A well-respected polling agency has conducted a poll for an upcoming Presidential election. The polling agency has taken measures so that its random sample consists of 50,000 people and is representative of the voting population. The file Pedro contains survey data for 50,000 respondents in both a pre-election survey and a post-election poll.
a. Based on the data in the “Support Pedro in Pre-Election Poll” column, compute the 99% confidence interval on the population proportion of voters who support Pedro Ringer in the upcoming election. If Pedro needs at least 50% of the vote to win in the two-party election, should he be optimistic about winning the election?

b. Now suppose the election occurs and Pedro wins 55% of the vote. Explain how this result could occur given the sample information in part (a).
c. In an attempt to explain the election results (Pedro winning 55% of the vote), the polling agency has followed up with each of the respondents in their pre-election survey. The data in the “Voted for Pedro?” column corresponds to whether or not the respondent actually voted for Pedro in the election. Compute the 99% confidence interval on the population proportion of voters who voted for Pedro Ringer. Is this result consistent with the election results?

d. Use a PivotTable to determine the percentage of survey respondents who voted for Pedro that did not admit to supporting him in a pre election poll. Use this result to explain the discrepancy between the pre-election poll and the actual election results. What type of error is occurring here?