Multiple Correspondence Analysis (MCA): Table Specifications Startup Panel - Multiple Correspondence Analysis (MCA) Tab

You can select the Multiple Correspondence Analysis (MCA) tab of the Multiple Correspondence Analysis (MCA): Table Specifications Startup Panel to access the options described here. Use these options to specify the input table for the multiple correspondence analysis.

Option Description
Input

Three data file format options are available in the Input group box:

Depending on the Input method selected, one or more of the following option buttons are available. You can tabulate variables with codes, or input a (stacked) table of frequencies with or without coding variables.

Variables (Factors in Burt Table) Displays the standard variable selection dialog box. You can select the variables that can be used in the analysis. The Burt table is computed for the categories of the selected variables.
Codes for grouping variables Displays the Select Codes for Coding Variables dialog box. You can use this dialog box to enter the codes (numbers or text labels) that define the categories for the selected variables.

The Codes for grouping variables button is only available after the Variables (factors in the Burt table) are selected.

Supplementary row id variable If the Raw data (requires tabulation) input method has been selected, you can click the Supplementary row id variable button to display the standard variable selection dialog box. Designate a variable that identifies cases in the active data set to be analyzed as supplementary rows.

The cases selected as supplementary rows are not used for the computation of eigenvalues and eigenvectors (see Computational Details), but coordinate values and corresponding statistics are computed for those rows.

Supplementary columns (variables) Displays the standard variable selection dialog box. Designate the variables to be analyzed as supplementary columns (variables). Unlike in simple correspondence analysis, where supplementary columns and rows can be added from the Correspondence Analysis Results dialog box - Supplementary points tab, in multiple correspondence analysis the supplementary columns must also define a valid Burt table.
  • In this case, specify all variables for the analysis using the Variables (Factors in Burt Table) button.
  • Select the subset of those variables that are to be treated as supplementary columns using the Supplementary columns (variables) button.

The variables selected as supplementary columns are not used for the computation of eigenvalues and eigenvectors (see Computational Details), but coordinate values are computed for those columns and reported in the spreadsheet and plots of coordinates.

Code for supp. rows You can enter the code or text label for the Supplementary row id variable that uniquely identifies the supplementary rows in the data set.
Variable with frequencies/counts Displays the standard variable selection dialog box. Select the appropriate variable that can be used in the analysis (the variable containing the measure of correspondence, similarity, confusion, association). Only positive values or zero are allowed in this variable (example, Statistica does not permit negative frequencies).
Variables with frequencies Displays the standard variable selection dialog box. Select the appropriate variables that can be used in the analysis.

The data in the selected variables (and cases) are expected to define a valid Burt table.

The Variables with frequencies button is only available when the Frequencies w/out grouping vars option is selected in the Input group box.

Specify structure of table Displays the Specify the dimensions of the table dialog box. Enter the factor names (such as survival, age, location) and the number of levels for each factor (such as 2, 3, and 3). When processing the data, Statistica automatically checks whether the respective data specifies a valid Burt table.

The Specify structure of table button is only available after Variables have been selected.

Compute all dimensions You can select this check box to compute all of the eigenvectors and eigenvalues associated with the current Burt table. If this check box is not selected, specify the Maximum number of dimensions to extract.
Maximum number of dimensions For large Burt tables, extracting all dimensions might not be useful and/or efficient. For many applications only a few dimensions might be required. You can specify the number of dimensions to extract here.

For examples and more detailed information on specifying tables for Multiple Correspondence Analysis, see Raw Data (Requires Tabulation), Frequencies with Grouping Variables, and Frequencies without Grouping Variables. See also, Multiple Correspondence Analysis.