Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: standard errors for varience components on a second order polynomial using 3 observations

 From "Mustillo, Thomas J" To "statalist@hsphsun2.harvard.edu" Subject st: standard errors for varience components on a second order polynomial using 3 observations Date Fri, 21 Jan 2011 21:29:20 +0000

```Hi. I have a panel dataset of 3 observations on an outcome variable, vote%, for a political party competing in 50 electoral districts. I use a two-level model (time nested in district), with time measured as 0, 1, and 2 to correspond with three consecutive elections. I'm trying to derive variance components on a second-order polynomial, but find that Stata does not yield all standard errors on the random effects. A plot of the data show a clear 2nd order polynomial form, and the estimates of the random effects seem sensible, but I wonder: Why the failure to compute the standard errors?
The model I estimate:
.xtmixed vote% time time^2 ||district: time time^2, mle cov(unstr)
Two examples of parties with problematic results:
Example 1:
------------------------------------------------------------------------------
Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
district: Unstructured     |
sd(t) |   4.523789   .8339028      3.152058    6.492478
sd(t_2) |   1.282918   .2985322      .8130666    2.024285
sd(_cons) |   4.851168   .7807332      3.538803    6.650224
corr(t,t_2) |  -.9799059          .             .           .
corr(t,_cons) |  -.9609035    .013391     -.9800929    -.923927
corr(t_2,_cons) |   .8866836   .0240602      .8292161    .9256002
-----------------------------+------------------------------------------------
sd(Residual) |   .4832775   .1806657      .2322674    1.005553
------------------------------------------------------------------------------
Example 2:
------------------------------------------------------------------------------
Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
districtid: Unstructured     |
sd(t) |   5.255947          .             .           .
sd(t_2) |     1.6143          .             .           .
sd(_cons) |   4.185948          .             .           .
corr(t,t_2) |  -.9894518          .             .           .
corr(t,_cons) |  -.9597653          .             .           .
corr(t_2,_cons) |   .9101291          .             .           .
-----------------------------+------------------------------------------------
sd(Residual) |   .8413581          .             .           .
------------------------------------------------------------------------------

*
*   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/
```