topic Re: PROC SURVEYREG AND DUNNETT in Statistical Procedures

PROC SURVEYREG AND DUNNETT

ANKH1 — Thu, 19 Sep 2019 21:01:41 GMT

Hello,

We are running the following code:

proc surveyreg data=sample1 varmethod=brr;
CLASS ANIMALCAT;
MODEL boxcox_FSDDFI1=ANIMALCAT SMK_202 /ANOVA;
LSMEANS ANIMALCAT/DIFF=CONTROL('50-74.99') MEANS CL ADJUST=DUNNETT PDIFF;
WEIGHT WTS_P;
REPWEIGHTS BSW1-BSW500;
RUN;

And we get the following output:

But we don't get an adjusted p-value for each comparison with control. What are we missing?

Thank you in advance!

Re: PROC SURVEYREG AND DUNNETT

Reeza — Thu, 19 Sep 2019 21:28:24 GMT

What does the log show? Notes and Warnings.

Re: PROC SURVEYREG AND DUNNETT

ANKH1 — Thu, 19 Sep 2019 21:34:48 GMT

It says the following:

WARNING: The ADJUST=DUNNETT p-values and confidence limits cannot be computed because the
Dunnett-Hsu approximation did not converge. Try ADJUST=SIMULATE.
NOTE: PROCEDURE SURVEYREG used (Total process time):
real time 27.27 seconds
cpu time 23.83 seconds

I used ADJUST= SIMULATE and got this

Is SIMULATE the same as DUNNETT? Can I report that p-value and say post-hoc DUNNETT was used?

Re: PROC SURVEYREG AND DUNNETT

Reeza — Thu, 19 Sep 2019 22:25:41 GMT

I don't think they're the same. Check the docs for the differences and for why it may not converge.

Re: PROC SURVEYREG AND DUNNETT

ANKH1 — Fri, 20 Sep 2019 15:54:31 GMT

I checked support SAS website and found this:

Problem Note 36073: ADJUST=DUNNETT DIFF=CONTROL multiple comparisons may not converge in PROC LIFETEST

"The Dunnett-Hsu approximation depends on an iterative algorithm to approximate the covariance matrix. In the k-sample tests scenario, the covariance matrix is singular, which causes problems with the iterative algorithm.

To circumvent the problem, use ADJUST=SIMULATE."

Should it be ok then to use SIMULATE? Do you recommend any documents that I can look at?