Why proc nlmixed gave different results?

lxl65 — Wed, 10 Jan 2018 15:19:55 GMT

dear all,

I have a question related to proc nlmixed. I am reading following SAS article:

https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_nlmixed_sect006.htm

The original SAS code in this article is:

   proc nlmixed data=tree;
      parms b1=190 b2=700 b3=350 s2u=1000 s2e=60;
      num = b1+u1;
      ex  = exp(-(day-b2)/b3);
      den = 1 + ex;

      model y ~ normal(num/den,s2e);

      random u1 ~ normal(0,s2u) subject=tree;
   run;

I changed the above code to following:

   proc nlmixed data=tree;
      parms b1=190 b2=700 b3=350 s2u=1000 s2e=60;
      num = b1+u1;
      ex  = exp(-(day-b2)/b3);
      den = 1 + ex;

      lh  = pdf('normal', y , num/den, s2e);
      ll  = log(lh);
      model y ~ general(ll);

      random u1 ~ normal(0,s2u) subject=tree;
   run;

The parameter estimates of two codes are almost identical except for one parameter "s2e". The "s2e" from original code is 61.5; however the "s2e" from modified code is only 7.8.

I thought the two codes should give same results. Can anyone point out what went wrong in modified code?

Thank you very much!

Re: Why proc nlmixed gave different results?

Rick_SAS — Thu, 11 Jan 2018 17:51:06 GMT

The models are equivalent, but the code from the doc is estimating the variance whereas your code is estimating the standard deviation. Note that 61.5139 = (7.8430)**2

topic Why proc nlmixed gave different results? in Statistical Procedures

Why proc nlmixed gave different results?

Re: Why proc nlmixed gave different results?