Basic Analysis and Graphing - SAS

Chapter 6 Performing Contingency Analysis 195
Correspondence Analysis
Table 6.5 Descriptions of the Analysis of Means for Proportions Options (Continued)
Display Options
Shows or hides the decision limits, decision limit shading, center line,
and needles.
Related Information
• “Example of Analysis of Means for Proportions” on page 201
Correspondence Analysis
Correspondence analysis is a graphical technique to show which rows or columns of a frequency table have
similar patterns of counts. In the correspondence analysis plot, there is a point for each row and for each
column. Use Correspondence Analysis when you have many levels, making it difficult to derive useful
information from the mosaic plot.
Understanding Correspondence Analysis Plots
The row profile can be defined as the set of rowwise rates, or in other words, the counts in a row divided by
the total count for that row. If two rows have very similar row profiles, their points in the correspondence
analysis plot are close together. Squared distances between row points are approximately proportional to
Chi-square distances that test the homogeneity between the pair of rows.
Column and row profiles are alike because the problem is defined symmetrically. The distance between a
row point and a column point has no meaning. However, the directions of columns and rows from the
origin are meaningful, and the relationships help interpret the plot.
Correspondence Analysis Options
Use the options in the red triangle menu next to Correspondence Analysis to produce a 3-D scatterplot and
add column properties to the data table.
Table 6.6 Descriptions of the Correspondence Analysis Options
3D Correspondence Analysis
Save Value Ordering
Produces a 3-D scatterplot.
Takes the order of the levels sorted by the first correspondence score
coefficient and makes a column property for both the X and Y
columns.
The Details Report
The Details report contains statistical information about the correspondence analysis and shows the values
used in the plot.