The Data

Assets	Sales	Name
620	2250	Giant
12500	1300	City
1600	6600	A&P
2550	260	FirstEmp
2800	270	BankSouth
5400	550	CalFirst
4200	17000	Kroger
5250	530	FirstTenn
3700	360	Marine
1100	3700	Stop&Shop
33000	3200	Mellon
9000	880	Norstar
1100	5100	SupGen
21000	2500	Norwest
750	2150	Turner
2500	6000	Wool

1. Start out...

... by entering this into DataDesk. Start the DataDesk program. Then create a new variable by going to the New submenu of the Data menu and selecting Blank Variable ( Data → New → Blank Variable ). Name your first variable Assets . Then create two more new variables named Sales and Company .

In order to enter data, click on the left data field. Enter a number, then tab to the second data field. Enter a number then use tab to move to the third data field. After entering a row of data, use enter to move to the next row. When finished, click on the little squares in the upper right corners to close the three data fields.

As you enter data, please keep in mind the difference between having your cursor vertical (e.g. for editing a particular entry) and horizontal for inserting new cases.

Please note you can bypass the entering of data in this exercise by downloading companies_orientation.dsk

2. Double click on the DATA

icon in the upper right corner of the screen to open the relation. Now compute the average value of assets for these companies. You do this by first selecting the icon labeled Assets and then selecting Reports from the Summaries submenu of the Calc menu ( Calc → Summaries → Reports ).

3. Now select

the entry Select Summary Statistics... of the Calculation Options submenu of the Calc menu ( Calc → Calculation Options → Select Summary Statistics ). Note that by checking and unchecking various boxes you can change the summary information which appears in the report you just calculated. (The changes won't take effect until you click OK in the dialog box.)

4. Create a plot

of assets vs. sales for this data. To do this, first select the Assets icon as y-variable. Then select the icon Sales as x-variable while holding the shift key down. Then choosing menu item Plot → Scatterplots gives the desired graph.

5. Look at the plot

you've just created. Notice the way it makes clear that there are two groups of companies with very different relationships of assets to sales. Doubleclick on the Company icon so you can inspect the names of companies. By choosing Modify → Palettes if necessary, open the Palettes menu. A rectangular box with 12 icons should appear on your desktop. Click on the box with the question mark. The information tool now lets you click on a dot in your plot window and find out which company it is. Try it a few times...

6. The lasso tool

in the upper left of the Palettes menu lets you draw a curve fully enclosing some of your data and thereby select those. Try to do this with the companies having high assets to sales ratios, and then look inside the companies variable window to see which companies this is.

7. Examine

the hyperview menu of the plot you just created. This is done by clicking on the small triangle just to the left of the plot window title. This contains a selection of commands that may go well with what you have just done. Try Add Regression Line which tries to add a line approximating the data. Note that because of the two disparate groups, this line doesn't work very well. We'll be studying this more later.

8. Sometimes data is easier

to understand after we transform it. By first selecting the Sales and Assets icons, and then choosing the Manip → Transform → Log() menu item, compute the logarithms of these two variables. Doubleclick on the two icons which result. Note that these derived variables appear as textual formulas, but that their hyperview menus include the often useful Show Numbers option to see what the actual values are. Create a plot of Log(Assets) vs. Log(Sales) and note how this is in some ways easier to read information off of than the previous plot.

9. Collect some of your results

in a layout, ( Data → New → Layout ) and add a few textual comments.