Controlling for Age in SPSS

Discussion in 'SPSS' started by Angela Sberna, Sep 22, 2011.

  1. I have 2 variables - EXCLBF (a dichotomous variable) and DERS (a
    continuous variable). I want to compare DERS means between those who
    answered yes (1) to EXCLBF and those who answered no (0) to EXCLBF.
    This can be accomplished through an independent t-test; however, I
    also want to control for a confounding variable, AGE. How do I control
    for AGE in SPSS? What test do I need to run?
     
    Angela Sberna, Sep 22, 2011
    #1
    1. Advertisements

  2. Angela Sberna

    Art Kendall Guest

    I would start with a scatterplot with DERS of the vertical axis and AGE
    on the horizontal axis with EXCLBF as colors for the markers.

    Then in the output window edit the scatterplot. See how well separate
    linear and loess fits look.
    If the loess fits are not glaringly different from the linear fits do a
    2 group ancova.
    UNIANOVA DERS by EXCLBF with AGE ...

    To give more refined advice we would need to know the meanings of your
    variables and whether there is reason to think that the relation of AGE
    to DERS would differ for the two groups so that you need a more complex
    model.

    Art Kendall
    Social Research Consultants
     
    Art Kendall, Sep 22, 2011
    #2
    1. Advertisements

  3. Angela Sberna

    Bruce Weaver Guest

    Great advice about starting with a plot.
    Here Art is hinting at something that should perhaps be said a bit more
    explicitly. The basic ANCOVA model he shows above imposes a linear
    relationship between Age and DERS, and it *forces* the slopes for the
    two groups to be equal (i.e., parallel lines). So if the plot suggests
    that the relationship is not linear, or that the lines are not parallel,
    then you need to consider a more complex model--e.g., a model that
    allows the two groups to have different slopes, or a model that allows
    the relationship to be curvilinear rather than linear, etc.

    HTH.
     
    Bruce Weaver, Sep 22, 2011
    #3
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.