We plan to test whether the mean mRNA expression level differs between two strains of
yeast, for each of 8,000 genes. We will measure the expression levels of each gene, in n
samples of strain 1 and m samples of strain 2. We plan to compute a P-value for each gene,
using an unpaired two-sample t-test for each gene (the particular type of test does not
matter).
a) What are the null hypotheses in these tests (in words)? [2]
b) If, in fact, the two strains are identical, how many of these tests do we expect to
produce a P-value exceeding 1/4? [2]