Evan:
This prety much depends on the proportion of citties that are left censored. If that proportion is
small than I wouldn't bother doing anything fancy about it. If the proportion is large or if I was
feeling nerdish I would first model the raw number of people in poverty with -tobit- (strictly
speaking this assumes normality of the distribution of number of peoples in poverty which can't be
true, but since that count is likely to be large I don't worry about it.). Than I would create
multiple imputed values for the censored observations, and create proportions. Than I would model
that proportion with -betafit- (I couldn't resist the self plug) for each "completed" dataset, and
combine the different estimates using the standard multiple imputation equations.
HTH,
Maarten
--- [email protected] wrote:
> Say that I've got the number of persons living in poverty. As a raw number, this is left
> censored at 100. Ideally, I want to look at the percentage of the population that is living in
> poverty. So not only do I fail to have a constant censoring value but, for a given city with
> fewer than 100 persons living in poverty, I am not certain if that city is left censored or
> right censored....
>
> Do you have any idea how to handle this? Would it be all too terrible to enter poverty = 100
> when poverty (raw) <100, and simply use OLS methods to analyze it?
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting adress:
Buitenveldertselaan 3 (Metropolitan), room Z214
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
___________________________________________________________
NEW - Yahoo! 360 � Your one place to blog, create, publish and share! http://uk.360.yahoo.com
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
