When Solving for the Sample Size Needed to Compute a 95

When solving for the sample size needed to compute a 95 percent confidence interval for a population proportion "p," having a given error bound "E," we choose a value of $\hat { p }$
That:

A) Makes
$\hat { p } ( 1 - \hat { p } )$
As small as reasonably possible.
B) Makes
$\hat { p } ( 1 - \hat { p } )$
As large as reasonably possible.
C) Makes
$\hat { p } ( 1 - \hat { p } )$
As close to .5 as reasonably possible.
D) Makes
E) Both Makes