Re: st: RE: RE: calculating standard deviation of age at death from period life tables

[Steve Samuels <sjsamuels@gmail.com>]

"Tricking" Stata's survival programs to accept age-period data, as
recommended in the 2004 Statalist post, may be unnecessary. The
user-written  -lifetabl- command by Carlos Ramalheira ("findit
lifetabl") will read a schedule of rates or of numerator & midyear
denominator counts. This is a specialized area, so I  recommend that
consultation with a demographer, or at least a recent demography text.
The example on German Rodriguez's Princeton page  reproduces
calculations from  SH Preston, P Heuveline, and T Guillot  (2000)
Demography : Measuring and Modeling Population Processes,
Wiley-Blackwell, NY.

I do question whether SDs are a good summary for the variability in
age at death, In most parametric survival models, SDs are parameters
for the log of the time variable. On the original scale, SDs are
usually increasing functions of the mean.

-Steve

On Mon, Mar 8, 2010 at 12:42 PM, Steve Samuels <sjsamuels@gmail.com> wrote:
> Tony is on the right track.  This requires a life-table solution.  The
> SD for age at death in period (or "current") life tables  is that of
> the ages at death for a synthetic cohort whose age-specific death
> rates are those observed for the period, in this case 2004.  This
> calculation requires information about the numbers at risk in each age
> during the year, not just about the numbers who died at each age.
>
> I've not constructed a period life table in Stata, but see:
>
> http://data.princeton.edu/eco572/periodlt.html
> http://www.stata.com/statalist/archive/2004-02/msg00127.html
>
> Getting from these results to the SD will take some work: from the
> table probabilities estimate the mean at death m; then estimate the
> mean of (age -m)^2
>
> -Steve
>
>
> On Mon, Mar 8, 2010 at 11:57 AM, Lachenbruch, Peter
> <Peter.Lachenbruch@oregonstate.edu> wrote:
>> Am I missing something?  Doesn't the Greenwood formula give this to you, and isn't it in the ltable command?

Johannes Schoder wrote:
>> I want to calculate the standard deviation of age at death from period
>> life tables:
>> So I know how many people died at age x, where x runs from 0,1,2...110:
>> For example, in 2004 in the US people 2000 died at age 0, 1900 at age 1,
>> 600 at age 2, etc.
>


-- 
Steven Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
845-246-0774

*
*   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/