SPSS Statistics

SPSS Statistics

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

 View Only
  • 1.  Dummy variables in linear regression

    Posted Fri October 13, 2023 11:06 AM

    Hello,

    in SPSS 29.0.0.0 I made dummies for a categorical variable with 3 levels using: transform -> Make dummy variables. i got 3 dummies, i included 2 in the regression model leaving the 3d as reference. at the same time i watched tutorials to make these dummies in a different way, more manual, and they said that i need to make only 2 dummies to include in regression model. the question is: is my way correct?

    other question: if i need to make interaction term in regression model with dummies, should i make interaction term for the reference dummies too?

    thanks in advance



    ------------------------------
    Judi Abouraied
    ------------------------------


  • 2.  RE: Dummy variables in linear regression

    Posted Fri October 13, 2023 11:28 AM
    What you did with CREATE DUMMIES is perfectly fine.  It is equivalent to the more manual process, since you only used two of them in the regression.

    As for interactions, you can create the two-way interactions and test them in the regression if you think such interactions would exist in your model.  You would make them a separate block in the regression in order to get a test statistic for the whole set.

    --





  • 3.  RE: Dummy variables in linear regression

    Posted Mon October 16, 2023 03:14 AM

    Thank You very much.



    ------------------------------
    Judi Abouraied
    ------------------------------



  • 4.  RE: Dummy variables in linear regression

    Posted Mon October 16, 2023 03:44 AM

    So, I have variable 1 with 3 dummies (1 (reference),2, 3), two dummies included in the regression and variable 2 with 2 dummies (A (reference), B) one included in the regression) both are independent. the dependent variable is continuous. I think there is interaction between variables 1 and 2. Could you kindly show me what my interaction term looks like?

    I would include in the regression:

    • dummies 2
    • dummy 3 
    • dummy  B
    1. dummy B x dummy 2
    2. dummy B x dummy 3

    and should I use a two-way interaction or is this model enough? if yes, would you give me ahint how?



    ------------------------------
    Judi Abouraied
    ------------------------------



  • 5.  RE: Dummy variables in linear regression

    Posted Mon October 16, 2023 10:13 AM
    I suggest that you switch to Analyze > General Linear Model > Univariate, where you can just treat those variables as factors and specify any desired interactions at the factor level in the Model subdialog.  The results would be the same as you would get from regression if you spelled it all out, but it is easier to specify the model this way and get all the necessary test statistics automatically.

    --