We are in the process of planning a national survey of businesses. This will use a stratified sample with one level of stratification (by state or territory). We anticipate a 20 - 25% response rate and expect to need to adjust the weights for non-response. We only propose to produce descriptive statistics from the data for reporting i.e. means and proportions. Can the complex samples module in SPSS correctly calculate the variances using adjusted weights? I have read the manual in depth and its a bit ambiguous on this point. What I would like to be able to produce are means/ proportions with associated RSEs and 05% confidence intervals.
Hoping someone on the forum can advise.
What you are asking for is post-sampling adjustments. The CS procedures don't provide for that. You could use the SPSSINC RAKE extension command to adjust the weights to control totals, but the CS calculations should then be considered as approximate. There are some other options we could discuss after March 6. If you want to follow up, you can contact me at jkpeck AT gmail DOT com.
Licensing problem SPSS Trial Version 6 Answers
Test of Significance 0 Answers