SPSS Statistics

 View Only
  • 1.  PLUM Procedure not Producing Output

    Posted 2 days ago

    I have SPSS Advanced Statistics and am running an ordinal logistic regression analysis. I can easily run multiple tests and immediately get an Output. However, when I run the Ordinal logistic regression with PLUM (Analyze > Regression > Ordinal), I can wait hours and an output never appears. On the status bar, it shows "Running PLUM..."

    I've done some basic troubleshooting and have tried some different view settings and options but cannot get this test to produce in the output. Does this test just take forever to run or are there other ideas for troubleshooting to get the output?

    Thank you!



    ------------------------------
    Bethany
    ------------------------------


  • 2.  RE: PLUM Procedure not Producing Output

    Posted 2 days ago

    It sounds like you're experiencing a frustrating issue with SPSS when running an ordinal logistic regression using the PLUM procedure. While the problem might be related to the complexity of your data or the specific model you're trying to fit, here are some steps you can take to troubleshoot and potentially resolve the issue. Additionally, you might find it helpful to use NS WhatsApp to quickly reach out to colleagues or professional groups who have experience with SPSS for real-time advice and feedback. This can often provide practical solutions or alternative methods to address the problem effectively. Here are some troubleshooting steps:

    1. Check Data Size and Complexity: Ordinal logistic regression can be computationally intensive, especially with large datasets or complex models. Ensure your dataset isn't excessively large and consider simplifying your model to see if that helps.

    2. Reduce Model Complexity: Try reducing the number of predictor variables in your model to see if it runs faster. If it does, you can gradually add predictors back to identify which ones might be causing the slowdown.

    3. Examine Data for Issues: Ensure your data is clean and doesn't contain any anomalies like missing values, outliers, or extremely skewed distributions that might be causing the PLUM procedure to struggle.

    4. Increase System Resources: If possible, run the analysis on a computer with more processing power and memory. SPSS can be resource-intensive, and more powerful hardware might help.

    5. Update SPSS: Make sure you are using the latest version of SPSS. Sometimes, software updates include performance improvements and bug fixes that could resolve your issue.

    6. Check for SPSS Settings: Verify your SPSS settings. Sometimes, changing the memory allocation settings in SPSS can help with performance issues. Go to Edit > Options > Data and increase the memory allocation if needed.

    7. Run Analysis in Smaller Batches: If your dataset is very large, consider running the analysis on smaller subsets of the data to see if it completes. This can help determine if the size of the dataset is the issue.

    8. Consult Documentation and Support: Refer to the SPSS documentation and user forums for any known issues with the PLUM procedure. IBM SPSS Support might also provide insights or patches for specific problems.



    ------------------------------
    Alex Hales
    ------------------------------



  • 3.  RE: PLUM Procedure not Producing Output

    IBM Champion
    Posted 2 days ago
    Without knowing more about the data and model specifications, I suspect that the problem with PLUM speed is due to bootstrapping.  For some reason, that slows the procedure down greatly.  Of course, bootstrapping takes a significant amount of time for any procedure, but the PLUM slowdown is really exceptional.

    So, assuming that you have this on, turn it off to see whether you get good performance.  PLUM without the bootstrap isn't much slower than regular regression on moderate-sized problems.

    --





  • 4.  RE: PLUM Procedure not Producing Output

    Posted 16 hours ago

    Alex, copyleaks.com flags your reply as AI-generated content.  I have no objection to AI-generated content per se.  But I do object to the posting of AI-generated content without clearly labeling it as such.  IMO, that is just another form of academic honesty.  YMMV.

    PS- I know that detectors of AI-generated content do not have perfect sensitivity and specificity.  And I know that my own "spidey senses" are not always correct.  So if the reply you posted was your own original work, I humbly apologize.



    ------------------------------
    Bruce Weaver
    ------------------------------



  • 5.  RE: PLUM Procedure not Producing Output

    IBM Champion
    Posted 15 hours ago
    I fed the original post to ChatGPT3, and it responded with exactly the rather unhelpful text that was posted.

    --