SPSS Statistics

SPSS Statistics

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

 View Only

  • 1.  Stuart Maxwell Test of Marginal Homogeneity--Paired Samples

    Posted Tue January 03, 2023 10:34 AM
    Would someone please show me the click by click method of running this test for paired pre and post samples in SPSS version 29? It looks like the interface has changed since I last ran it. 
    Thank you. 

    --
    Megan O. Smith, MPH
    Program Evaluator
    Smith Consulting Group Services, LLC


    #SPSSStatistics


  • 2.  RE: Stuart Maxwell Test of Marginal Homogeneity--Paired Samples

    Posted Tue January 03, 2023 02:10 PM
    Analyze > Nonparametric Tests > Legacy Dialogs > 2 Related Samples...


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

    Original Message


  • 3.  RE: Stuart Maxwell Test of Marginal Homogeneity--Paired Samples

    Posted Fri September 27, 2024 09:00 AM

    Rick,

    The MH test in Nonparametrics of SPSS is not the Stuart-Maxwell test, which is a MH test for nominal data, but a MH test for ordinal or even discrete interval data (with same values for both variables, to form a square contingency table). The test that is in SPSS - is described (with a bit different, but equivalent formula to that of SPSS Algorithms gives) in Agresti "Analysis of Ordinal Categorical Data", 2nd ed., 2010, page 227. This test ("Comparing Marginal Mean Scores", per Agresti) is much close in results - as my probes show - to the paired-samples Student's T test.

    Stuart-Maxwell test is absent in SPSS, but it is simple co compute using, for convenience, MATRIX sesion.



    ------------------------------
    Kirill Orlov
    ------------------------------

    Original Message


  • 4.  RE: Stuart Maxwell Test of Marginal Homogeneity--Paired Samples

    Posted Fri September 27, 2024 10:41 AM

    Stuart-Maxwell marginal homogeneity test for nominal data. Use this SPSS syntax.

    matrix.

    comp tab=
        {1520,  266,  124,  66;
          234, 1512,  432,  78;
          117,  362, 1772, 205;
         36,   82,  179, 492}. /*a k x k frequency (contigency) table (data example taken from Stuart, 1955)

    comp k= ncol(tab).
    comp rt= rsum(tab).
    comp ct= csum(tab).

    comp d= (rt-t(ct))(1:(k-1)).
    *print d.

    comp S= -(tab(1:(k-1),1:(k-1))+t(tab(1:(k-1),1:(k-1)))).
    loop i= 1 to k-1.
     comp S(i,i)= rt(i)+ct(i)-2*tab(i,i).
    end loop.
    *print S.

    comp chi2= t(d)*inv(S)*d.
    print chi2.
    *print sqrt(chi2).
    comp sig= 1-chicdf(chi2,k-1).
    print sig.
    end matrix.



    ------------------------------
    Kirill Orlov
    ------------------------------

    Original Message


Related Content