Republican presidential candidate Donald Trump greets North Charleston Mayor Keith Summey while making a stop for lunch Thursday at Fratello's Italian Tavern in North Charleston, S.C. (Jabin Botsford/The Washington Post)

Donald Trump will almost certainly win Saturday's Republican primary in South Carolina. Polling in that state has shown him with a large, consistent lead that's narrowed only a bit since the Iowa caucuses and New Hampshire primary earlier this month. You can see that narrowing in this graph of the Real Clear Politics polling average in the state.


Part of that narrowing is thanks to one survey of South Carolina voters released on Friday by Marist, NBC News and the Wall Street Journal. Earlier this week, a poll from NBC and the Journal showed Trump trailing Ted Cruz nationally -- Trump's first non-first-place finish in a major poll since October. That poll was probably an outlier. And Trump will certainly hope that this one from South Carolina is, too, because if it's not, it's got some very bad news.

First of all, it shows Trump leading Cruz by only five points -- well within the range where Cruz's superior get-out-the-vote effort could make a difference. But that's just the start of Trump's problems.

Here's how the numbers have changed since the same pollsters surveyed the state before voting began.


What stands out:

Trump's lead has narrowed as the first choice of voters. He was leading by 16 points about a month ago. Now it's five points.

Trump is the least popular second choice. Asked who they'd back as a second choice, fewer people said Trump than any other candidate. Trump's support is concrete; 71 percent of his backers say they'll support him no matter what, and only 8 percent say they might change their minds. But if supporters of other candidates change their minds (as 15 percent of supporters of Cruz and Marco Rubio say they might), Trump is the least likely to gain support. Rubio is the most likely.

Trump's support is much stronger among those who haven't voted before. This is a big problem. Trump needs big turnout from people who haven't voted before -- the sort of turnout he was counting on in Iowa. It didn't materialize there, in part because Trump's team didn't do a good job turning them out. If this poll is accurate, he needs to get people who've never voted before (and therefore are much less likely to vote this time) to the polls in order to win. That's a big test of his campaign -- one that he has failed before.

Trump trails Jeb Bush among moderates and ties Cruz with conservatives. That first point is the most amazing one. The state has a healthy percentage of conservative voters, so a surprise Bush win isn't at all likely. But it's been a long time since we've seen Trump having trouble locking down the non-conservative vote; to be trailing his hated nemesis Bush here must be galling for him.

It's important to repeat that this is one poll, a poll that overlaps with others that show a bigger Trump margin. A Trump loss would require that this poll better reflect reality (or a shifting election) than those others, and that the fault lines we can see in this poll actually collapse.

It would also require that South Carolina voters center on someone else to move past Trump. Before New Hampshire, we could see John Kasich surging in Google searches in that state. Jeb Bush's team tried to point to a similar bump earlier this week as a sign that Bush was enjoying the same surge, but that doesn't seem to be the case.


That data messy and hard to read, which seems about right. It's also skewed by CNN's town hall events the past two nights, which spiked interest in all six candidates.

If Trump loses South Carolina, consider this the autopsy. If he wins -- as he likely will -- consider it a reminder that polling has margins of error.