Re: st: using several years of survey data

[Steven Samuels <sjsamuels@gmail.com>]

Maria, ordinarily I would say that your set up is correct.  However, sometimes in this kind of study (I'm not familiar with yours), the population is randomly divided into five groups of about equal size; then one group a year is surveyed. In 10 years, each group would appear twice. In this case, I don't know if the standard sampling formulas apply. I suggest that you contact the study statisticians for some guidance. 

Steve
sjsamuels@gmail.com


On Mar 1, 2011, at 4:22 PM, Maria E. Montez Rath wrote:

Hello,

I'm using 10 years of the National Inpatient Sample (NIS) and have a
question on how to define the strata.

The NIS is a stratified, single-stage cluster sample of hospitals.
Every year, about 20% of US community hospitals are sampled. Sixty
strata are defined depending on things like geographical region and
teaching status, and from each a random sample of hospitals is taken.
For each hospital all discharges are then selected.

The data provided includes the strata and weights. I set the data by

svyset hospid [pweight=discwt], strata (nis_stratum)

When using several years, shouldn't I create a new strata variable
that incorporates the year so that the strata is different per year?

gen str yearstratum=string(year,"%04.0f") + string( nis_stratum,"%04.0f")

svyset hospid [pweight=discwt], strata (yearstratum)

I think this is necessary in order to obtain the correct standard errors.

Can anyone correct me if I'm wrong or confirm that what I'm doing is right?

Thank you,

Maria

*
*   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/


*
*   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/