Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: nested data

From   Natasha Sarkisian <>
Subject   st: nested data
Date   Tue, 27 Sep 2005 16:36:10 -0400


I have a nested dataset and I need advice on what types of analyses
would be appropriate to use with it.  The data are collected from groups
of siblings; in each family, each sibling answers questions regarding
each of the other siblings.  So in each family:
Obs 1  Sibling 1 rates sibling 2
Obs 2  Sibling 1 rates sibling 3
Obs 3  Sibling 2 rates sibling 1
Obs 4  Sibling 2 rates sibling 3
Obs 5  Sibling 3 rates sibling 1
Obs 6  Sibling 3 rates sibling 2

Obviously, these observations are not independent.  First of all, there
are family groups, but that's easy to deal with.  But what do I do about
the fact that siblings are rating each other?  I could make the family
to be the third level unit, individual siblings -- the second level, and
their ratings of specific "target" siblings -- the first level (i.e.
group together obs 1 and 2; obs 3 and 4; obs 5 and 6). But the lack of
independence remains with such grouping, because target siblings are the
same -- i.e. we could also group obs 1 and 6; obs 2 and 4; and obs 3 and
5. So each observation is cross-nested, but not in a typical sense of
belonging to two different types of groups on the same level (e.g.,
schools and neighborhoods, or boys and girls, etc.) -- they just belong
to two "groups" of the same type at once.  If anyone has any advice, I'd
appreciate it.

Natasha Sarkisian
*   For searches and help try:

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