Data Set 1 Data set name: evans.dat Variable 1 = number of coronary heart disease cases-CHD 2 = total number at risk in stratum 17 274 15 122 7 59 5 32 1 8 9 39 3 17 14 58 3 = serum catecholamine level (0 = low, 1 = high)-CAT 4 = age group (0 = below 55, 1 = 55+ )-AGE 5 = electrocardiogram abnormality (0 = normal, 1 = abnormal)ECG 0 0 0 0 1 0 0 0 1 0 1 1 0 0 1 1 0 1 0 1 1 1 1 SampIe Questions A. Fit a logistic model that predicts CHD from the predictors CAT, AGE, andECG.

B. Fit a logistic model that predicts CHD from the predictors, CAT, AGE, ECG, CAT x AGE, and CAT x ECG.

C. Determine the estimated odds ratio for the effect of CAT adjusted for AGE and ECG based on the model fit in Question A.

D. Determine the formula for the estimated odds ratio for the effect of CAT adjusted for AGE and ECG based on the model fit in Question B.

Use this odds ratio formula to derive individual odds ratio estimates for various choices of AGE and ECG.

E. Fit the same model as used in Question A, but code the CAT variable as -1 =low, + 1 =high. Compare your results in Questions A and C with the corresponding results using the new coding.

F. Test for interaction using a likelihood ratio test.

G. Test for the significance of CAT adjusting for AGE and ECG using both Wald and LR tests.

H. Obtain 95% confidence intervals for the effect of CAT adjusted for AGE and ECG.

1. Use a hierarchical backward elimination strategy to obtain the "best"

model.