Can we still trust election surveys?
Monday's midterm election was capped off by a night of surprises. As more ballots were counted, the gap between pre-election surveys and partial results grew wider.
The biggest unexpected shifts involved senatorial candidates Bam Aquino and Kiko Pangilinan, who ranked 12th and 17th in Pulse Asia Research, 15th and 16th in Social Weather Stations, 12th and 18th in Octa Research, and 12th and 15th in WR Numero.
As of writing, Aquino is ranked 2nd in the partial results, garnering over 20 million votes. Pangilinan is 5th, with over 15 million votes. The closest survey to reflect this was PUBLiCUS Asia, which ranked Aquino 2nd and Pangilinan 3rd.
“Ang survey ay isinagawa tatlong linggo bago ang eleksyon, but that doesn’t exactly explain it,” Pulse Asia President Ronald Holmes said during an interview with ANC on Tuesday. As we wait for more comprehensive data, Holmes said a possible theory is the decline in the fill-up rate during the April survey. This means surveyees did not fill up the entire top 12, with the majority reporting that the number of candidates they will vote for may still change.
According to Rye Ranjit, Professor at the University of the Philippines Diliman’s Department of Political Science and Octa Research Fellow, Aquino’s win in the vote-rich Lucena-Lingayen Corridor in Luzon was reflected in Octa’s survey, where the candidate placed 12th. “He was never not in the survey,” Ranjit told GMA News. “He optimized and maximized. Last minute, nag-organize siya.”
Ranjit added, “In Central Visayas, which is the stronghold of the Dutertes, Bam Aquino was in the top 5. He ran a very good campaign [that] resonated with a broad base of supporters.”
The limitations of pre-election polls
Political analyst Jean Franco told PhilSTAR L!fe that pre-election surveys provide only "snapshots" of opinion when they are conducted, making it uncertain how views evolved in the following weeks.
“What happened in the past two or three weeks is that those surveys taken, we don't know about. That's the mystery,” he said.
In fact, the monthly surveys leading to the final April poll foreshadowed the eventual falloff of candidates like Ben Tulfo, Manny Pacquiao, Willie Revillame, and Abby Binay. Holmes said that while these candidates appeared in their final survey in April, they were already gradually sliding down the Magic 12 from the early rounds in February and March.
Another upset was senatorial candidate Atty. Rodante Marcoleta. He was ranked 6th in partial results but only made the Magic 12 in Octa and PUBLiCUS Asia surveys. Dr. Mahar Mangahas of SWS said in a GMA News interview that while it’s convenient to attribute this to an endorsement from the Iglesia ni Cristo, it is not enough to grant this big leap. “5% lang [ng total voters] ang INC voters,” he said.
Holmes further theorized that Marcoleta, who is part of the “Duter10” slate with survey frontrunners Bong Go and Bato Dela Rosa, may have gotten support from areas dissatisfied with the current administration.
This seems probable as only six senatorial bets from Alyansa para sa Bagong Pilipinas, formed by allies of President Bongbong Marcos, made it to the top 12.
“Usually when we look at midterm elections, dahil may advantage [sila], 8-9 of the administration’s candidates ang pasok sa Magic 12,” said Cleve V. Arguelles, WR Numero President and CEO and Assistant Professorial Lecturer in the Department of Political Science and Development Studies at De La Salle University Manila.
The six Alyansa candidates who made it are Pia Cayetano, Ping Lacson, Lito Lapid, Tito Sotto, Erwin Tulfo, and Camille Villar. Arguelles added in an interview with ABS-CBN News that this poor performance from administration-backed bets indicated dwindling trust in the current president.
“We learn as we go along,” explained Dr. Mangahas of SWS. “So now we know how much things can change at the last minute.”
Reevaluating survey methods
Political analyst Dennis Coronacion told L!fe that survey firms must recalibrate their methodology. “Kailangan mas maging nuanced ‘yung survey instruments nila.”
Multiple variables can affect how surveyees respond to pre-election polls. “Surveys can be affected, for instance, by social desirability bias. Kung ano ang tingin mong gusto ng nakikinig, 'yun ang ilalagay mo,” said analyst Franco.
The research firms assured that they are conducting internal studies to understand and properly address the survey gaps. “Sa Octa, we need to get the resources to run surveys closer to the elections, and we don’t have those resources,” Ranjit said. “All the survey firms could have gotten it right if they had more rounds. The election we’re seeing is dynamic.”
Holmes reminded that survey responses are confidential, and it can never be 100% definitive. “We [have a] 95% confidence interval. Sa 100 survey na gagawin namin, limang beses maaaring may weird na resulta. Hindi ibig sabihin noon na walang integrity, na hindi na dapat pagkatiwalaan.”
This contrasts with the criticism leveled by senatorial candidate Atty. Luke Espiritu, who on Monday evening described pre-election surveys as "mind-conditioning" and deemed SWS and Pulse Asia's polls "beyond erroneous" and "anomalous" on Facebook.
In response to such criticisms and the recent election surprises, SWS’ Dr. Mangahas assured that surveys are not just done to collect senatorial preferences, but also overall voter sentiment, including what issues they consider when choosing candidates. Perhaps the surprises of the midterm elections will urge research firms to update methods to better reflect the shifting realities of Filipinos.
