Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Mixed model using splines - can RE for selected knots only be tested at level 3


From   "Buzz Burhans" <buzzb3@earthlink.net>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Mixed model using splines - can RE for selected knots only be tested at level 3
Date   Thu, 8 Sep 2011 13:30:42 -0600

I am running a mixed model which is similar to a growth curve model, with
three levels of nesting (measurement occasion in cow in herd).  I am using
-xtmixed-, with a set of 12 spline knots created over the time frame of the
measurement occasions as the RHS variables.  The dataset is large, with a
large number of cows and herds at level 2 & 3, which makes modeling the RE
slow and problematic.  I am interested in both the fixed effects of the
spline knots, and in the random effects of the knots across herds (level 3).
Selected specific knot RE are of more interest than others (those at the
beginning of the curve, and another at a midpoint).  Modeling all the knots
as RE is extremely slow and has convergence problems.  

Would it be appropriate to model only specific knots of interest as RE, or
must I test all knot RE for inclusion in the model? All knots would remain
in the fixed effects equation, my question is about the need for inclusion
of all knots vs. selected knots in the RE equation at the 3rd level.

Thanks for any insight.

Buzz

Buzz Burhans, Ph.D.

Dairy-Tech Group
So. Albany, VT / Twin Falls ID

Cell: 208-320-0829
ID Fax: 208-735-1289
VT Fax: 802-755-6842

Email: buzzb3@earthlink.net



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


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index