Please discuss on the list if the /weight= x principle is what we want. I have met various needs for weighting, but assume that the simple frequency weight shown here is sufficient.

Regards

Hi everybody, I think there are two different situations where a WEIGTH variable should be necesary:

The first situation is when working with data from epidemiologial surveillance. It is very usual that data sent to a upper level (for example National level from a regional level) are sent in an aggregated format. In order to be able to analyze this kind of data a command in the form of sumfreq disease /sum=COUNT (where COUNT is the weigth variable) is essential.

The second situation is when working with data comming from a complex sampling design (Not a simple random-sample strategy). In this case it is necesary to specify not only the WEIGHT variable, but the STRATA and Primary Sampling Unit. The sintax for this situation could be something like: CFREQ VAR1 /W=WEIGHTVAR /S=STRATVAR /PSU=PSUVAR

Greetings, Pedro

Pedro Arias Bohigas John Ericssons gatan, 12 5 tr 112 22 Stockholm Sweden Phone (cell)+46(0)734421130