From Rosa Gini To statalist@hsphsun2.harvard.edu Subject st: question about standardizing Date Thu, 22 Feb 2007 18:01:05 +0100

i have some doubts about standardizing rates in stata.

i've always used -dstdize-. in fact what this command does is standardizing binomial variable, doesn't it'? variance is calculated on lines 387-391 of the dstdize.ado file (i'm reading the file from base distribution of stata 9), and the formula is that of a binomially distributed variable.

this make it suitable for standardizing *proportions*.

i've discovered recently that in stata 9 standardization is also possible for means, with the -mean- command and the -stdize, and a new command is available for proprtions, namely the -proportion- command. if i understand the command -ratio- is a generalization of the previous, in case data are aggregated.

now the question are:

1) what if i need to standardize actual rates, i.e., poisson variables?
2) what if i have a dataset for aggregated means? for example: i need to calculate the average length of stay in hospital for a given intervention, but i only have aggregated length of stay, for example

state age tot sub
al_los jects
- ----------------------------
1 65 500 50
2 65 613 57
3 65 802 75
1 75 470 40
2 75 715 37
3 75 950 60

i guess that in order to compute confidence intervals i would also need the sums of the squares of the lengths of stay (assuming length of stay normally distributed), am i correct? well if i had it, would it be possible to use a stata command to produce standardized means with confidence intervals?

thanks for any suggestion!!

rosa gini

*
* 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/