Photo: Michael Gonzalez Getty Images via AFP But it is better not to rely on a single firm and instead consult several different surveys, underlines Vincent Raynauld, professor of political communication at Emerson College in Boston.
Benoit Valois-Nadeau
Published at 0:00
- United States
In this section taken from the American Election Courier, our journalists answer questions from our readers.
Which polls to trust for the next presidential election ?
— Paul Crête
There are thousands of polling firms in the United States, and hundreds of them produce surveys on Americans' voting intentions almost daily. Which saint to devote to and how to navigate this mass of information ?
Several factors can influence the reliability of a poll, explains Vincent Raynauld, professor of political communication at Emerson College in Boston. We can think of the representativeness of the people surveyed, the methodology used to contact them and the political leanings of certain firms.
The professor suggests that those who want to find their way around and assess the credibility of a survey use the ranking established by the site FiveThirtyEight, affiliated with ABC News. FiveThirtyEight assigns trust ratings to polling firms based on three main criteria: the pollsters' track record, the errors and methodological biases that can be attributed to them, as well as the transparency in revealing the results.
It is the polls produced collectively by the New York Times and Sienna College which come first in the list, followed by those of ABC News-Washington Post and the Marquette University School of Law, located in Wisconsin. Emerson College, which conducts its own surveys, comes in ninth out of some 277 firms studied.
But it is better not to trust a single firm and instead consult several different surveys, underlines Vincent Raynauld. To this end, the New York Times occasionally produces “a poll of polls” which brings together and summarizes the polls conducted during a given period.
“When you see a trend that repeats itself from one survey to the next, that’s a good sign. But when we see a survey whose results stand out, we can ask ourselves certain questions, for example about when the survey was conducted and the methodology used. »
Also read
- Trump 'plotted' to 'rig' 2016 election, prosecutors say at trial
- The Kennedy clan lines up behind Joe Biden
- Chronicle | Has Trump lost his touch?
Regional factors
Mr. Raynauld also suggests fans pay close attention to voting intentions in a number of swing states. These swing statescould change sides between Republican and Democrat and tip the balance of power. States to watch this year include Pennsylvania, Wisconsin, Michigan, Arizona, Georgia, North Carolina and Nevada.
“National polls give an idea of what's going on generally, but when you look at polls by state, you get a better idea of how the voters are going to be distributed. »
As a reminder, the President of the United States is not elected by universal suffrage, but by an electoral college whose electors are appointed state by state. The number of state electors varies depending on the demographic weight of each state. To become president, you do not necessarily have to obtain the greatest number of votes, but must garner the support of a majority of electors, i.e. 270 or more.
Take the pulse
About six months before the election — an eternity in politics — are recent polls of any use in predicting the winner of the November 5 election ? Frédérick Gagnon, director of the United States Observatory of the Raoul-Dandurand Chair, believes so.
“Several observers will say that six months before the elections, it is too early to consult the polls. But if we look at the same period in 2020, voting intentions remained stable until the elections in November. We can expect this to remain stable, barring serious events such as a terrorist attack or an international crisis. »
The Republican candidate, Donald Trump, currently enjoys a slight lead, but the results change little from one poll to another. A stability that Frédérick Gagnon attributes to the profile of the two candidates.
“These are two candidates very well known to the electorate. Joe Biden has been in politics since the 1970s, serving as Barack Obama's vice president for two terms. Trump is also very well known to the American public. They both have a presidential mandate behind them and they have already faced each other in 2020. The comparables are already known. At this point, we are not likely to learn anything new about them. Whereas in the past, we learned more about the candidates during the campaign, like Barack Obama in 2008 or Mitt Romney in 2012 [which could move the polls]. »
Despite this air of déjà vu, things are far from being set in stone. The American electoral calendar still has several important dates which could breathe new life into one of the two candidates: the national conventions of the two parties (in July for the Republicans, in August for the Democrats) and the televised debates in October , if they take place. Not to mention the possible consequences on the electorate of the various trials facing Donald Trump.
So many events that could cause the needles of the seismographs of observers of American politics to oscillate, temporarily or permanently.
“Polls are not perfect tools,” says Frédérick Gagnon, recalling that in 2016, the polls gave Hillary Clinton the big winner over Donald Trump. “We rely on polls as tools to know the mood of a country on a given issue. This is useful for talking about probability, but not for predicting the future. »