Discussion:
Q: An investigator wishes to study the effect of cigarette smoking on the development of myocardial infarction (MI) in women. A sample of 2000 disease-free current smoking women and 100 disease-free ex-smoking women, aged 50-59, are identified in 1996, and the incidence of MI between 1996 and 1998 is noted at follow-up visits 2 years later. Investigators find that 40 currently smoking women and 10 ex-smoking women have developed the disease.
a. Is a one-sample or a two-sample test needed here?
b. Is a one-sided or a two-sided needed here?
c. What is the hypothesis?
d. Which of the following test procedure should be used to test this hypothesis?
i. Χ2 test for 2x2 contingency tables
ii. Fisher's exact test
iii. McNemar's test
iv. One sample binomial test
v. One sample t test
vi. Two sample t test with equal variance
e. Carry out the test procedure(s) mentioned in (d) and report the p-value.
f. Compute the odds ratio and its 95% confidence interval.
g. Compare the results from (e) and (f). Are they consistent? What is your final conclusion for the investigator?