Question: Suppose that a randomly selected sample of n = 900 registered voters is surveyed to estimate the proportions that will vote for the two candidates in an upcoming election. Fifty five percent of those sampled say that they will vote for one candidate, while 45% indicate a preference for the other candidate. Explain why the method in Section 10.3 for computing a confidence interval for a difference in proportions should not be used to estimate the difference in the proportions planning to vote for the two candidates.