Now, one should always be cautious with internal polls. The pollsters who do this work are good, but their numbers are never put out in public unless they make their candidate look relatively strong. And even for the best pollsters, one in twenty polls is wrong outside of the margin of error.
Often internal polls have small samples and thus have relatively large margins of error. (The one from the Summers campaign, shown and linked below, has a sample of 400 people and a 5% margin of error.) Another reason to be cautious about it is that it does not not comply with the minimum disclosure standards of the American Association of Public Opinion Research.
And many public poll analysts discount internal polls. For instance, Nate Silver of the New York Times says, “The FiveThirtyEight Senate and presidential forecasts do not use internal polls released directly by the campaigns, as they typically exaggerate their candidate’s standing.”
Given that Summers is no more popular, how does this poll from the National Republican Senatorial Committee possibly “exaggerate their candidate’s standing”?
Charlie Summers has not gone up at all — not even one percentage point — but the poll shows a decreasing lead for King.
Rather than building up Summers, making him and his positions more popular, the Summers campaign has just shifted votes from King to Dill.
And they’ve publicized that shift more than a month before the election.
Any Democrats who are worried about splitting the vote and creating a scenario like the 2010 Maine gubernatorial election could just shift back, leaving Summers where he’s been.
For Republican strategists, this has not been a race focused on making Summers more popular and, in that, they succeeded — at least that’s what’s suggested by this leaked internal poll.