SPSS Statistics

 View Only

  • 1.  Comparing two likertscale questions

    Posted Mon September 25, 2023 10:43 AM

    In my questionnaire I asked two statements on a Likert scale (5 point). These statements are related to each other. Statement two is answered more negatively than statement one, I want to investigate the correlation between the two. Can this be done in a cross table (chi-square + observed and expected) or is there a better test in SPSS?



    ------------------------------
    Danny Weeda
    ------------------------------


  • 2.  RE: Comparing two likertscale questions

    IBM Champion
    Posted Mon September 25, 2023 03:32 PM
    A crosstab would be a reasonable start.  Include the residuals in the cells to see where there are unusual cells.  On the Statistics subdialog, choose statistics appropriate for ordinal data.  You can also look at nonparametric correlation. 

    Graphically, you might find a fluctuation plot useful (Graphs > Fluctuation Chart).  If you don't see that on the menu, you can install it via Extensions > Extension Hub.


    --



    Original Message


  • 3.  RE: Comparing two likertscale questions

    Posted Sat September 30, 2023 04:12 AM

    Got inspired once again by randomly reading these pages. The fluctuation plot: wonderful!

    Is there a way to have the count written within the squares (or next to them, if squares too small), instead of relying on the legend only?



    ------------------------------
    Jos Blykers
    ------------------------------

    Original Message


  • 4.  RE: Comparing two likertscale questions

    Posted Sat September 30, 2023 05:49 AM

    I was too fast (and enthusiastic) - I can add data labels to the squares through the Output - Data Editor.



    ------------------------------
    Jos Blykers
    ------------------------------

    Original Message


  • 5.  RE: Comparing two likertscale questions

    IBM Champion
    Posted Sat September 30, 2023 10:09 AM
    I'm not sure what you mean, but using the Chart Editor, you can do Elements > Show Data Labels to get the counts to appear for each point.  Be careful, though, to pick the right count variable to display as there may be two of them both labelled Count.

    IMO, though, adding the count values is likely to obscure the patterns shown in the chart with just the symbols.  

    Also, in the Chart Editor, you can change the symbol for the points to any of the shapes supported in GPL.


    --



    Original Message