Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: censored random normal with a fixed mean and standard deviation


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: censored random normal with a fixed mean and standard deviation
Date   Wed, 21 Oct 2009 23:59:09 -0400

Scott Merryman <scott.merryman@gmail.com> et al.:
Thanks to Kit Baum, you can now:

ssc inst cnormp
clear all
set obs 100000
cnormp 25 30, c(0)
g y=max(0,rnormal(r(m),r(s)))
su

The program currently does not do a lot of error checking (e.g. that
the censoring point is less than the desired mean), but will
eventually be updated to produce informative error messages in various
nonsensical cases.  If anyone has any ideas about how to improve on
the method, I'd be interested to hear them.  Probably there is an
easier solution that simply did not occur to me--I am currently
solving the system of nonlinear equations by brute force using Mata's
optimize().

On Tue, Oct 20, 2009 at 2:48 PM, Scott Merryman
<scott.merryman@gmail.com> wrote:
> In generating censored random normal variate with a fixed mean and
> standard deviation is there a way to determine what the mean and
> standard deviation should be for the uncensored normal.
>
> For exmaple, if one wants a cenored normal (at 0) with mean 25 and
> standard deviation 30, then
>
>  -gen yield = max(0, rnormal(25,30))-
>
> will result in a mean too large and standard deviation too small.  In
> this example values of -rnormal(13.5,43.75)- work, but is there way to
> determine what the uncensored mean and standard deviation should be?
>
> clear
> set obs 1000000
> set seed 123456
> gen yield1 = max(0, rnormal(25,30))
> gen yield2 = max(0, rnormal(13.5,43.75))
> sum
>
>
> Scott

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index