Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
st: minimum number of obs for xtmixed
From
"Miller, Daniel P" <[email protected]>
To
"[email protected]" <[email protected]>
Subject
st: minimum number of obs for xtmixed
Date
Thu, 9 Jun 2011 10:12:01 -0400
Hello all -
I am running a multi-level growth curve model using xtmixed of the sort:
xi: xtmixed y i.x*time i.x*time2 $covariates || id:time time2, var mle
where x is my main categorical independent variable interacted with time [time] and time-squared [time2].
There are as many as four waves of data for each respondent, although there is a good deal of missing data.
My question is about the minimum number of observations reported by xtmixed= .
The model results provide the following:
Mixed-effects ML regression
Number of obs = 31082
Group variable: id Number of groups = 8506
Obs per group: min = 1
avg = 3.7
max = 4
How can a "group" (person) contribute 1 observation to the model?
When I create a variable to measure the number of observations, I get different but more sensible results:
After running the model, I create a flag:
gen insample=1 if e(sample)
Then:
sort id
by aid: gen count =_N if insample==1
(34045 missing values generated)
. tab count
count | Freq. Percent Cum.
------------+-----------------------------------
2 | 581 1.87 1.87
3 | 6,903 22.21 24.08
4 | 23,598 75.92 100.00
------------+-----------------------------------
Total | 31,082 100.00
Which would seem to indicate that the minimum number of observations is 2, but most people contributed 3 or 4.
Thanks much in advance,
Dan Miller
Boston University
*
* 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/