Optimal Scaling by Alternating Least Squares
The combination of measurement level and number of sets that you choose in this dialog box determines which of the Optimal Scaling by Alternating Least Squares procedures you perform.
Measurement Level. You can specify the optimal scaling level for variables used in the analysis.
- All variables multiple nominal. All variables in the analysis have category quantifications that can be different for each dimension.
- Some variable(s) not multiple nominal. One or more variables in the analysis are scaled at a level other than multiple nominal. Other possible scaling levels are single nominal, ordinal, and discrete numeric.
Number of Sets of Variables. You can specify how many groups of variables are to be compared with other groups of variables.
- One set. The data contain one group of variables.
- Multiple sets. The data contain more than one group of variables. If this option is selected, Nonlinear Canonical Correlation is chosen.
Selected Analysis. The combination of choices for Measurement Level and Number of Sets of Variables yield either multiple correspondence analysis, categorical principal components analysis, or nonlinear canonical correlation analysis. The settings for each procedure are:
- Multiple Correspondence Analysis. Select All variables multiple nominal and One set.
- Categorical Principal Components Analysis. Select Some variable(s) not multiple nominal and One Set.
- Nonlinear Canonical Correlation Analysis. Select Multiple Sets.