Professor W. Joseph Campbell, author of ‘Disappearing in a Gallup: Poll Failure in the US Presidential Election’, in an interview with BBC News Brazil, says that political attempts to punish research institutes are ‘pointless’ and seek the segment’s ‘new gold standard’.
“It’s no exaggeration to say that the election polls market is going through a crisis,” says Professor W. Joseph Campbell, who has devoted himself to the study of these polls in the United States in recent years, to BBC News Brazil.
Book author Getting Lost in a Gallup: Voting Failure in the US Presidential Election (Something like “Lost in the Galup: Poll Fails in the U.S. Presidential Election”) and Campbell, professor of communications at the American University of Washington, says the always difficult and costly job of conducting election polls has faced new challenges in recent years—because people are reluctant. participating in conversations because of their political orientation or because they can no longer stand unsolicited phone calls from unknown numbers.
Whatever the reasons, the reality is that US research institutes are experiencing an unprecedented series of mismatches between expectations about voter behavior and the reality revealed by polls.
“In 2020, poll failures were the worst in 40 years since 1980, in Ronald Reagan’s election over Jimmy Carter. Joe Biden gained nearly 4.5 percent of the popular vote in 2020, with some coming pretty close to that. But most have stayed away, with some estimating that he’s 10 points ahead of Donald Trump,” says Campbell, who has worked as a journalist for 20 years and has a PhD in communications. “In the last five presidential elections, the polls underestimated the Republican party vote as a whole.”
Image: EPA
In Brazil, the debate over election polls intensified after the first round on Sunday 10/10. The distance between candidates Luiz Inácio Lula da Silva (PT) and Jair Bolsonaro (PL) was less than the previous day’s polls had shown. The Datafolha and Ipec polls showed a nearly 14 percentage point advantage for the PT over the current president – Datafolha estimated the valid votes for Bolsonaro at 36% and for Ipec at 37%, which is roughly 42 million valid votes on a simple account. The current president actually had 51 million valid votes, well above the polls’ margin of error of two percentage points, or 43.2%.
Additionally, in the state government dispute, current governor Cláudio Castro (PL) won the election in the first round in Rio de Janeiro, with main polls predicting a runoff against Marcelo Freixo (PSB). ). In Sao Paulo, Tarcísio de Freitas (Republicans) took first place, while second behind Fernando Haddad (PT) in the polls – they are now fighting for the second round. Also in Rio Grande do Sul, when the polls reversed the rankings, Onyx Lorenzoni (PL) took first place and Eduardo Leite (PSDB) took second place. Lorenzoni and Leite will also compete in the second round.
This Tuesday (10/04), Attorney General Anderson Torres stated that he has asked the Federal Police to open an investigation into research institutes. Another Bolsonaro government minister, Communications’s Fábio Fábio, urged supporters of the president not to respond to election polls, which he accused of being “weapons of manipulation”.
BBC News Brazil video interviewed W. Joseph Campbell on Monday (10/03), but the next day asked the American professor for a position on the penalties proposed by politicians against research institutes.
Referring to the text of the American Constitution that guarantees free elections, the pollster said, “It seems pointless to recommend penalties for electoral polls that miss the mark – and it’s probably political arrogance. In the United States, such attempts quickly clash with First Amendment protections.” expression.
Shortcomings and new formats
Campbell says research institutes’ response to the recent crisis in the US has been split in two ways.
“Some well-known institutes have actually decided to withdraw from the polls, including the company Gallup, which is probably the best-known name in the polling industry. George Gallup founded the institute in the 1930s and it has become the most important, but from a series of tough elections that culminated in 2012. Then Gallup came out of this industry,” the professor said, noting that the institute predicted that year Mitt Romney would be slightly ahead of Barack Obama, who eventually won the election. .
Since then, the institute has conducted other types of research, such as Americans’ values, behavior, and consumption preferences. The same path was chosen by the Pew Research Center company, which announced in 2016 that it would no longer project votes in national and state elections.
Picture: DESCRIPTION
“But the most common response from institutes was to rigorously review their work and try new approaches. This experiment was very impressive. This segment is looking for a new gold standard,” he says, explaining that institutes are testing and combining data. social media posts, online recruitment of repeat interviewers, and demographic surveys.
The researcher explains that random-dial phone surveys are the gold standard of surveys in the US, but have a very low response rate.
“With the advent of telemarketing and caller ID, people don’t want to answer a call from an unknown number.”
“The internet has become dominant in the way polls are conducted. That may be for better or worse, but either way, the internet is still not considered the gold ballot standard.”
Another example of this review that the US election research market is going through is a report published in 2021 by the American Association for Public Opinion Research, which represents institutions and researchers in the industry. The report, signed by nearly 20 experts, examined “poll errors” in the 2020 US presidential election. The document states that a single, definitive reason for the failures cannot be cited – a real possibility, but difficult to prove, the refusal of a significant portion of Republican party supporters to participate in the polls encouraged by the former President’s speeches. Donald Trump is against these polls.
According to W. Joseph Campbell, conservative voters’ greater “doubt” about election polls has to do with the growing lack of confidence in the press in this political arena.
“The links between research institutes and the press can be very strong. In the US, major media companies such as the New York Times, Washington Post and CNN have their own survey operations or collaborate with other companies in the field.” .
In Brazil, the Datafolha institute is part of the same business group as the Folha de S. Paulo newspaper, and the Globo group frequently commissions election polls from specialized institutes.
Self-confident
In addition to the reluctance of potential respondents, which is a relatively new phenomenon, Campbell reminds that surveys are subject to always-existing problems.
“I show in my book that there are records of mistakes made in election polls as early as 1936,” he says, reminding that there are limitations to comparing scenarios in different countries because electoral and political systems can be very different as well as very different. socioeconomic contexts.
“It cannot be ruled out that voters have changed their minds in the last days before the election, or that undecideds have chosen a particular candidate. These late changes are very difficult for polls to catch.”
After the first round in Brazil, research institutes offered possible explanations for the discrepancies between the percentages of voting intentions they achieved and the concrete votes candidates received, such as the last-minute vote migration from Ciro Gomes (PDT) and Simone Tebet. A high abstention that may have further affected (MDB) and Lula voters.
Image: GETTY IMAGES
Representatives from Ipec, Datafolha and Quaest insisted there were no “errors” as the polls were portraits of voter intent during the interviews, not a predictor.
- How do research institutes explain Bolsonaro underestimated in polls?
- Hypotheses explaining Bolsonaro’s votes that the polls did not predict
But for Campbell, the purpose of polls is to make their results as “credible” as possible, even closer to the vote.
“Polls aren’t always wrong, sometimes they’re right, sometimes more or less. But they’re wrong quite often, and surveys are considered a risky business,” says the American.
“I think those in the research industry in particular will understand that their field is going through a lot of challenges right now. It’s not easy to do research, to do it well and successfully. It’s very expensive, time consuming, and response rates are very, very low. … So that’s the playing field these days. There are many factors that make it very difficult.”
Researcher, author of other books such as Getting it wrong: Debunking the Biggest Myths in American Journalism (“The Mistake: Debunking the Greatest Myths in American Journalism” in its free translation) also warns that errors in election polls are often associated with journalistic errors.
“Journalists rely heavily on polls, and poll results tend to shape the narrative of political campaigns and elections,” says the professor.
– This text was published at https://www.bbc.com/portuguese/geral-63140085.
Did you know that the BBC is also on Telegram? subscribe to the channel.
Have you watched our new videos on YouTube?? Subscribe to our channel!
source: Noticias
