CHAPTER 18 A Yes-or-No Proposition: Logistic Regression 251

Understanding the Basics of

Logistic Regression

In this section, we explain the concepts underlying logistic regression using an

example from a fictitious animal study involving data on mortality due to radia-

tion exposure. This example illustrates why straight-line regression wouldn’t

work and why you have to use logistic regression instead.

Gathering and graphing your data

As in the other chapters in Part 5, we present a real-world problem here. This

example examines the lethality of exposure to gamma-ray radiation when given

in acute, large doses. It is already known that gamma-ray radiation is deadly in

large-enough doses, so this animal study is focused only at the short-term lethal-

ity of acute large doses. Table 18-1 presents data on 30 animals in two columns.

TABLE 18-1

Radiation Dose and Survival Data for 30 Animals,

Sorted Ascending by Dose Level

Dose in REMs

Outcome 0

Lived; 1

Died

Dose in REMS

Outcome 0

Lived; 1

Died

0

0

433

0

10

0

457

1

31

0

559

1

82

0

560

1

92

0

604

1

107

0

632

0

142

0

686

1

173

0

691

1

175

0

702

1

232

0

705

1

266

0

774

1

299

0

853

1

303

1

879

1

326

0

915

1

404

1

977

1