120 CHAPTER 7 Correlation, Regression, and Logistic Regression
Table 7.4
Hemoglobin Level ( H b ), Packed Cell Volume ( PCV ), Age,
and Menopausal Status for 20 Women *
Subject
number
Hb (g/dL) PCV (%) Age (years) Menopause
(0 = no, 1 = yes)1 11.1 35 20 0
2 10.7 45 22 0
3 12.4 47 25 0
4 14.0 50 28 0
5 13.1 31 28 0
6 10.5 30 31 0
7 9.6 25 32 0
8 12.5 33 35 0
9 13.5 35 38 0
10 13.9 40 40 0
11 15.1 45 45 1
12 13.9 47 49 0
13 16.2 49 54 1
14 16.3 42 55 1
15 16.8 40 57 1
16 17.1 50 60 1
17 16.6 46 62 1
18 16.9 55 63 1
19 15.7 42 65 1
20 16.5 46 67 1
- From Chernick and Friis ( 2003 , p. 286, table 12.10).
logistic regression analysis. The purpose of this data is to fi t a logistic
regression model to see if the odds of becoming anemic differ for
women under 30 years of age compared with women over 30. Female
patients with hemoglobin levels below 12 g/dL were classifi ed as anemic.
We see from the data that two out of the fi ve women under 30 were
anemic, while only 2 of the 15 women over 30 were anemic. None of
the women experiencing menopause were anemic. It is because during
menstruation, younger, nonmenopausal women have blood and hemo-
globin loss, while postmenopausal women would not. So it was hypoth-