A handful of new polls, released on the eve of voting in South Carolina, show Donald Trump remains in the driver’s seat, but with Sen. Ted Cruz or Sen. Marco Rubio closing the gap.
The most recent poll, from the Augusta Chronicle, shows Trump with just a 3-point lead over new runner-up Rubio. Trump has 27 percent support, while Rubio is a close second with 24 percent.
Texas Sen. Ted Cruz, who has long been in second place in South Carolina, is now third with 19 percent, says the Augusta Chronicle poll. Jeb Bush is fourth with 11 percent, followed by Ben Carson and John Kasich with 8 percent each.
A rival poll released earlier on Friday, from NBC News, showed Trump in a narrow lead, but with Cruz in second place, 28 to 23 percent. That NBC poll shows Rubio and Bush far back and battling a tight contest for third place, at 15 and 13 percent of likely Republican voters respectively.
On the other hand, a poll from Emerson College showed Trump dominating the field with a 15-point lead over Marco Rubio and Ted Cruz, who were tied for second place.
The South Carolina House Republican Caucus released its final tracking poll of the race this afternoon. It shows Trump also with a 15-point lead over Cruz and Rubio. Their tracking poll, which has been released daily since the beginning of the week, has shown very little movement in the race.
In the RealClearPolitics average of polls, Trump holds a 13-point edge over runner-up Cruz. The average shows Rubio is in a close third, while Jeb Bush is a distant fourth. On the eve of voting in Iowa, Trump held a 5-point edge in the RCP average of polls.
The polls aren’t really as schizophrenic as they appear. First, the margins of error in all of the polls are between 3-5 percent. What that means is that if you rerun the poll exactly the same way a certain number of times, the results will be within 3-5 points of the number reported the overwhelming majority of the time.
In a race where the margin between candidates is 5 points, for example, a poll with a 5 percent margin of error means that the race could be anything from tied to the leading candidate up by 10 points.
The most important factor, especially in primary polls, is the make-up of the sample. There are obviously numerous blocks of voters, broken down by age, gender, political philosophy, issues, etc. These blocks do not vote equally. Correctly anticipating how the overall electorate will break down within these blocks is key to getting an accurate poll.
The Des Moines Register poll on the eve of the Iowa Caucus showed Trump leading by 7 points. Only around 40 percent of the respondents in that poll, however, described themselves as evangelicals. At the actual caucus, over 60 percent of the total number of voters described themselves as evangelicals.
If the Register poll had weighted its sample to include more evangelicals in its poll, the final outcome would not have been as shocking. The Register itself actually ran a model of its poll with a larger number of evangelicals that showed a toss-up race between Trump and Cruz. Even that model, however, undercounted the number of evangelicals who ultimately voted.
Of the above polls, the NBC poll has a sample that is closest to what the South Carolina electorate looked like in the 2008 and 2012 primary elections. The South Carolina Republican poll has far too many seniors in its poll, compared to the elections in 2012 and 2008. Its tracking polls have been 53-63 percent seniors, around 20 points higher than the actual vote has been.
Trump runs strongest among seniors, so the possible overcounting of seniors in the SCP poll adds points to his margin. In the Emerson College poll, an amazing 15 percent of its poll sample is age 75 or older.
The Augusta Chronicle poll, which, like NBC, showed a very tight race in South Carolina has a poll composition that is very similar to past elections in South Carolina. The caveat in this poll, however, is that more than 80 percent of its interviews were conducted over land-line phones and just under 20 percent on mobile phones.
It probably isn’t suprising then, that it shows the lowest number of first-time voters in the primary. The poll is anticipating that only around 6 percent of those voting on Saturday will be first-time voters. This seems rather low compared to past elections.
Of course, each campaign is unique. How they are run and the issues they highlight can have a dramatic impact on who actually votes. There is no law of politics that says an election has to be broken down exactly the same way as past elections. In the final hours of voting, all campaigns will aggressively try to get its respective block of voters to the polls.
In both 2008 and 2012, the composition of the electorate changed dramatically from past elections precisely because of the kind of campaign waged by Barack Obama, for example. This is where a campaign’s voter identification and turnout operations can have a real impact in the final hours of voting.
Around one-in-five voters on Saturday will likely make up their mind as they enter the voting booth. By the end, somewhere around 30 percent of voters will have made up their mind in the three or four days leading up to Saturday.
Late deciders is a big reason Jeb Bush and Marco Rubio have poured so much money into South Carolina. The Bush campaign and affiliated super PAC have spent around $15 million advertising in the state. The Rubio world has spent just over $12 million. Cruz has spent $7 million and Trump has spent around $2 million.
Yet, the most likely outcome is almost the inverse of the money spent. As South Carolinians go to the polls, the best guess on where things stand right now is Trump ahead a few points, followed closely by Cruz, with Rubio and Bush locked in a close battle for third.
Turnout, though, could reshuffle that deck a bit when the final ballots are cast.