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: problem with _statsby_
From
"Feiveson, Alan H. (JSC-SK311)" <[email protected]>
To
"[email protected]" <[email protected]>
Subject
st: problem with _statsby_
Date
Wed, 22 Sep 2010 09:54:55 -0500
Hi - I am trying to run a complicated -xtmixed_ command with -statsby-, but it appears that _statsby_ doesn't like the "cov(exchangeable)" option.
Contains data from braslet_long_092110.dta
obs: 19,197
vars: 11 22 Sep 2010 09:41
size: 575,910 (99.4% of memory free)
----------------------------------------------------------------------------------------------------------------
storage display value
variable name type format label variable label
----------------------------------------------------------------------------------------------------------------
ord float %9.0g
outcome byte %9.0g group(meas)
sub str1 %9s Astronaut
fd int %8.0g Time of Flight
braslet str3 %9s Braslet
hrep float %9.0g session # (for y4 = HR only)
beatno byte %9.0g heartbeat #
condition byte %9.0g cond baseline, Valsava, or Mueller
y float %9.0g
ib byte %8.0g
isub float %9.0g group(sub)
---------------------------------------------------------------------------------------------------------Sorted by: ord outcome
Here- The -xtmixed- command works fine for one combination:
. xtmixed y ib if condition==1 & outcome==19 ||isub: R.ib,cov(exchangeable) ||fd: R.ib,cov(exchangeable) nolog
Mixed-effects REML regression Number of obs = 84
-----------------------------------------------------------
| No. of Observations per Group
Group Variable | Groups Minimum Average Maximum
----------------+------------------------------------------
isub | 8 6 10.5 12
fd | 14 6 6.0 6
-----------------------------------------------------------
Wald chi2(1) = 8.34
Log restricted-likelihood = -341.96484 Prob > chi2 = 0.0039
------------------------------------------------------------------------------
y | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
ib | 24.11126 8.350805 2.89 0.004 7.743986 40.47854
_cons | 176.8267 10.03752 17.62 0.000 157.1536 196.4999
------------------------------------------------------------------------------
------------------------------------------------------------------------------
Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval]
-----------------------------+------------------------------------------------
isub: Exchangeable |
sd(R.ib) | 24.10437 7.628245 12.96338 44.82014
corr(R.ib) | .6203086 .3331783 -.323844 .9454376
-----------------------------+------------------------------------------------
fd: Exchangeable |
sd(R.ib) | 18.48145 4.967922 10.9126 31.29998
corr(R.ib) | .8079071 .1894934 .0514915 .9752792
-----------------------------+------------------------------------------------
sd(Residual) | 9.483174 .8960756 7.879937 11.4126
------------------------------------------------------------------------------
LR test vs. linear regression: chi2(4) = 116.96 Prob > chi2 = 0.0000
But when I try it with -statsby- , i get an error:
. statsby _b[ib],by(condition outcome): xtmixed y ib ||isub: R.ib,cov(exchangeable) ||fd: R.ib,cov(exchangeable) nolog
invalid 'cov'
r(198);
Yet, when I remove the "cov(exchangeable)" option, -statsby- works:
. statsby _b[ib],by(condition outcome): xtmixed y ib ||isub: R.ib ||fd: R.ib, nolog
(running xtmixed on estimation sample)
command: xtmixed (y ib) (isub: R.ib) (fd: R.ib, nolog)
_stat_1: _b[ib]
by: condition outcome
Statsby groups
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
...............--Break--
r(1);
I guess I could do this by hand, but I wonder if anyone else has encountered similar problems with -statsby- .
Al Feiveson
*
* 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/