BH Statistics

October 28, 1999

- Create a derived variable for each original quantitative variable so that
all the variables have values on similar scales (see Mary's example -- it
makes each value a percentage of the largest value). Note that Housing and
Crime need to be adjusted so that, in their derived variables, larger means
better. As they are now defined, larger means worse.

- Decide on weights for each variable. One way of weighting
variables is to pick a weight for one variable and make all the
other weights relative to it. For example, if you want Crime to be
twice as important as Housing, then make Crime's weight twice as
large as Housing's weight. Then make all the other variables'
weights some multiple or fraction of Housing's weight.

- Explain why you weighted the variables the way you did.

*Write this to be handed in. Include each variable's weiht and why you gave it that weight relative to the others. Don't worry about being "right." We just want to understand how carefully you've thought about this.*

- Create a derived variable that gives each city a total
weighted score. A total weighted score is the sum of each
variable's value times its respective weight.

- Evaluate your derived total score variable (
*Evaluate Derived Variable*in the*Data*menu.) Select*City name*and your derived total score variable. Then select*Make Variable Table*in the*Manipulate*menu.

What are the top 5 cities according to your weighting system?

Please save this work on your floppy disk and bring your disk to
class. The easiest way to do this is to open the *Places Rating*
data from your floppy disk. Then, when you select File>Save, your
work will be saved on your floppy.