CS103-Sp01 day 19

CS 103 Day 19 Monday May 7, 2001

Class notes: Get this page as a handout
Chapter 4, Graphing

I don't think SPSS charts are as bad as Cronk does. They are not quite as flexible as Excel's, but they do some things easily that Excel will not. He's left out much of the good stuff.

Learning SPSS:

The file Height.sav is already in SPSS Stuff. String (text, alphabetic) variables are possible in SPSS, and are useful especially for labeling cases in scatterplots. See Problem A for how to enter.

You have done 4.1, pp. 31-35 Bar, pie, Histograms. See Asst. 14 problem B for how to do Boxplots.
4.2, pp 35-37 Scatterplot. Go through the text example, Set markers by "Sex", and (Not in text) also Label Cases by "Name"
Editing a Chart: turn to p. 31, and follow the directions there to open the Chart Editor. (or double-click the chart)

Chart Editor: Another window! You change the looks of a chart in the Chart Editor .

We have windows simultaneously for the Data Editor, the Output Viewer, the Chart Editor, and possibly Help, all with buttons on the bottom taskbar.

When the Chart Editor is running, the chart in the output window will be grayed over.
Close the Chart Editor (File>Close) when you have the chart the way you want.

Changing Chart Labels, Axis ranges, etc: Chart menu. Investigate. And/or Double click a chart element to get a menu about it.
Buttons: (a couple) in the toolbars

All the symbols are the same shape, which will not distinguish gender when printed in black and white. Click on a Male point , then click on the Marker button (it shows an *). In the Markers menu, choose triangle, or some other shape, and Apply. (Then Close the Markers menu) (The Format menu gives you the Markers too.)
Point labels: Usually we don't want all the points labeled; only some interesting ones. To see any individual point label, click the Point ID button (far left bottom row of buttons--the square--hovering with the mouse on any button will give a written description). Your cursor will change to a decorated square. Click a point in the graph. It will be labeled with the "Name". Click again to unlabel. (You can label as many at once as you can click on)

Label any points that lie outside the "herd". (To stop labeling points and get your usual cursor back, click the Point ID button again.)

Fit Line(s)="trendline", "regression line" Chart/Options, Fit Line section.

You can fit separate straight lines to the males and females by checking the Subgroups box. Do that. You can fit a line to the whole data set at the same time, by checking the Total box as well. Do that too.
(Do the Fit options button to make sure Linear Regression is selected)
(To get the equation requires doing the Simple Regression procedure, getting a whole different set of output. We won't. Cronk p.43-5 tells what to do and how to read it.)

Line style: Fit lines come up in different colors, unsuitable for distinguishing in black and white. Click on a line, or on the line in the legend, till the line in the legend shows barbell ends. Get the line style menu from Format>Line style, or from the --- (dotted line) button.
End of Chart Editor

Clustered bar chart for means, 4.3 See problem D.

Get this page as a handout

Assignment 14 Due Friday, Day 21, May 12

A, B, C all use Height.sav.

A. String variable, Text file. Add to Height.sav a new variable "Name"; in the Variable View window, New stuff: set the Type as String. (another word for alphabetical). Give the 16 people short names (A, B, C,...O, P will do) Save as Height-names.sav. New stuff: Now "Save as" a Tab-delimited (.dat) file. (This is an ASCII, text, file, with tabs between the items.) Open Notepad (Start menu>Accessories), find and open your file, (you'll need the type "All files" to find it), and print it to hand in.

B. New stuff: Boxplots (side by side). ( Before doing this, Add a case to Height, a 6-foot (72 inch) 140 pound female "Q".)

Explore: Plot:

Boxplots

New stuff:

Hand in.

C. Scatterplot Redo the scatterplot with "height" on the x-axis and "weight" on the y-axis. As before, Set markers by "Sex", and Label Cases by "Name" (The predictor or explanatory variable should go on the x-axis, the dependent on the y-axis. Height can cause weight, but not usually the other way around.) Now open the Chart Editor. (See notes)

a) Add your name to your chart in a footnote (Chart/Footnote menu).

b) Label any points that stand away from the herd.

c) Add Fit lines to the males and females separately.

d) Change the Markers and one of the lines so they can be distinguished in black and white printing.

Close the Chart Editor and print the chart to Hand in.

D. Clustered bar chart for means: Follow along with pp. 37-8, "Advanced Bar Charts"
The data is shown and explained on p. 59. I have entered it and it is in
Class Material\CS103\SPSS Stuff\GRADES.SAV.
Hand in the graph shown on p. 38. Answer this: Which instructor seems to have done the best job?

To Sievers Home Page

CS103-Sp01/day19.htm 5/3/01

This page belongs to Sally Sievers who is solely responsible for its content. Please see our statement of responsibility.