Logistic Regression: A Self-learning Text, Third Edition (Statistics in the Health Sciences)

This estimate indicates that smokers have 1.8 times as

high a risk for getting hypertension as nonsmokers,

controlling for age, sex, cholesterol, and occupation.

30. The rare disease assumption.


31. The odds ratio is a legitimate measure of association
    and could be used even if the risk ratio cannot be
    estimated.


32. OR (OCC controlling for AGE, SEX, SMK, CHOL)d
    ¼e
    ^b
    ;whereb^¼ 0 :5309 is the coefficient of OCC in the
    fitted model
    ¼expð 0 : 5309 Þ
    ¼ 0 : 5881 ¼ 1 = 1 : 70 :
    This estimate is less than 1 and thus indicates that
    unemployed persons (OCC¼0) are 1.70 times more
    likely to develop hypertension than are employed
    persons (OCC = 1).


33. Characteristic 1: the model contains only main effect
    variables
    Characteristic 2: OCC is a (0, 1) variable.


34. The formula exp(bi) is inappropriate for estimating
    the effect of AGE controlling for the other four
    variables because AGE is being treated as a
    continuous variable in the model, whereas the
    formula is appropriate for (0, 1) variables only.

Chapter 2 True-False Questions:

1. F: OR¼exp(c)


2. F: risk¼1/[1þexp(a)]


3. T


4. T


5. T


6. T


7. T


8. F: OR¼exp(bþ 5 d)


9. The model in logit form is given as follows:

logit PðXÞ¼aþbCONþg 1 PARþg 2 NPþg 3 ASCM

þd 1 CONPARþd 2 CONNP

þd 3 CONASCM:

10. The odds ratio expression is given by
    expðbþd 1 PARþd 2 NPþd 3 ASCMÞ:

Chapter 2 669