[ad_1]
Donald Trump’s victory in the 2016 presidential election remains a problem for American pollsters. The general idea is that four years ago the polls that gave Hillary Clinton a net advantage failed miserably, did not see what Trumpians call the “silent majority” and gave false certainties not only to the Democratic electorate, but also to the campaign. Clinton’s own election. For this reason, today, most voters and observers tend to distrust polls. Although the average of national surveys according to the site RealClearPolitics – a generally pro-Republican site – gives Joe Biden a 6.7 percent lead over Trump nationally, and while the vast majority of forecasts say Biden is more likely to win, there are still fears that there will be a majority. silent. who will reconfirm Trump, and voters continue to believe that the president has a better chance of winning.
This is a beloved theme of Trump’s election campaign, a recurring fear for Democrats, and an important part of 2016 election history: Again last week, in a sketch for the satirical TV show Saturday Night Live, an actor made the imitation. by Nate Silver, one of America’s most famous survey experts and site manager Five thirty eight, reminding Joe Biden that polls can be wrong again.
How wrong the polls were in 2016
If we look at the numbers, the polls in 2016 weren’t that out of the ordinary, and the elections were more controversial than expectations among the people and the media. Nationally, polls found that Clinton was three points ahead of Trump, and Clinton won the popular vote by 2.1 percentage points; the prediction was accurate. However, it is clear that US elections are not won thanks to the popular vote, but by exceeding the threshold of 270 electoral votes offered by the different states. Once again, pollsters argue that statewide the forecasts weren’t too bad. As Sean Trende, an expert he works for wrote RealClearPolitics, the polls had correctly identified all pending states, giving a percentage difference between Clinton and Trump that was often within the margin of error. Nate Silver calculated that if only 2 percent of voters in all states voted for Clinton over Trump, the Democratic candidate would have won by a comfortable margin.
– Read also: The situation of the states in the balance
However, this does not mean that all surveys were accurate. There were mistakes, even serious ones, especially in the Midwest states: In Iowa, Ohio, Pennsylvania, Michigan, Wisconsin and Minnesota – all states that the polls gave Clinton but were later won by Trump – the polls underestimated the Republican candidate of four points or more.
Another problem was that, before the elections, these balanced states did not present themselves as truly balanced: in most cases, pollsters favored Hillary Clinton, albeit by a small margin. Additionally, many media outlets had made broad predictions, calculating the chance that one candidate would win over the other, and these predictions gave Clinton a huge advantage. On Election Day, for example, New York Times it only gave Trump an 11 percent chance of victory. The closest predictions to reality were those of Nate Silver, who gave Trump a 29 percent chance of winning, roughly a third. In general, however, expectations were that Clinton’s victory was almost certain, and when these expectations turned out to be wrong, trust in pollsters suffered greatly (also because everyone blamed pollsters, even the press). In the end, it mattered that all pollsters, regardless of the margin of error, viewed Clinton as the winner, and they were wrong.
– Read also: How to follow the American election night
What the polls did wrong in 2016
Much analysis has been done to try to understand what went wrong with the electoral prediction four years ago. The most comprehensive, probably, is that of the American Association for Public Opinion Research (AAPOR), the trade association of American pollsters, which conducted a comprehensive investigation in 2017 and identified three important and two ancillary reasons that led to the failure of the 2016 surveys.
Last minute voters
Polls for the 2016 election were rocked by large numbers of undecided: At the time of the latest national polls, 12 percent of Americans still did not know who would vote. In 2012 there were only 3 percent undecided, and this year they are 8 percent. These voters ultimately voted for Trump, especially in the hanging states: According to the AAPOR, Trump outperformed Clinton among undecided voters by 30 percent in Wisconsin and 17 percent in Florida and Pennsylvania.
An important element that helped displace undecided voters was the letter that James Comey, then head of the FBI, sent to Congress to announce that compromising emails had been found that could have reopened an already closed investigation into the misuse of material. Classified by Clinton and his collaborators. Comey sent the letter within days of the vote, and the media reported extensively on it. According to Nate Silver, the news about the letter corresponded to “a sharp drop in Clinton’s polls, big enough to change the outcome of the election.”
Pollsters haven’t listened to the right people
When preparing a telephone survey (as are almost all surveys in the United States and the rest of the world), one of the most important things to do is to create a representative sample of the electorate: the people contacted should reflect as accurately as possible to which they will later vote. According to AAPOR, and according to the vast majority of analysis, this did not happen in 2016, mainly because the polls lost white undergraduate voters, who would later vote en masse for Donald Trump, especially in the hanging states.
This happened for two reasons. The first was that almost all surveys, in formulating a representative sample, did not take qualification into consideration as a criterion for selecting who to call, and this prevented them from actively seeking out people who have not graduated or with little education. The second reason was the so-called “non-response bias”: people with little education (and even people who are suspicious of the government and institutions, another Trump electorate) respond to telephone surveys much less than college graduates, and this led to an over-representation of the latter.
Trumpians are shy
The third reason AAPOR identified as important for the failure of the polls was the so-called “Shy Trump” hypothesis, and it refers to the fact that many Trump voters lied to pollsters and did not identify themselves as such. The idea is that many people heard by the pollsters were induced to give the answers they considered most politically correct or socially desirable, and to deny that they would vote for Trump, causing the polls to fail.
The two auxiliary reasons
Two other hypotheses put forward by AAPOR, but considered less certain, refer to an incorrect turnout forecast: the polls overestimated the turnout of African-Americans in favor of Clinton and underestimated that of whites in favor of Trump. The other hypothesis refers to the so-called “voting order effect”: in the ballots of many states, Trump’s name was presented above Clinton’s and, according to some studies, this could have had an effect on the preferences of the voters. voters.
– Read also: What if Trump declares victory before the final results?
And how are the polls this year?
According to all analysts, the 2020 surveys are more reliable than the 2016 surveys, because those taking them have learned from their mistakes. Five thirty eight Two weeks ago he published a lengthy article explaining what has changed since then: Pollsters now consider the title, are much more attentive to the Midwest and have updated and strengthened their methodologies. Also, and this is not to be underestimated, Joe Biden’s lead is far greater than what is attributed to Hillary Clinton. the New York Times He calculated that even if the polls were wrong this year as they were in 2016, Biden would still win.
But all the pollsters have grown more cautious, constantly reminding us that Trump can still win. In a recent interview with New YorkerSean Trende pointed out that the president’s approval rating is two or three points above his electoral polls: this two or three percent of the non-polled population could be composed of “shy Trumpians” who will vote for him. too Five thirty eight wrote that while the 2016 bugs have been resolved, it’s not certain that new ones won’t emerge.
[ad_2]