Re: st: anchoring vignettes and gllamm problem

[Schuyler Blasy <schuyler.blasy@gmail.com>]

Dear Steve,

Thank you very much. I truly appreciate your reply.

I just wanted to mention that today I modified Professor King's data.
I cut out all but 16 observations, much like what my data would be,
and I ran into the same problem. It seems that I just do not have
enough observations to run this program.

Again Steve, thank you very much. You confirmed what I seemed to find
today too.

Best wishes,
Sky

On Fri, Jul 26, 2013 at 7:08 PM, Steve Samuels <sjsamuels@gmail.com> wrote:
>
> To be clearer, For non-linear models, you need at least 10 observations
> per predictor/
> parameter.  See: www.psychosomaticmedicine.org/content/66/3/411.full.pdf.
> So, you have enough for one predictor. So it's not surprising that you
> cannot
> fit your model.
>
> Steve
>
> I'd say the problem is small sample size; there is information
> insufficient
> information for finding a solution.
>
> Steve
>
>
> On Jul 26, 2013, at 1:14 PM, Schuyler Blasy wrote:
>
> Hello everyone,
>
> I am having troubles with using -gllamm- command in applying it to
> anchoring techniques; my Stata 12 will just continue to process and
> not complete the computations. I am running on a windows 7 machine
>
> Here is a bit of background. I am a just a masters student doing an
> internship for a nonprofit that wants self-assessment surveys. I am
> trying to incorporate the anchoring vignette technique into the
> evaluation, but I am not too familiar with the Stata code and data
> analysis part of it.
>
> I found the Gary King article, and his simulated response/data with do
> file and I was able to run it successfully on my computer.
> http://www.gllamm.org/examples.html
>
> However, when I tried to create my own data set and do file, Stata
> gets hung up on the gllamm part and just never finishes computing.
> Professor King’s data set and do file finish in about 4 to 5 minutes,
> but mine has gone on for 10 hours with no end in sight.
>
> There are differences in Professor King’s and mine data. Mine uses 3
> vignettes with a scale of 0 through 10, and one self-assessment.  But,
> I believe Professor King uses 5 vignettes with a scale of 1 through 5.
>
> Also, I only tried my data set with 16 total observations, whereas
> Professor King’s data set contained hundreds.
>
> The part that Stata gets hung up on is:
>
> gllamm xsay s_Treatment s_age s_male s_educyrs i1
> > i2 i3, i(id) link(soprobit) s(het) ethresh(thresh)
> > constr(1) init;
>
> Can anyone give advice or an idea why my data set is not completing
> processing?
>
> I sincerely appreciate any advice or input anyone has on this problem.
> I have to admit, I am a Stata novice at best but I am learning, and I
> am just a masters student, but I’m trying to do my best.
>
> Appreciatively yours,
>
> Schuyler Blasy
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/