S.E. for unweighted mean in complex surveys

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

RonVega
Posts: 4
Joined: Tue Jul 03, 2012 1:47 pm

S.E. for unweighted mean in complex surveys

Postby RonVega » Tue Jul 03, 2012 2:32 pm

Hi everyone,

I have a complex survey design where counties are part of a state. I know the population and stratification and am able to calculate the mean and standard error for each of my subpopulations using the syntax in SPSS 20:

CSDESCRIPTIVES
/PLAN FILE='C:\Users\documents\complex.csaplan'
/SUMMARY VARIABLES=satisfaction
/SUBPOP TABLE=GroupNumber_Recode DISPLAY=LAYERED
/MEAN
/STATISTICS SE COUNT
/MISSING SCOPE=ANALYSIS CLASSMISSING=EXCLUDE.

My question arises when I want to calculate a mean for the total sample. If i just use the complex survey descriptive for the entire sample, this mean is weighted by the sample size of the county. I want to calculate a total mean where all the means for each county are equally weighted and get a standard error for this value. Does anyone have suggestions for how to accomplish this?

To clarify, if one county has 1000 people with mean satisfaction 3.0 and the next has a 10 people with mean satisfaction 5.0. If I just run a "mean" the result would be close to 3. I want an equally weighted by county so the mean would be 4.0 and find a standard error for this value.

Any help would be greatly appreciated.

Thanks,
Ron

Who is online

Users browsing this forum: No registered users and 1 guest

cron