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

Chapter 12 True-False Questions:

1. F: The outcome categories are not ordered.


2. T


3. T


4. F: There will be four estimated coefficients for each
   independent variable.


5. F: The choice of reference category will affect the
   estimates and interpretation of the model parameters.


6. Odds¼exp[a 2 þb 21 (40)þb 22 (1)þb 23 (0)þb 24 (HPT)]
   ¼exp[a 2 þ 40 b 21 þb 22 þ(HPT)b 24 ]


7. OR¼exp(b 12 )


8. OR¼exp[(5020)b 21 ]¼exp(30b 21 )


9. H 0 :b 13 ¼b 23 ¼b 33 ¼b 14 ¼b 24 ¼b 34 ¼ 0
   Test statistic:2 log likelihood of the model without
   the smoking and hypertension terms (i.e., the reduced
   model), minus2 log likelihood of the model contain-
   ing the smoking and hypertension terms (i.e., the full
   model from Question 6).
   Under the null hypothesis the test statistic follows an
   approximate chi-square distribution with six degrees
   of freedom.


10. ln

PðD¼gjXÞ

PðD¼ 0 jXÞ



¼½agþbg 1 AGEþbg 2 GENDER

þbg 3 SMOKEþbg 4 HPT

þbg 5 ðAGEGENDERÞ

þbg 6 GENDERSMOKEފ,

whereg¼1, 2, 3.

Six additional parameters are added to the interaction

model (b 15 ,b 25 ,b 35 ,b 16 ,b 26 ,b 36 ).

Chapter 13 True-False Questions:

1. T


2. T


3. F: each independent variable has one estimated coef-
   ficient


4. T


5. F: the odds ratio is invariant no matter where the cut-
   point is defined, but theoddsis not invariant


6. T


7. F: the log odds (D1) is the log odds of the outcome
   being in category 1 or 2, whereas the log odds ofD2is
   the log odds of the outcome just being in category 2.


8. T: in contrast to linear regression, the actual values,
   beyond the order of the outcome variables, have no

686 Test Answers