--- Ian Wakeling <ian.wakeling@HANANI.QISTATS.CO.UK> wrote: > Nancy, > > > Date: Wed, 27 Apr 2005 11:13:12 -0400 > > From: Agustin Calatroni <acalatr@UMICH.EDU> > > Subject: Re: NLmixed > > > > Nancy, > > > > Recode varLogi as varLogi-1 and proc nlmixed will converge. > > > > data titi; > > set titi; > > varLogi=varLogi-1; > > run; > > > > As you are using the syntax: > > random u ~ normal(0,s2u) subject=region_n; > > I think you should also sort your data set by the variable > region_n as the random statement is assuming a different > subject every time its value changes. > > > As to why I will let somebody with more experience in non-linear > mixed > > models explain. > > > > HTH, > > >

Ah, good pick-up Ian. You are absolutely correct. Still, there appears to be no random effect variance component in these data. And results are still unstable across the different variance component parameterizations. (See my previous post for more on these latter comments.)

Dale

--------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra@NO_SPAMfhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 ---------------------------------------

__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com