leading to
^pp¼
x
n
ðsample proportionÞ
- For the Poisson distribution,
Lðx;yÞ¼
Yn
i¼ 1
yxiey
xi!
leading to
yy^¼
P
xi
n
¼x ðsample meanÞ
- For the normal distribution,
Lðx;m;yÞ¼
Yn
i¼ 1
1
s
ffiffiffiffiffiffi
2 p
p exp
ðximÞ^2
2 s^2
"
leading to
mm^¼x
Matched Case–Control Studies Pair-matched case–control studies with a
binary risk factor were introduced near the end of Chapter 3. To control con-
founding factors in this design, individual diseased cases are matched one to
one to a set of controls, or disease-free persons, chosen to have similar values
for important confounding variables. The data are represented by a 22 table
(Table 4.12) whereðþ;Þdenotes (exposed, unexposed) categories. Letybe
the odds ratio associated with the exposure under investigation;n 10 was shown
to have the binomial distributionBðn;pÞ, where
TABLE 4.12
Case
Control þ
þ n 11 n 01
n 10 n 00
172 ESTIMATION OF PARAMETERS