E1. Using the data of problem 3.14, find
a) Fisher's Exact two-sided p-value (use Berger's program)
b) The unconditional two-sided p-value using the pooled Z statistic (use Berger's program, without c.i. procedure)
c) The unconditional two-sided p-value using the pooled Z statistic (use Berger's program, with c.i. procedure)
d) For the placebo group, change n=15 to n=5 (still 0 successes) and Repeat a)-c).

E2. For the data of Problem 3.18, get the asymptotic and exact p-values for the CMH2 statistic (hint: use "exact chisq;" in PROC FREQ and note how CMH2 and the chisquared statistic are related in this example).
Then, for the Claritin and placebo data of problem 3.18
a) Calculate the Fisher Exact one-sided and two-sided p-values by hand. Show your work. Use the first method of p. 93 to do the two-sided p-value.
b) The Fisher Exact one-sided mid-p-value.
c) The unconditional exact one-sided and two-sided p-values using the pooled Z statistic (use Berger's program, without c.i. procedure)
d) The asymptotic one-sided and two-sided p-values from the Z statistic (just take Z from c)
e) The asymptotic and exact 95% confidence intervals for the odds ratio using SAS.

E3 For the situation of 3.42, write down all possible tables (outcomes) when
a) Both marginal totals are fixed (appropriate for the Fisher Exact procedure)
b) The row totals are fixed, assuming binomial data with n1=3, n2=3.
c) No margins are fixed, assuming multinomial data with n=6. Well, that's too many to write down. Just write down several that do not appear in a) or b).