Re: Create weights from multiple variables

[MR]

All,

It seems I am getting rusted. I am not able to recall how to create a weight variable based on three variables. I know that SPSS python has Rake syntax but it does not create interlocked weights. I have census and sample data by gender by age and 4 region as below. I need to create one weight variable to reflect census proportion. Unfortunately SPSS Rake would not do this  and it seems I have to do it manually but honestly I forgot the method. Can anyone help?

                     Region 1      Region 2    Region 3   Region 4

Male Age 1

Male Age 2

Male Age 3

Male Age 4

Female Age1

Female Age2

........

Mike

[Kirill Orlov]

If by "interlocked weights" you mean that you know target proportion for every cell in your Region X Age X Gender table - not just marginal proportions - then it is univariate and not multivariate weighting. You just have to create the single grouping variable with 4 X 4 X 2 = 32 categories and weight this one variable. It is simple weighting and you can calculate the weight correction by hand. Or use for example my macro "weighting groups" (rivita.ru/spssmacros_en.shtml) which performs univariate weighting. I don't know if RAKE extension command performs univariate weighting, probably it does.

[sedlacev]

You cau use also this SPSS macro:
SPSS multivariate weighting