SPSS Statistics

SPSS Statistics

Your hub for statistical analysis, data management, and data documentation. Connect, learn, and share with your peers! 

 View Only

  • 1.  Complex Sample Analysis Preparation Wizard

    Posted Mon November 11, 2024 12:09 PM
    Edited by Ian Mon November 11, 2024 12:47 PM
    Hi I'd really appreciate some advice setting up the following scenario, doesn't seem that complicated but I must be doing something wrong.
    I have a data set with 447 cases.
    My weights are intended to correct for under/over representation across seven regions compared to the actual population. The overall sample rate is about 5%. It's a finite population of about 10,000. Each of the 447 cases is given one of seven weights, which I calculated by dividing the share of the  population by the share of the sample. The range of weights is about 0.3 to 2.2.
    I'm then analysing the data using SPSS version 30.0, Complex Samples module.
    I'm not sure I've set it up correctly in the Analysis Preparation Wizard.
    I've tried a few different things
    1) just set the weights as the weights variable (only estimation method option then is With Replacement)
    2) also set the strata as Hubs
    3) changed the estimation method to Without Replacement
    4) you can then add the Hub population numbers
    But the p values look really big, much bigger than I was expecting the design effect to be.


  • 2.  RE: Complex Sample Analysis Preparation Wizard
    Best Answer

    Posted Mon November 11, 2024 12:37 PM
    Edited by Lorraine Rizzuto Wed November 13, 2024 02:48 PM

    Update: I seem to have worked this out - when I tried weights which gross (scale up) to the population as well as re-weight to the population proportions the p values and design effect are much more in line with my own estimates. Not sure why that would be be the case but seems to have been the issue.



    Original Message


Related Content