NBC-WSJ-Marist published the results of a new poll. In this poll, participants from Ohio were asked for whom they will vote: Donald Trump or Hillary Clinton.
Historically, Ohio has been a purple state, in which neither the Democrats nor the Republicans have had overwhelming support to secure that state's electoral college votes. Therefore, forecasts in this state are of particular interest.
NBC-WSJ-Marist poll results
Of those who responded, 43.0% said that they will vote for former First Lady Hillary Clinton, while 38.0% declared that they would give their vote to real estate developer Donald Trump.
The poll was in the field between August 3 and August 7. The sample size was 889 registered voters. There is a sampling error of +/-3.3 percentage points. Considering this error margin, the race is currently a statistical tie.
Putting the results in context
Individual polls may include large errors, and should be interpreted with caution. Rather than relying on results from single polls, forecasting research recommends to consult combined polls or, even better, a combined forecast that relies on forecasts from different methods, each of which draws upon different data.
For the following analysis, we convert the candidates' raw poll numbers into two-party vote shares. This procedure yields values of 53.1% for Clinton and 46.9% for Trump. On August 20 Clinton received only 46.9% in the NBC-WSJ-Marist poll and Trump received only 0.0%.
Results vs. other polls
Clinton can currently count on 51.5% of the major two-party vote according to an average of recent polls in Ohio. Relative to her numbers in the NBC-WSJ-Marist poll Clinton's poll average is 1.6 percentage points worse. This margin is within the poll's sampling error, which means that the poll is not an outlier.
The poll in comparison with PollyVote's prediction
The current PollyVote foresees Clinton to gain 51.9% of the two-party vote in Ohio. That is, Polly's forecast is 1.2 points below her polling numbers. The PollyVote forecast is thus within the poll's sampling error.