John L. Korey,
California State Polytechnic University Pomona and
JeDon Emenhiser, Government and Politics,
Humboldt State University
© The Authors, 1998; Last modified 14 August 1998
Note: SPSS is a trademark of SPSS Inc., and Excel is a trademark of Microsoft Corp.
This set of exercises uses a file containing information about members of the California legislature and their districts. These exercises focus on some of the graphic capabilities of SPSS. For exercises using the same data that focus on statistical procedures, see Chapter 3 of the REPR instructional module on the California legislature.
The data include both the senate and the assembly. In these exercises, we will work with the assembly only. Click on "Data," on "Select Cases," on "If condition is satisfied," and on "If." In the dialog box, type "chamber = 2." Click on "Continue" and on "OK."
1. Choose a categorical variable from the data (e.g., "ETHNIC," member's ethnicity). Click on "Graphs" and on "Bar." Choose all of the defaults provided, moving the variable (such as "ETHNIC") into the "Category Axis" box. Repeat this process for "Line," "Area," and "Pareto" graphs. Do the same for a "Pie" graph, but move the variable into the "Define Slices by" box. Which graph provides the most meaningful representation of the data?
2. Repeat the above, but this time play around with some of the defaults to see how doing so changes the results. In some instances, you may need to supply additional information. For example, for stacked graphs, you will need a second variable under "Define Stacks by." (Try "PARTY.")
3. Click on "Graphs" and on "Histogram." Move the same variable you have been working with into the "Variables" box. Click on "OK." How do the results compare with those you obtained using a bar graph? Repeat the procedure, but this time use a continuous variable such as "ROLLCALL" (an index of members' voting records, scaled so that 0 represents the most liberal member and 100 the most conservative).
4. One might reasonably hypothesize that the higher the per capita income of a district ("PRCAPINC") the more conservative would be the voting record of its representative ("ROLLCALL"). Click on "Graphs," then on "Scatter." Choosing all defaults, move the dependent variable "ROLLCALL" into the "Y axis" box, and the independent variable "PRCAPINC" into the "X axis" box. Surprised by the results? Repeat the process, but this time move "PARTY" into the "Label Cases by" box, click on "Options," "Display chart with case labels," "Continue" and "OK." The results will be messy, but should give you a clue as to what is happening. Repeat the "Scatter" procedure, but this time moving "PRCAPINC" out of the "X axis" box, and moving "PARTY" out of the "Label Cases by" box and into the "X axis" box.
5. Try making additional scattergrams. For example, use Prop 209 on the X axis with each of the other propositions (210, 215, 217, and 218) on the Y axis. Which graphs slope upward, which downward? Why? Which propositions are most closely related? Compute regression and correlation statistics and compare your estimates with them.
6. After you have created a graph in SPSS, you can edit it by pointing your mouse at the graph, clicking the right mouse button, then left clicking on "SPSS Chart Object" and on "Open." Try this with some of the graphs with which we've been working. For example, with "Scatter," try going into the chart editor and clicking on "Chart," "Options," "Total," "Fit Options," "Display R-square in legend," "Continue," and "OK."
6. As you might expect, SPSS graphics capabilities have both strengths and limitations compared to other applications. Try exporting this data set in Excel format, then go into Excel, with the help of any of several good Excel manuals use some of its graphics capabilities, and compare the results to those obtained with SPSS. To export the data, click on "File" and "Save As." Next to "Save as type" choose "Excel (.xls)," type in a file name, and click on "Save."