1. The dataset ELEMENTS.SYD contains neutron activation data on shell-tempered sherds from two Mississippian sites in Tennessee: Bussell Island (at the mouth of the Little Tennessee River) and Mound Bottom (in the Nashville Basin). The concentrations of 10 elements are provided in parts per million (ppm). Your mission is to analyze the differences in average composition between the two sites. Please do (at least) the following in SYSTAT. (If you run into trouble, you can look at the command file Ex03a.SYC for hints on how to proceed; note that the TRANSPOSE command is very useful in this instance.)
a. Construct and juxtapose dot plots showing the average composition at each site. Also try a multiway dot plot, superposing different plotting symbols on a single set of axes. Use a linear scale for concentration.
b. Now construct and juxtapose dot plots using a logarithmic scale. Is the log or the linear scale more appropriate for the comparison at hand, and why?
c. Compare the two sites by means of a Tukey mean-difference graph.
2. The file KNOLL.SYD contains data from Indian Knoll, a Late Archaic site in Kentucky. The following variables are coded: LEVEL, the designation of a 6-inch level, numbered (negatively) in sequence from the top to bottom; BURIALS, the percentage of all burials that were found in that level; BEADS, the percentage of all disc shell beads that were found in that level (associated with burials); and CONCHS, the percentage of all large, conch-shell artifacts that were found in that level (associated with burials). These data were extracted from Figures 1 and 2 in Howard Winters's article “Value Systems and Trade Cycles in the Late Archaic” (in New Perspectives in Archaeology, edited by Binford and Binford, 1968). (Although the percentages for each variable should theoretically add up to 100%, they don't in this case because of the errors in estimating the numbers from Winters's graphs. Don't worry about these errors; they make little difference for present purposes.) Winters uses these data to argue that the availability of shell ornaments fluctuated cyclically through time. Your job is to explore and re-evaluate his data by means of the time-series and smoothing techniques we have discussed.
a. Graphically illustrate how each of these variables behaves through time. (You can best do this with the PLOT command, using the LINE option to “connect the dots” with a line.)
b. Smooth each of the variables with the “4253H, twice” algorithm and graphically show your results. What's going on in these data? Do you see cycles like Winters did? [Hint: Smoothing may be accomplished by means of a SYSTAT command file called 4253H.SYC, which can be found on the course web site. Just open this file in the SYSTAT command pane (File | Open), run it (File | Submit Window), and follow the prompts. This will create a dataset containing variables named SMOOTH and ROUGH, which you then should examine graphically. In order for this command file to work properly, select Edit | Options, click on the “General” tab, and make sure that the box marked “Perform substitutions specified by TOKEN commands” is turned ON. To avoid needless output, you might also want to click on the “Output” tab and turn OFF the box marked “Display statistical quickgraphs.”]
c. Print the file 4253H.SYC and, by annotating
the printout, briefly explain how it works. (Hint: this is a command
file that uses automatic token substitution.)
Datasets and command files for this exercise (right-click to download, then delete the ".txt" suffix):