I haven't tried it, but I think this procedure does not adjust for correlation.  ... View more

@lvm I was referring to @Ksharp response in my post about normality, I agree with your response.  ... View more

You said:   "in testing hypothesis, the part of robustness is ability of the procedures to control Type I error rate of a test to close to the significance level α"   That's correct.  So what is question you are asking?  Is it about the robustness of some particular statistical method?  If so, what method?   The only way I know of to measure robustness is through simulation--and simulating enough datasets that violate an assumption to adequately test is difficult.   Steve Denham ... View more

Hello - On behalf of Technical Support:   "The reason why the Seasonal Component and the Seasonally-Adjusted series are missing could be releated to no non-missing Seasonal-Irregular component (SIC) values associated in your data. The Seasonal Component is computed as the average of the SIC for a given seasonal period. (Details on the computation of each of the components can be found here: http://support.sas.com/documentation/cdl/en/etsug/68148/HTML/default/viewer.htm#etsug_timeseries_details07.htm)"   Technical Support has also created an example to illustrate this point further:   data test;   set sashelp.air(where=('01dec1958'd <= date <='01oct1960'd)); run;   proc timeseries data = test outdecomp = out;   id date interval=month;   decomp orig tcc sic sc ic sa/mode=mult;   var air; run;   proc expand data=test out=expandout method=none;   id date;    convert air=tc / transformout=( cd_tc 12 );    convert air=my_tc / transformout=(cmovave 12 movave 2 trim 6);    convert air=s  / transformout=( cd_s  12 );    convert air=i  / transformout=( cd_i  12 );    convert air=sa / transformout=( cd_sa 12 ); run;   proc print data=out; run;   proc print data=expandout; run;   Thanks, Udo ... View more

@plf515,   Your intuition is right on the nose.  The advantage to the means model approach is that I don't have to come up with code and level labels, which is big for me because I am really lazy.   I think means model approaches were touted by Milliken and Johnson because the Type IV hypotheses (designed for missing cell analyses) in PROC GLM were not unique, whereas the means model resulted in unique hypothesis tests.  Now that the LSMESTIMATE statement is available, I can see a lot more analyses going this route.   Steve Denham ... View more

This may be something as simple as replacing the '.'s in the GENERAL option with zeroes:   covtest general 0 1 0 0 -1;   Steve Denham ... View more

I didn't attend that SGF talk, so I don't know the details, but that statement seems to incite unnecessary worry.  I regularly use MERGE without a BY statement to horizontally concatenate data that I then use to use in creating ODS graphics. (See the end of this article for a discussion.) I've never had a problem.   Philosophically, you have to accept that the first observation in data set 'A' might have nothing to do with the first observation in data set 'B'. Of course, I wouldn't do a statistical analysis of the combined data unless I knew that the observations lined up.   Perhaps a more fair poem would be:   To horizontally concatenate, MERGE without BY is really great. But if the variables don't relate, a statistical analysis is tempting fate.     ... View more

  Yes, so, these 20 volunteers all tested the 4 different solutions at different study days. So that's why it is repeated (with the same subjects and the same solutions for every subject).    ... View more

Actually, there is one more possibibility that may be even easier than using the macro i suggested above. because your time-dependent variable is piecewise constant, you can make your dataset such that you split observations at the time where the covariate change it values. That is, you should have a entry and a exit variable such such that that the total observation time is the same. be aware that person that has event should be censored at time of change of covariate and only have a eventtime on the original eventtime. Then you can use phreg with delayed entry: model (entry exit)*event(0)=covariates Then you avoid creating timedependent variables inside phreg. The method will be fast if you use the FAST-option : PROC PHREG DATA=mydata FAST; ... RUN; It was introduced in the latest release from august this year, but included in the university edition. Without the fast-option the method with delayed entry will not be faster than using timedependent variables. ... View more

Well, certainly type is going to cause problems with all those IVs to be used.  You can either drop that type or try to combine it with another.   Beyond that, I think you need to also look at the IVs and their distribution. ... View more

You've gotten some good advice regarding MEANS.  Especially the advice to BE CAREFUL.   In your question, though, you used MEANS only as an example.  If you  want to do regression, I suggest either QUANTREG or ROBUSTREG.   Oddly enough, I wrote a paper on these. Should more of your PROC REGs be QUANTREGs and ROBUSTREGs?   In addition, note that the outliers are often the interesting stuff - automatically excluding outliers would mean that we never discovered black holes. :-). ... View more

It is the same thing. Since your reference level is not part of your regression (dropped), removing insignificant dummy is essentially the same as combining it with your reference. So you just have new reference variable implicitly.   ... View more

Ok, time to talk to tech support. I'll report back on what they say. ... View more

Thanks again! ... View more