Data taken from a study to investigate the effects of smoking on cervical cancer are stratified by the number of sexual partners (Table E1.9).
|
TABLE E1.9
|
|
|
|
|
|
Cancer
|
|
|
Number of Partners
|
Smoking
|
Yes
|
|
No
|
|
Zero or one
|
Yes
|
12
|
|
21
|
|
|
No
|
25
|
|
118
|
|
Two or more
|
Yes
|
96
|
|
142
|
|
|
No
|
92
|
|
150
|
(a) Calculate the odds ratio associated with smoking separately for the two groups, those with zero or one partner and those with two or more partners.
(b) Compare the two odds ratios in part (a). When the difference is confirmed properly, we have an effect modification.
(c) Assuming that the odds ratios for the two groups, those with zero or one partner and those with two or more partners, are equal (in other words, the number of partners is not an effect modifier), calculate the Mantel-Haenszel estimate of this common odds ratio.