To study factors that affect the recurrence of heart attacks
(HA), an investigator collected data from 20 HA victims. The
investigator fits a logistic regression
model with an indicator of a second HA within one year (1 = HA; 0 =
no HA) as the binary outcome. There are two predictors: change to
recurrent stroke
x1 = 1 if the patient completed an anger management program;
0
otherwise x2 = anxiety score (0 = low anxiety, 100=high
anxiety)
The computer output is given in Table 3 below:
(i) For patients that have completed the anger management
program, justify if
high anxiety associated with an increased probability of a second
heart
attack.
(ii) Determine if there is a statistical evidence that shows an
anger
management program is associated with a reduction in the
probability of a
second heart attack.
(iii) Justify why linear regression is inappropriate for
modelling the probability
of a second heart attack.