Page 165 - Statistics II for Dummies

Chapter 8: Making Predictions by Using Logistic Regression

Which method to use to compare? 149

Sorting out similar situations
Data come in a variety of forms, and each form ✓ If you want to estimate one numerical
has its own analysis to use to make compari- variable based on another, use simple
sons. It can get difficult to decide which type of linear regression (see Chapter 4).
analysis to use when.
✓ If you want to estimate one numerical vari-
It may help to sort out some situations that able using many other numerical variables,
sound similar but have subtle differences that use multiple regression (see Chapter 5).
lead to very different analyses. You can use ✓ If you want to estimate a categorical vari-
the following list to compare these subtle, but able with two categories by using a numer-
important, differences:
ical variable, use logistic regression, which
✓ If you want to compare three or more is the focus of this chapter, of course.
groups of numerical variables, use ANOVA
(see Chapter 10). For only two groups, use a ✓ If you want to compare two categorical
t-test (see Chapters 3 and 9). variables to each other, head straight for a
Chi-square test (see Chapter 14).

All this evidence helps confirm that your model fits your data well. You can
go ahead and make predictions based on this model for the next individual
that comes up, whose outcome you don’t know (see the section “Estimating
p” earlier in this chapter).

7/23/09 9:28:37 PM
13_466469-ch08.indd 149
13_466469-ch08.indd 149 7/23/09 9:28:37 PM