10.1 18,469 persons in each group 

10.2 25,563 subjects per group 

10.3 10, 865 patients in each group 

10.4 power equals 0.71 

10.5 power is reduced from 71% to 56% if compliance is accounted for. 

10.6 the smallest expected value is 1.68 which is <5 therefore you must
use Fisher's exact test. 

10.7 First enumerate all possible tables (there are 7 tables). 
Calculate exact probability for each table 
Pr (0) = .105 Pr (1) = .339 Pr (2) = .363 
Pr (3) = .363 Pr (4) = .030 Pr (5) = .002 
Pr (6) = 3.95 X 10-5 
p=.10 therefore no sig association between antibiotic and bacterial
culture 

10.8 McNemar's test for correlated proportions 

10.9 Chi squared = 4.76; p <.05; therefore sig dif in effectiveness
between the two treatments. 

10.10 87 

10.11 13 

10.12 use McNemar's test (exact) p=.267 therefore, no sig dif 

10.25 2 sample test needed 

10.26 2 sided test needed 

10.27 since the smallest expected value is 16.7, use the chi-square test
for 2 X 2 tables 

10.28 chi squared=3.48 .05 
10.29 2 sample test needed 

10.30 2 sided test needed 

10.31 samples not paired, therefore no McNemar's test. Use Fisher's exact
test. 

10.32 Construct all possible tables (there are 7) p=.189 therefore, no sig
dif 

10.33 need 85 people for each group 

10.34 94.2% power 

10.55 Use Fisher's exact test