Refer to Table 3.3.

a. Fit the loglinear model of homogeneous association. Report the estimated conditional odds ratio between smoking and lung cancer. Obtain a 99% confidence interval for the true odds ratio. Interpret.

b. Test goodness of fit of the model. Interpret.

c. Consider the simpler model of conditional independence between smoking and lung cancer, given city. Compare the fit to the homogeneous association model, and interpret.

d. Fit a logit model containing effects of smoking and city on lung cancer. Use the smoking effect to estimate the conditional odds ratio between smoking and lung cancer. How does this compare to the estimate for the loglinear model?

e. Use

(b) to test the hypothesis of a common odds ratio between smoking and lung cancer for these eight studies. How does the result compare to the Breslow-Day test in Section 3.2.4?