SPSS Statistics

Multiple correspondence analysis and Cluster Analysis

  • 1.  Multiple correspondence analysis and Cluster Analysis

    Posted Fri September 25, 2020 12:52 PM
      |   view attached
    Goodmorning everyone,

    I have some problems in linking the Multiple correspondence analysis object scores and the cluster analysis.
    I have 15 dummy variables measuring Performance Management System characteristic used by 111 firms. Once entered in the software I discretize the variables in two categories (for replicating the original dummy variables), hence i plot the variables in the joint category plot and I save the onject scores. All variables have discrimination measures which are above 0.150, no-one is discarded from the analysis.

    In the joint category plot I can identify 4 different categories of Performance Management System, which are coherent with the literature, therefore the variables display in the two dimension in a correct way.

    Moreover, when i plot the object scores of the two dimension in the hierarchical cluster analysis (ward method), the suggested number of clusters is also 4, hece it justifies my four categories of PM.

    However, when I make the cross tabulation of my original 15 dummy variables with the cluster membership variable, they displays differently with respect  to the categories of PMS. Hence I have clusters that are not coherent with the 4 categories identified and variables are mixed up among all clusters.

    Do I do something wrong? Is there a passage that I have skipped?
    Thak you so much for any answer in advance.
    I attach the four categories of SPM

    Filippo Ferrarini