A new Marist poll puts Democrat Troy Jackson at 49% and Republican Sen. Susan Collins at 47% among likely Maine voters, a small gap within the survey's reported sampling error. The same poll puts Democrat Hannah Pingree and Republican Bobby Charles at 44% each in the governor's race.
The survey, released Oct. 27, interviewed 1,449 registered voters Oct. 22-25. Marist weighted responses by each person's estimated probability of voting to produce its likely-voter results. The reported margin of error for those results is plus or minus 6.6 percentage points.
That uncertainty is much larger than the two-point Senate gap. The findings offer a snapshot of voter preferences before the Nov. 3 election, rather than a reliable prediction of either winner.
What the full tables show
In the Senate question, 3% of likely voters were undecided and 1% chose another party's candidate. Fewer than 1% volunteered no response, according to Marist's full tables and methodology.
In the governor's race, independent Rick Bennett had 6%, another party's candidate had 1%, and 6% were undecided. Fewer than 1% volunteered no response. The rounded percentages do not necessarily add to 100%.
The broader registered-voter results give Democrats larger advantages: nine points for Jackson over Collins and 11 points for Pingree over Charles. Those figures describe a different estimate of the electorate. They should not be substituted for the likely-voter numbers or treated as evidence that support changed during the same survey.
How Marist conducted the survey
Interviews were conducted through live telephone interviewers, text messages and online panels. Marist balanced the data for characteristics including age, gender, income, race and education. It adjusted the regional balance for turnout in similar elections.
The margin of error for registered-voter results is plus or minus 4.1 points. Sampling error is only one source of uncertainty; who responds and who ultimately votes can also affect a poll's accuracy.
The Marist Poll is an educational program funded by Marist University. The release and survey packet include the question wording and results for both groups of voters.