[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

# st: conditional GLLAPRED in 3-level logit model

 From "Mark Hebblewhite" To Subject st: conditional GLLAPRED in 3-level logit model Date Tue, 4 Oct 2005 22:53:38 -0600

```Dear Stata list serve,

I  am writing to ask for advice about using GLLAPRED after GLLAMM to obtain
conditional probabilities in a three-level (locations within individual
wolves within wolf-packs) logit model with a random coefficient for an
analysis I am doing on wolf resource selection.

My model structure is :

gllamm useavail (some other variables X1- - X3) disthgh, i( wolfuid packid)
nrf(1,2) eqs(wolfhuman wolfhuman packhuman) fam(binom) link (logit) nip(8)
trace
where wolfhuman =1 and packhuman is set to be a random coefficient for var
disthgh at the 3rd level (i.e. pack). So random intercepts for wolf and
pack, and a random coefficient for disthgh at the packlevel.

disthgh is a continuous variable describing distance to high human use
trails and roads.

My general question is how to obtain conditional predictions for the 3rd, or
pack level, using GLLAPRED. I have acheieved this for a two-level model, but
am running into conceptual problems for 3-level models.

1) At a general level, I am unclear how many generated variables I need to
designate - do I need to specify GLLAPRED varname, u for all three levels of
latent variables? i.e. GLLAPRED varname, u(var1 var2 var3) where var1 2 and
3 are defined by me as, and this leads to my second question,

2) How to determine what to set the value of the various random effects at.
In my case, I have 5 wolf packs each with different mean distance to high
human use (from 0.2km to 4.4km) - would it be reasonable to set these then
representing the mean values for disthgh as the values at which to evaluate
my conditional predictions at? This seems to be the case in one of the
examples in the new stata book, but then elsewhere I get the impression its
just set at values of SD's. So at a general level I am unclear it seems
about GLLAPRED.

I seem like I might be missing something quite basic here, but hopefully my
question is not too trivial to obtain some advice.

Gratefully yours,

Mark Hebblewhite
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Mark Hebblewhite, PhD Candidate
Department of Biological Sciences
University of Alberta
Edmonton, Alberta
T6G 2E9
Office phone: 1-780-492-0083
Home phone: 1-780-431-1476
email: mark.hebblewhite@ualberta.ca
project website: http://ursus.biology.ualberta.ca/yhtelkwolfproject/
personal webpage: http://www.ualberta.ca/~mh12
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```

 © Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index