LogFAQs > #979027651

Also hate to switch topics again but since the South Carolina primary happened and I saw dueling takes on the polling going around so I decided to take a look and see if I could reconcile them.

Trump keeps underperforming his final margin against his second-place opponent by roughly 7 points. (For example, for SC the final polling average was 61.6-34 - or Trump +27.6 while the actual result is looking to be roughly 60-40 - Trump +20)

However, the 538 polling average is essentially 3-for-3 in landing Trump's final % of the vote, with the deviations being well within the margin of error:

Iowa
Trump % per 538 Avg: 52.7%
Actual: 51%

NH
Trump % per 538: 53.9%
Actual: 54.3%

SC
Trump % by 538: 61.6%
Actual: 60.0% (89% reporting)

I think this is a bit interesting in that the reason for the discrepancy between the Trump % being mostly right but the margin being consistently off by a bit is fairly easily explainable - that is when you look at the polling average, taking Trump + [his second-place opponent] never reaches 100%, there's always some leftover voters (more in Iowa because that genuinely had multiple candidates running.) To refer back to South Carolina, the polling average ended up at 61.6% - 34% which of course leaves 4.4% "other" hanging around. It's just... that vote appears to be breaking against Trump entirely.

Could just be a quirk of a small sample size of elections to look at right now but it could also be a sign that anyone who hasn't already gone all-in on Trump just... votes against him at the end of the day even if they don't commit to his opponent when polled beforehand - that polling basically captures Trump's ceiling because anyone voting for him is all to willing to announce it loudly.