(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 reflect Americans' views ahead of elections (Photo: Reuters). When it comes to polling ahead of 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 data is 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 showed Harris leading, others said Trump was 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. In addition, many people only make their 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 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. And sampling is sometimes imperfect. In the 2016 election, voters with lower levelsof education were underrepresented, in part because they responded at lower rates than those with higher levels of education. Polling centers also tend to have fewer people answering their phones. "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 get 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 wouldn't participate in polls and even accusing her of "brainwashing" children. A New York Times analysis of data from the past four decades shows that polls can be skewed 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 interested in 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 that group of voters were included, the poll would have shifted "about 1.25 percentage points toward Trump." This year, SCRI allocated more interviews to voters who were 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 FiveThirtyEight , calculated the error in polls in the 21 days leading up to 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 tend to be the most accurate. This is not surprising, given the country’s larger population, which makes it easier to sample. In addition, Senate and gubernatorial 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 tend to be the most accurate. But for political observers, this data is of little use, as all three states are predicted to vote for Harris. They will be more interested in the results from 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 the present, Nevada's margin of error is only 3.3 percentage points, lower than the national average over the same period (3.9 points). This is quite surprising because Nevada is considered a difficult state to poll because people here tend to move more frequently and work more irregular hours than other states. In contrast, polling data in the Northern battleground states is less reliable, although still more accurate than average. Since 1998, the margin of error 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 error. 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, Hawaii’s average error has increased by 10.4 percentage points. This may be due to Hawaii’s low level of political interest, as well as its ethnic diversity, which makes sampling more difficult. “What makes states easy or difficult to poll? Size and population are certainly key factors. But the data also clearly show that experience matters. The states that poll most accurately are the states that are polled most often,” Rakich said.
Comment (0)