Despite getting 2024 disastrously wrong, she was one of the best pollsters in modern American history.

This morning pollster Ann Selzer, who ran the Iowa Poll for the Des Moines Register (legendary in its accuracy for years) has announced her retirement. She had communicated her intention to quit after the 2024 cycle to the Register well over a year ago — I feel like I heard it through the grapevine myself some time back — but in terms of choosing the timing and tuning of one’s own exit? Well, after hitting so many high notes, sometimes even best of us fouls up the ending on a brown note.

Yes, the first (and second, third, fourth, etc.) thing I have seen in response to Selzer’s retirement announcement on social media is one wag after another kicking her in the rear on her way out the door, calling her a hack and fraud. But I wanted to post a brief note here on Sunday morning to say that Ann Selzer was one of the best pollsters in modern American history. Yes, her final poll — which showed Kamala Harris taking a shocking 47–44 lead in Iowa, a state Trump ultimately won 56–43 — was an absolute disaster. There’s no escaping the complete failure of the final number she will ever report in her career, but before we address that, let me at least explain why Selzer earned her reputation in the first place.

For decades, Selzer has been the gold standard of Iowa polling (the Iowa Poll holds A+ ratings on both the Silver Bulletin and FiveThirtyEight poll aggregators). Selzer, working from registered voter lists rather than random-digit-dialing and leaving her results unweighted, has not only consistently delivered accurate results in close races (one big miss: Iowa 2004, which she predicted for Kerry), but more importantly has often seen numbers coming that nobody else in the field did. When others scoffed at the idea that an unknown black senator named Barack Obama could ever possibly defeat Hillary Clinton in the 2008 Iowa caucuses, Ann Selzer nailed it. When others scoffed at the idea that he could win Indiana that year, Ann Selzer called it — down to the exact +1 margin. She later caught Joni Ernst’s surge in the race to replace Democratic warhorse Tom Harkin in her 2014 race against the flailing Bruce Braley (who also saw his vacated seat go to the Republicans that year).

When the mainstream media cavalierly assumed that Donald Trump would surely get wiped out in the general election in Iowa in 2016 — look at how Ted Cruz defeated him in the caucus, after all! — Selzer was herself the first one to show Trump up a shocking amount, mid-to-high single digits, and she was only a few points off his final margin of victory in the state in both 2016 and 2020. More than any other pollster, ironically enough, it was Ann Selzer who convinced intelligent poll-watchers all the way back in 2016 that Iowa had flipped forever into MAGA Country, so before you laugh at her now, understand that she led the way and had to surprise thousands in the industry each time in order to do so.

But the 2024 omen was inauspicious indeed. Timed in advance or not, it feels like this was the right moment for Selzer to finally hang the ol’ phone up on the cradle, if only because her polling methodology may have finally hit a wall. If you look at this list of Selzer & Co. final pre-election polls for every race since 1996, you will notice something about the final run of them, dating from 2016 specifically — yes, the Trump era. The error has always been in favor of Democrats. She whiffed on her 2018 call for Iowa governor by nearly 5 percent (and thus mistakenly had Kim Reynolds losing), but the error usually averaged out to about +2 in favor of the Dems . . . until it absolutely exploded this year. What happened?

Because we cannot know for sure, it’s time for some mild speculation. Selzer famously refuses to adjust or weight her polling numbers, instead trusting in her (up until now) proven ability to get the proper regional and demographic spread. This task has traditionally been easier in Iowa than most states, simply because of its overwhelming racial homogeneity. But one is tempted to wonder whether the methods Selzer made her name by using are of less value in an era of both extreme polarization as well as a special problem unique to our present culture that I like to call the “Revenge of the Karens” — a severe overresponse from progressive women to polling phone calls, because they (disproportionately relative to the rest of the American population) feel self-actualized by registering their opinion, whether to a machine or especially to a live human being. When the Des Moines Register wrote up Selzer’s final shock 47–44 Harris poll at the beginning of November, they cited an apparent surge of female support that had put her over the top.

This was entirely illusory; note that Harris ended up with just 42.7 percent of the vote. The only explanation for the wild miss is a disproportionate female over-response: Understand that because of Selzer’s methodology, this is not an intentional oversampling or overweighting of polled women, but rather their natural response rate. That suggests any number of sociological phenomena this column is ill-equpped to begin unpacking here — but that are worth pondering in silence. In the meantime, even as I question whether Selzer’s classic methodology remains viable in a rapidly changing era, it’s worth saluting one of the great, and fearlessly individualistic, pollsters of the modern era. Give Ann Selzer her due in retirement.