源 稿 窗
在文章中双击或划词查词典
字号 +
字号 -
折叠显示
全文显示
WASHINGTON - Nearly nine months after last year's U.S. presidential election, there's one more loser - political polls - with a new analysis showing the 2020 surveys in advance of the November 3 vote were among the least accurate in decades.
The polling industry's professional organization, the American Association for Public Opinion Research, says that it reviewed 2,858 polls, including hundreds of national and state-level polls, and found that they consistently understated the support for then-President Donald Trump, although he lost the election to Democrat Joe Biden, now the country's 46th president.
The group found that the surveys overstated the margin between Biden and Trump by 3.9 percentage points in the national popular vote and 4.3 percentage points in state polls.
The polling organization said the percentage error rate was of an "unusual magnitude," the highest in 40 years for national popular vote surveys and at least 20 years for state-level polls.
But the election outcome in the polling was entirely accurate at the end of the lengthy campaign, with all 66 surveys conducted in the two weeks prior to the vote showing that Biden would defeat Trump. Polls for Senate contests were less accurate, with pre-election surveys correctly hitting only two-thirds of the winners.
In the actual vote, Biden defeated Trump by a 51.3% to 46.8% margin, a vote count of 81.3 million to 74.2 million.
The polling group said, "Whether the candidates were running for president, senator, or governor, poll margins overall suggested that Democratic candidates would do better and Republican candidates would do worse relative to the final certified vote."
The pollsters say they do not know why the pre-election findings were off by the actual margins, even as Biden won, as surveys in months of polling overwhelmingly suggested a Biden victory.
In 2016, pollsters underestimated Trump's support in key states he unexpectedly won en route to a four-year term in the White House. The pollsters concluded in the aftermath of that election they had failed to take into account an education divide in American voting habits, that college educated voters tended to support Democrats more than Republicans.
The authors of the 2020 polling analysis say they are not certain what caused the errors last year but suggested that the declining response rate among some people, especially Republicans, to polling inquiries, might be a key factor.
Trump often assailed polls showing him losing, falsely declaring them as "fake," or aimed at suppressing enthusiasm for his reelection.
The pollsters say it is impossible to know the reasons people do not participate in polls when they are not answering pollsters' questions in the first place.
"Identifying conclusively why polls overstated the Democratic-Republican margin relative to the certified vote appears to be impossible with the available data," the report says.