Re: st: Imputing, interpolating, or otherwise finding missing data?

[L Robinson <lysistrataphobia@yahoo.ca>]

Thanks for the reply Maarten!

If I use ice the regular way, and I use the raw data (not the growth rate) I end 
up with a completely linear trend. The data fit better to a logarithmic 
equation, but I'm not sure how to tell ice to use that, or try that... 
otherwise, I think I'm making progress! 


Lori



----- Original Message ----
From: Maarten buis <maartenbuis@yahoo.co.uk>
To: statalist@hsphsun2.harvard.edu
Sent: Wed, June 30, 2010 11:50:29 PM
Subject: Re: st: Imputing, interpolating, or otherwise finding missing data?

--- On Thu, 1/7/10, L Robinson <lysistrataphobia@yahoo.ca> wrote:
> I have a dependent variable that nicely follows the
> equation y=-91.78*ln(x) + 
> 433.73. y is a growth rate, x is time(1,2,3, etc). 
> 
> I can't seem to find a way to incorporate the values of this
> equation into either ipolate or ice.

-ice- needs to estimate that equation, as it implements a 
technique called Multiple Imputation, which requires that 
you not only add some noice around the predictions to represent 
uncertainty due to the error term, but also some noice to 
represent uncertainty in the parameter estimates. So in -ice-
you should not be able to fix the parameters. However, this
noice representing the uncertainty in the parameters is
usualy small compared to the uncertainty in the error term,
so I would just use -ice- in the regular way.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/




*
*   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/