Re: st: predicting consumption

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Wed, 9/3/11, gemini mtei wrote:
> I am trying to predict household total consumption from
> the national household budget survey to a small survey
> that we conducted but didn't collect consumption. I have
> used a linear model (OLS) as follow,
<snip> 
> The model is giving me R-square of .55 and i have done all
> diagnostic tests and it seems fine. I have used the split
> half method for validation of the predicted consumption but
> (i.e. selecting a random sample from the households survey,
> run consumption model and predict into the remaining sample
> then compare with actual consumption) the problem i am
> facing is the model over predicts consumption for the
> households with low consumption while it under predict for
> households with higher consumption. 

That is to be expected when using this type of regression 
imputation, which is why you should not do it.

One option would be to stack the two datasets and do a 
multiple imputation for income. The problem I would expect
there is that that imputation model must at least include
all covariate you later want to use in your model of interest
and I guess that the big dataset does not include them all
(why else would you go for the smaller dataset?). 

We can do a bit to fix mistakes in data collection, but
there are limits, and you just hit one of them. At some point 
we just have to except that if data was not collected, then 
that information just does not exist. If we want to solve 
that, we will just have to get our hands dirty and start 
collecting the data we want. (Or, and that is often more 
practical, change the research question so it can be answered
with the available data.)

Sorry for this not very optimistic message,
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/