(Dan Tri) - Polling agencies facing many challenges ahead of the US election have also made adjustments ahead of this year's election.
Poll results often do not fully represent Americans' views ahead of elections (Photo: Reuters). When it comes to polling before the 2016 US presidential election, many Americans would consider it a huge mistake. Most polls at the time said Hillary Clinton would win. The winner was Donald Trump. In fact, polling organizations underestimated Trump’s support in battleground states. However, the New York Times points out that national polling figures are quite similar to the popular vote, which Hillary Clinton is ahead of. With a close election like 2024, polling is even more difficult. Some polls show Harris leading, others say Trump is ahead. "The truth is that polls - and the models that rely heavily on polls to predict outcomes - cannot confidently predict what will happen on November 5," Brian Klaas, associate professor of global politics at University College London (UCL), wrote on the Atlantic news site.Difficulties surround In every election, poll results have a certain degree of discrepancy with the final result. This is inevitable because polling units can only estimate who will actually vote. Besides, many people only make decisions when they get to the ballot box. Some unexpected events can also appear at the last minute. From the lesson of 2016, political analysts have pointed out the weakness of public opinion polls is that they cannot fully identify "blind spots" in voters' views. Many people feel ashamed and do not dare to admit that they will vote for Mr. Trump, the most controversial candidate in modern American history. As a result, the data collected is inaccurate. In fact, the New York Times analysis also found a higher level of bias in elections with Mr. Trump. “Trump’s name on the ballot, for some reason, makes it harder to poll,” said Berwood Yost, director of the Franklin & Marshall College Polling Center in Pennsylvania. In addition, sampling is sometimes imperfect. In the 2016 election, voters with lower levels of education were not fully accounted for, in part because they responded at lower rates than those with higher levels of education. The rate of people answering the phones of polling centers also tended to decline. "People aren't answering the phone. Ten years ago, you might have to call 20 people to get the person you want. Now, it's double that: You have to call 40 people to find the person you want. So polls take longer and cost more money," Rachael Cobb, a political science professor at Suffolk University, told CNBC. The polarization of the electorate also makes polling more difficult. Lonna Atkeson, a professor of public opinion at Florida State University, said she has received emails saying she won't participate in polls and even accusing her of "brainwashing" children. Data from the past four decades analyzed by the New York Times shows that polls can be biased in both parties. Accuracy levels also vary widely even in close elections. For example, after a big 2020 upset, pollsters are making relatively accurate predictions ahead of the 2022 midterms. Ahead of this year’s presidential election, pollsters are looking for ways to improve their predictions, some looking at different groups of voters, while others are more concerned with impatient voters who don’t answer all the questions. Candidate Kamala Harris and opponent Donald Trump (Photo: AFP). Don Levy, director of the Siena College Research Institute (SCRI), which conducted the poll with the New York Times , said that if the group of voters were included, the poll would have shifted "about 1.25 percentage points toward Trump." This year, SCRI allocated more interview slots to voters classified as "rural, likely to vote for Trump." "If you think of them as M&Ms — Trump voters are red — we added some red M&Ms to the jar," Levy told CNBC.Which states are most likely to be wrong? Polling accuracy also varies from state to state. Nathaniel Rakich, an election analyst for the website FiveThirtyEight , calculated the error in the 21 days before the election in each state since 1998. The conclusion is that some states are more accurate than others. With an average margin of error of 3.3 percentage points, national polls are the most accurate. This is not surprising, given that the country’s larger population makes it easier to sample. In addition, Senate and governor races, which are limited to one state, are more likely to be inaccurate than presidential elections. When it comes to individual states, Colorado, Virginia, and Oregon are the most accurate. However, for political observers, this data is of little use, as all three states are predicted to vote for Ms. Harris. What they will pay more attention to are the results in the four battleground states in the “Sun Belt” of the western and southern United States: Nevada, North Carolina, Arizona, and Georgia. These four states are among the places with the lowest margin of error, ranging from 3.8 to 4.1 percentage points. If we only count from 2016 to now, Nevada's error rate is only 3.3 percentage points, lower than the national rate for the same period (3.9 points). This is relatively surprising because Nevada is considered a difficult state to poll because people here tend to move more frequently and work overtime more than other states. In contrast, polling data in the Northern battleground states is less reliable, although still more accurate than average. Since 1998, the error rate in Pennsylvania and Wisconsin is 4.6 percentage points, while in Michigan it is 4.9 percentage points. Wisconsin in particular is known as a state that is very prone to recording erroneous results. In 2020, ABC News/Washington Post once rated Mr. Biden leading this state by 17 percentage points. In the end, Mr. Biden only beat Mr. Trump by 0.83 percentage points. However, Wisconsin is not the most difficult state to poll. The top three states on this index are Oklahoma, Wyoming, and Hawaii. Since 1998, the average error in Hawaii has increased by 10.4 percentage points. This may be due to Hawaiians' low level of political interest, as well as their ethnic diversity, which makes sampling more difficult. "What makes states easy or difficult to poll? Size and population are certainly key factors. However, the data also clearly show that experience is an important factor. The most accurate polling states are the states that are polled most often," Rakich said.
Comment (0)