SPSS Statistics

 View Only
  • 1.  How to extract cases used in a hierarchical linear regression?

    Posted Mon March 27, 2023 09:26 AM

    Hi,

    I ran a hierarchical linear regression with pairwise deletions using SPSS Statistics version 27.0.1.0. What I am currently trying to do is extract the participants/cases used in this regression so I can run further analyses on them (e.g., demographics). I cannot figure out how to extract these. The total used in the regression is 114 participants. Is there a way to identify these 114 participants used in this regression while looking at my entire sample size, so I can run further analyses on these specific participants?

    Thank you



    ------------------------------
    William Campo
    ------------------------------


  • 2.  RE: How to extract cases used in a hierarchical linear regression?

    Posted Mon March 27, 2023 05:35 PM

    Can I ask which procedure you ran? We don't have a hierarchical linear modeling procedure as such. Without knowing, and given that you said you used pairwise deletion, I guess I would save out predicted values or residuals into new variables and identify cases in the model that way.



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------



  • 3.  RE: How to extract cases used in a hierarchical linear regression?

    Posted Mon March 27, 2023 06:12 PM

    On SPSS, I went to analyze > regression > linear. I then added my independent variables using a total of six blocks before pressing "OK" to run the regression.

    I did try saving the predicted values to identify which were used in the regression, however after counting the cases that were saved, there were only about 80 when it should be 114. This is with pairwise deletions. Even with list wise deletions, the cases saved were fewer than the outputted total. Do you know if I am doing something wrong this way or if there is a better way to go about identifying the cases used in the regression?



    ------------------------------
    William Campo
    ------------------------------



  • 4.  RE: How to extract cases used in a hierarchical linear regression?

    Posted Mon March 27, 2023 10:42 PM

    When you use PAIRWISE, the cases used in REGRESSION largely depend on the amount and patterns of missing data within each of the 6 blocks being entered. One thing that might help understand what those look like would be to request /DESCRIPTIVES N. Look for that matrix of pairwise N's near the top of the REGRESSION output.

    Keep in mind, too, when using /SAVE, that the dependent variable is predicted (given sufficient data for X in each block entered) even if the Y value itself is missing.

    If you wish, send me the data and the REGRESSION, and I'll take a look at it and see if I can tell what might be going on. It's possible that Missing Value Analysis and/or Multiple Imputation might give some valuable information on the patterns of missing data and their effect on the REGRESSION.



    ------------------------------
    Rick Marcantonio
    Quality Assurance
    IBM
    ------------------------------