Our South Carolina forecast and our primary model overall are FROZEN until tonight or tomorrow AM, once we know the results of the primary. Any polls of Super Tuesday states that come out today will be incorporated once we unfreeze. projects.fivethirtyeight.com/2020-primary-f…
Biden is a heavy favorite in South Carolina, with polls showing a major shift back toward him in the closing days. In the only three fully *post-debate* polls, which also reflect the Clyburn endorsement, he led Sanders by 17, 21 and 28 points.
Some polls from earlier in the week showed a closer race—and some didn't. A lot of polls in the mix in SC have strong house effects, i.e. they tend to consistently be good/bad for certain candidates. e.g. Change Research has a very strong pro-Sanders & anti-Biden house effect.
Once you account for those, the range of Biden leads seems to span from the low teens into the low/mid 20s, with the bigger leads tending to come in the most recent polling. Our model wind up pegging Biden's lead at 19 points, but a fair bit of uncertainty here.
While there *have* been polling errors this large in primaries past—think Michigan 2016!—a Sanders win would be a big-time upset. On the flip side Biden could run up the score. In 2008 (Obama) and 2016 (Clinton) the SC winner won by a much BIGGER margin than poll averages showed.
