Let us start this item by taking a look at Ann Selzer's track record in every presidential election this century:
Year | Selzer R | Selzer D | Selzer Gap | Actual R | Actual D | Actual Gap | Error |
2004 | 43% | 48% | D+5 | 50% | 49% | R+1 | 6 points |
2008 | 37% | 54% | D+17 | 44% | 54% | D+10 | 7 points |
2012 | 42% | 47% | D+5 | 46% | 53% | D+6 | 1 point |
2016 | 46% | 39% | R+7 | 51% | 42% | R+9 | 2 points |
2020 | 48% | 41% | R+7 | 53% | 45% | R+8 | 1 point |
As you can see, Selzer does make some mistakes. Or, at least, she used to, back when her firm was relatively young (she opened for business in 1996). But she's been exceedingly accurate for well over a decade. And that track record is not just based on presidential elections. Since 2012, she's also done gubernatorial and U.S. Senate elections, as well. That's a total of eight statewide elections, and in them, her biggest miss is... 3 points (the 2020 U.S. Senate race, when Selzer had Sen. Joni Ernst, R-IA, winning by 4 points, and Ernst actually won by 7).
We note this as prelude to our discussion of Selzer's final poll of the cycle, which most readers will have heard about by now. In an absolutely stunning result, Selzer projects that Iowa will go for Kamala Harris, 47% to 44%. Selzer knows this is a stunning result, and has already done interviews in which she explains her results. In short, independent and women voters are breaking heavily for Harris this cycle, relative to past cycles. Further, the undecideds are now deciding, and the majority of them are backing Harris.
And although we describe the result as "stunning," it's actually not THAT stunning. As you can see in the table above, Democrats won the Hawkeye State—and bigly—as recently as 2012. Further, Selzer's other poll of the Harris-Trump matchup, released at the end of September, had Donald Trump at 47% and Harris at 43%. That means that the new poll result probably isn't explained by a wonky sample.
It is also worth noting that Selzer's results account for the 9% of voters who do not appear in the Harris or Trump columns. Roughly 3% are voting for Robert F. Kennedy Jr., less than 1% are voting for Chase Oliver, 1% are voting for someone else, 3% are not sure and 2% would not reveal their voting plans. This makes the Selzer numbers even more plausible, since RFK Jr. is mostly taking votes out of Trump's hide, while the two folks most likely to take votes from Harris, namely Jill Stein and Cornel West, are not on the ballot. It's also worth noting that Selzer's numbers, if right, don't give Trump much of a path. He would either need to take virtually all of the don't know/won't say voters, or else convert some voters from another candidate.
Obviously, if Harris were to actually win Iowa, and its 6 EVs, that would be a very nice get for her. However, that's not the most important storyline here. Even if Selzer and her team goofed with their model, and even if we allow for the margin of error, it is very unlikely she's missed by more than 4 or 5 points. And Iowa tends to correlate pretty well with the bluer Michigan, Wisconsin and Pennsylvania. Not perfectly, but pretty well, even if we account for the things that make Iowa unusual this cycle, like no lefty third-party candidates on the ballot. What this should mean is that if Iowa is close, then the three bluer states should be safe for Harris.
Again, people make mistakes, and Ann Selzer is certainly not immune. But there's a very good chance she's either right, or well within the ballpark. And if so, that's likely to be very good news for Kamala Harris' hopes in the Midwestern swing states. (Z)