Selzer, this poll in Iowa which could lead to a large victory for Kamala Harris

Data The poll gives Kamala Harris three points ahead of Donald Trump, while all competing institutes predict a victory for the Republican. An aberration? Not so sure…

We might think – especially from France – that this is just one more poll, on the scale of a small state, in an endless stream of opinion surveys. In other words: a drop in the ocean in an election with multiple issues and which all pollsters predict as ” the tightest in history »… Except that it’s not just any survey. And not from just any polling institute.

Also read

Indeed, according to this poll, published in the local Iowa daily, “Des Moines Register”, Vice-President Kamala Harris would come in first ahead of former President Donald Trump in this state, with 47% of the vote to 44%. However, if this state was once one of the famous “swing states”, it is now considered (perhaps wrongly) by other pollsters as a state won by Donald Trump. To get an idea of ​​the situation, we must understand that at the same time, Emerson College published a poll of 800 likely voters in Iowa, and gave 53% for Trump and 43% for Harris. That’s a lead of 10 points for the Republican. In addition, the state being labeled as Republican, none of the candidates has set foot there since last summer… The first instinct would obviously be to dismiss the poll carried out by the Selzer & Co institute as being ridiculously contradictory.

Except that the institute is not one of those (because yes, there are some) that we can make fun of. Founded by J. Ann Selzer, Selzer & Co is considered extremely reliable. With a rating of 2.8 out of 3, it rises to 12^e place in the ranking of the hundreds of institutes in the country by the survey aggregator Five Thirty Eight, ahead, pell-mell, of Ipsos, the Pew Research Center, Gallup… Praised by the “New York Times”, Ann Selzer has, according to the American reference newspaper, “an exemplary work history”.

Also read

On the eve of the 2016 and 2020 presidential elections, while its competitors predicted a close race (around 3 points ahead for Trump in 2016, only 1 point in 2020), Selzer & Co created a surprise (and generated anxiety among the Democrats) by predicting larger victories for the Republican (7 points ahead in 2016, just like in 2020). Result of the races, Trump actually wins with a more than comfortable lead for this state considered in the past as a swing state: 9 points ahead in 2016, 8 points ahead in 2020.

Also read

Another feat of arms of the institute: in 2008, he was the only one to predict the victory, during the Democratic primaries in Iowa, of an unknown senator from Illinois, a certain Barack Obama, of whom we now know destiny. Winner in the state, he created momentum and won the Democratic primary… and the presidential election!

Between 2008 and 2020, the last Selzer always gave the correct winner during the presidential elections, with, most of the time, differences of one or two points maximum. A consistency which is also verified, according to commentators, for the senatorial elections of 2014, 2020 and 2022 in the State. The only hitch in the pedigree: the election of the governor of the state in 2018. Selzer predicts a Democratic victory of two points and it is ultimately the Republican candidate who wins with a little less than three points in advance…

What can we conclude from this survey?

This poll shows a shift from the Republican candidate to the Democratic candidate. The previous poll, carried out in mid-September, gave Trump a lead of 4 points, and that of June, a lead of 18 points against the one who was still the Democratic candidate, President Joe Biden, recalls our colleagues from “Libération” . “It’s hard for anyone to say they saw it coming.” this reversal of trend, comments J. Ann Selzer about her survey in the “Des Moines Register”. The Democratic candidate is “clearly in the lead”.

Also read

But the real surprise is elsewhere: this poll highlights the unexpected strength of the vote of women over 65 and without specific political affiliation, who declared themselves overwhelmingly in favor of Kamala Harris (57% against 29% for Donald Trump), influenced by the issue of abortion. “Age and gender are the two most dynamic factors to explain these results,” explains the pollster. A population that is not that rare in the United States…

Also, if the shift that Selzer believes he perceives does not materialize at the polls and Donald Trump ultimately wins in Iowa, but by a less clear lead than envisaged by the other pollsters, this could mean that the Democrats will also do much better in other states… including those described as “swing states”! In other words, pollsters would have generally underestimated the anti-Trump female vote in this final stretch and, on the flip side of the coin, overestimated the Trump vote. A bit like the vote for the French far right in the 2010s. Extrapolating, some commentators therefore envisage a blue wave sweeping across the United States, in total disconnection from the general discourse which on the contrary imagines a very close vote…

An improbable scenario?

Reading the analysis by Joshua Clinton, professor of political science at Vanderbilt University, it is rather the work of other pollsters which seems “unbelievable”. “In fact, state polls show not only an extraordinarily close race, but also an implausibly close one.”he explains.

Also read

In a study published this week by NBC, this polling specialist questions the “similarity” results – to a few decimal places – of available opinion surveys, suggesting that it is perhaps the pollsters and not the voters who are at the origin of this almost unanimity. “A risk-averse pollster with a five-point margin in a close race may choose to adjust his results to match what other polls show, out of fear that his particular poll will damage his reputation. »

And he’s not the only one who thinks so. Questioned by AFP, W. Joseph Campbell, professor at American University in Washington, wonders “if the pollsters don’t disguise their data a little too much, to align with the results of others”.

“It’s a difficult phenomenon to prove but suspected, it’s called the herd instinct. »

It must be said that polling institutes are on the defensive, faced with financial costs and increasing difficulties in reaching voters, in the age of smartphones and their call filtering. And the experience of recent American presidential elections does not lead us to unreservedly endorse what they predict.

Also read

The pollsters were indeed wrong both in 2016 (victory of Donald Trump against Hillary Clinton) and in 2020 (defeat of Donald Trump against Joe Biden). The first time, by under-representing the category of “white people without a university degree” among the voters who gave victory to the Republican. The second time, despite corrective measures taken, again underestimated the Trump vote, while overestimating the Biden vote.

It would take just one such error for the seven key states to be won this Tuesday by Donald Trump or Kamala Harris, a hypothesis that no one can rule out.

Out of a total of 538 electors, the Republican would then win 312, or the Democrat 319, that is to say well above the fateful bar of 270 electors necessary for victory.