The file ARCHAIC2.SYD contains data on assemblage composition from 10 Early Archaic sites on the Atlantic slope of the eastern U.S., the same data used in the cluster-analysis exercise. However, in this case, the dataset has been re-configured into a form more suitable for correspondence analysis. The variables in the file are the following: SITE$, site name; TOOL$, tool type; and N, the count of that tool type at that site. If you think of ARCHAIC.SYD (the cluster-analysis dataset) as a spreadsheet, then ARCHAIC2.SYD (the correspondence-analysis dataset) is the same data reconfigured so that each cell in the original spreadsheet now appears as a separate row.

In order to do a correspondence analysis, you must first tell SYSTAT that N is a “frequency variable” (Data | Frequency), i.e., that N represents the frequency (count) with which each particular combination of SITE$ and TOOL$ occurs.

Once you’ve done this, carry out a correspondence analysis of these data (Statistics | Data reduction | Correspondence analysis). The two variables to use in this analysis are SITE$ and TOOL$. In SYSTAT’s terminology, the variable you specify as the “dependent variable” at the outset becomes the “row variable” in the output listing; similarly, the variable you specify as the “independent variable” at the outset becomes the “column variable” in the output listing.

• Interpret the pattern you see in the correspondence plot. (Note: it may help you to make scatter plots of the sites and tools separately in the form of a biplot. This will decrease the overlap and visual “clutter” that occurs when sites and tools are plotted on the same axes. Doing so requires that you save the output of the correspondence analysis in a separate file, then use these output data to generate the biplot. A command file that shows how to do this, ARCHAIC2.SYC, can be found on the course web site.)
• Next, interpret the numbers you see in the printed output for both rows and columns. Discuss how well the correspondence analysis model fits the data by considering the values of mass, quality, and inertia. Also discuss how the original variables relate to the factors (and vice versa) by by looking at the values of the Contributions and squared correlations.
• How do the results of your correspondence analysis compare to those of your cluster analyses of the same data in a previous exercise?

Datasets and command files for this exercise (right-click to download):

• Archaic2.syd
• Archaic2.syc

Back