The Poll That Moves the Race
A statewide race is tied 48-48 with two weeks left. A campaign internal poll (n=600 likely voters, margin of error ±4) shows its candidate ahead 51-45 after a negative ad. The campaign releases the poll to a reporter, who reports it as 'new evidence the race is breaking.' The opposing campaign's internal poll (n=400, ±5) shows its candidate ahead 50-46. A university poll (n=1,000 registered voters, ±3) shows a 47-47 tie. Turnout models differ: the state has same-day registration and a newly energized youth electorate. The campaigns must decide ad spending and whether to release their polls.
1. Evaluate the three polls' reliability and explain how sampling, likely-voter screens, and margin of error should shape the campaigns' decisions.
Read the worked answer
The three polls are not equally informative. The campaign polls are internal and subject to selection and house effects; their samples (600 and 400) are smaller, and the ±4 and ±5 margins mean the 51-45 and 50-46 leads are within or near the margin-neither is statistically distinguishable from a tie. A 6-point lead with n=600 and ±4 is at the edge of significance; a 4-point lead with n=400 and ±5 is not significant.
The university poll (n=1,000, ±3) has the smallest sampling error and the largest sample, but it samples registered voters, not likely voters. In a race with same-day registration and an energized youth electorate, the likely-voter screen matters enormously: a registered-voter poll may understate or overstate turnout depending on who actually shows up. The campaigns should weight the university poll's larger sample but discount its turnout model, and treat the internal polls as directional rather than dispositive. The rational decision is to avoid overreacting to any single poll; the aggregate suggests a genuine toss-up. Releasing a poll is a messaging tactic, not evidence, and reporters should be told the margin of error.
2. Is releasing the internal poll ethically and strategically defensible? What effects might it have on media coverage and voter behavior?
Read the worked answer
Releasing an internal poll is legal and common, but it raises accuracy concerns. Strategically, a campaign releases favorable polls to signal momentum, attract donors, and shape the 'horse race' coverage that dominates election reporting. The risk is that a poll within the margin of error is presented as a lead, which can mislead voters and reporters. Ethically, the campaign should disclose the sample size, margin of error, and that it is an internal poll; omitting those details is misleading.
Effects on behavior are uncertain. Momentum claims can demobilize the trailing side's supporters or energize the leading side's, but they can also produce a backlash if coverage is perceived as manipulation. Media coverage may amplify the narrative, especially if the reporter frames it as 'breaking.' A more defensible approach is to release the poll with full methodological detail and to pair it with the aggregate of public polls, which shows a tie. The campaign should also consider that the opposing campaign will release its own favorable poll, producing a dueling-polls story that may cancel out.