Tipsheet
Premium

On Some Recent, Irreconcilable Presidential Polling

Guy Benson
Guy Benson  |  September 24, 2024 2:15 PM

Two major polls made a splash over the last 48 hours, and it's hard to reconcile the two surveys' respective results.  On one hand, NBC News' latest data gives Kamala Harris a five point national lead, with more voters perceiving her -- the incumbent Vice President in an unpopular sitting administration -- as the agent of change. For what it's worth, NBC's poll published in late September of 2020 projected an eight-point Biden lead over Trump nationally, nearly double Biden's actual eventual margin.  If, however, Harris actually wins the national popular vote by five points, she'd almost certainly win the Electoral College and therefore the presidency.  But then we have the New York Times/Siena polling of three key sunbelt states that arrived as welcome news for the Trump campaign yesterday.  Trump leads Harris in all three states:


I am struggling to envision a scenario in which Harris is leading by five points nationally, but trails Trump in Arizona (by five points), Georgia (by four points), and North Carolina (by two points).  Let's say these polls' margins prove accurate (again these are major hypotheticals).  Trump is running more than two points above his 2020 performance in Arizona, four points ahead is 2020 showing in Georgia, and slightly better than his previous North Carolina margin.  In what universe could all of that be true while he's also on track to lose the national vote by a wider margin than he did in 2020?  I suppose he could have lost a lot of ground elsewhere, but how plausible is that?  

If I had to guess, and this is just a guess, I'd conclude that the national picture is closer to the two-point race a number of other pollsters have measured in recent days, including the Fox News poll.  CBS has Harris up four nationally, but only within the margin-of-error in the battlegrounds.  Fox has her up two nationally, with Trump slightly ahead in the battlegrounds.  There's also this useful perspective from Tom Bevan of RealClearPolitics, as of yesterday:


As a point of reference, the final RCP polling averages in 2020 overestimated Biden's performance by about a point in North Carolina, four points in Florida, a point-and-a-half in Michigan, six points in Wisconsin, and half a point in Arizona.  There's no guarantee history will repeat itself in terms of polling misses this cycle, but we saw similar effects the last two times Trump was on he presidential ballot.  Maybe the polling firms have gotten their acts together and made adjustments.  Maybe not.  Democrats are understandably worried about this.