Re: st: Mim, T values, and Probability of T values

[Richard Williams <Richard.A.Williams.5@ND.edu>]

At 01:18 PM 2/23/2010, Eduard Bonet wrote:
> Hello.
>
> I am running OLS regressions on a multiply imputed dataset and have
> come across an outcome I don't quite understand.
> For certain estimated coefficients with T values that are clearly
> higher than 1.96, the probability of those T values is higher that
> 0.05. Does anyone knows why should I get these outputs?
> (I am sorry I cannot reproduce the problem as the data I am using is
> under embargo).

First off, the critical value for t varies with the degrees of 
freedom, so if the sample is small the critical value will be higher than 1.96.

Second, degrees of freedom are a little weird with MI.  If you have 
Stata 11, look at the manual entry for mi estimate.  On p. 47 it says

"Finally, mi estimate reports a coefficient table containing the 
combined estimates. Unlike all other Stata estimation commands, the 
reported significance levels and confidence intervals in this table 
are based on degrees of freedom that is specific to each coefficient. 
Remember that the degrees of freedom depends on the relative variance 
increases and thus on how much information is lost about the 
estimated parameter due to missing data. How much information is lost 
is specific to each parameter and so is the degrees of freedom."

If you use the -dftable- option, you can see what the d.f. are for 
each coefficient.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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/