Statalist


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

st: AW: xtmixed and multilevel models


From   "Martin Weiss" <martin.weiss1@gmx.de>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: AW: xtmixed and multilevel models
Date   Sun, 20 Dec 2009 17:30:59 +0100

<> 

" I am a bit confused about the number of levels in a multilevel model. In
Stata v 10 manual XT page 293 (xtmixed). In example 4 they describe a "2
level model" with individuals, states and regions. To me it looks like a 3
level model. What´s going on?"



The 10 manual says quite clearly on page 286 (quote is from the 11 manual):
" Finally, using formulation (2) and its multilevel extensions requires one
important convention of terminology. Model (2) is what we call a one-level
model, with extensions to two, three, or any
number of levels. In our hypothetical two-level model with classes nested
within schools, the schools
are considered the first level and classes, the second level of the model.
This is consistent with
terminology used elsewhere, e.g., Pinheiro and Bates (2000), but differs
from that of the literature
on hierarchical models, e.g., Skrondal and Rabe-Hesketh (2004). In that
literature, our schools and
classes model would be considered a three-level model, with the students
forming the first level,
classes the second, and schools the third. Not only is there one more level
(students) but the order
is reversed."



HTH
Martin

-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Tomas Lind
Gesendet: Sonntag, 20. Dezember 2009 17:28
An: statalist@hsphsun2.harvard.edu
Betreff: st: xtmixed and multilevel models

Dear listers,

I am a bit confused about the number of levels in a multilevel model. In
Stata v 10 manual XT page 293 (xtmixed). In example 4 they describe a "2
level model" with individuals, states and regions. To me it looks like a 3
level model. What´s going on?

I can´t find any example in xtmixed-regression with explanatory variables on
a group level (for example on level 2 in a 2-level model). If I want to
explain why a random intercept varies between group 2 units because of
differences in socioeconomic status between groups do I proceed like this?

I generate a new variable to the dataset "group_mean_ses" which is the mean
ses within group. All individuals in a group have the same value.

Now I add the variable group_mean_ses as an explanatory variable in the
xtmixed model (don´t worry over the variables, this is not a real study).

xtmixed  health  age  sex  ses   group_mean_ses  ||  group


If I have a random slope on some individual level variable, say age, and
want to explain that with group_mean_ses should I generate a cross-level
interaction variable.

generate gmsXage = group_mean_ses * age

and then specify the model as

xtmixed  health  age  sex  ses   group_mean_ses  gmsXage  ||  group : age


All comments or ideas are welcome.


/Tomas







*
*   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/


*
*   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   |   What's new   |   Site index