNum=1 to &nObs;   do j=1 to &nVars;   x{j}=ranuni(1);   end;   linp = 10 + 11*x1 - 10*sqrt(x2) + 2/x3 - 8*exp(x4) + 7*x5*x5   - 6*x6**1.5 + 5*log(x7) - 4*sin(3.14*x8) + 3*x9 - 2*x10;   TrueProb = 1/(1+exp(-linp));   if ranuni(1) < TrueProb then y=1;   else y=0;   output;   end; run; proc logistic data=SimuData;   effect splines = spline(x1-x&nVars/separate);   model y = splines/selection=stepwise;   store sasuser.SimuModel; run; data test;   array x{&nVars};   do j=1 to &nVars;   x{j}=0.15;   end;   drop j;   output; run; proc plm restore=sasuser.SimuModel;   score data=test out=testout predicted / ilink; run; data testout;   set testout(drop=x1-x&nVars); run; proc print data=testout; run; You can find other examples here. SAS/STAT 13.2 User's Guide Example Programs (Sample Library) ... View more

Hi Niam, Thanks for your question.  You seem to have two distinct wants for your model. The first, and your primary research question seems to be whether funding influences publications.  If that is the case, then your FE estimate with the university type dropped is effectively controlling for the university type and all other idiosyncratic confounding effects specific to the professor.  So your estimates, assuming the modeling specified correctly, should be consistent. If you are interested in the marginal effect of type and you want to still control for some idiosyncratic professor effect, you will need to relax some of the assumptions on the way that university type interacts with professor specific effects.  If willing to do this, you can employ another estimator known as a Hausman-Taylor estimator. Currently we do not have super convenient access to this estimator in ETS (but look for it soon ).  In the meantime, if you look at this document, page 210 http://eco.cueb.edu.cn/upload/2012/9/616432590.pdf then you will see how to estimate this in SAS.  It is a multistep process.  You can also find IML code here. Copyright C 2012 by R. Carter Hill and Randall C. Campbell "Using SAS for Econometrics"     by R. Carter Hill and Randall C. Campbell (2012)     John Wiley and Sons, Inc  Chapter 15 Again, the key assumption for this model to make sense is that there is no correlation between the individual professor characteristics and the university type (which seems a little implausible to me though this is your call)         Best of luck-Ken      ... View more

Hello, I assume you are following this example SAS/ETS User's Guide Example Programs. As you can see, after SIMILARITY gives you the similarity matrix, then you can cluster in the same way you would use cross sectional clustering routines.  Use PROC CORR, CLUSTER, whatever you wish. Similarity has a number of utilities but all are related to temporal ordering.  Typical methods of clustering ignore the ordering.  In the time series version of this clustering we are looking for variables(series) that we can treat as a group, rather than observations that we treat as a group.  The SIMILARITY procedure effectively transposes this information (with some other tweaks) so the clustering can be done on the variables.  If you were to use clustering directly (a perfectly sensible practice for some uses) then you would effectively be looking for intervals that behave similarly.  This might be perfectly reasonable for some sort of time series segmentation but that is not what we are showing in this example.  Hope this helps-Ken ... View more