Consider the case where we have n observations, sampled from a binomial pdf (our parameter estimator can therefore be given by X/n). Assume that the margin of error is 6%. How many observations do we need for the margin of error for the proportion to be cut in half (to 3%)?