Should we trust the polls?
We’re smack-dab in the middle of #Halalan2022, and the latest surveys from public opinion polling bodies are showing that BBM has pulled way ahead of the competition in the presidential race—but newer and (supposedly) more reliable methodologies say otherwise.
Survey results are in: BBM is leading
Pulse Asia’s most recent survey released on February 13th showed that 60% of respondents chose Bongbong Marcos Jr. as their preferred presidential bet, with VP Leni Robredo lagging behind in second place at 16%, Manny Pacquiao and Isko Moreno at 8%, and Ping Lacson at 4%.
As they often do during election season, the polls themselves are coming into question, specifically, how they have remained static despite fervent campaign efforts from all candidates. Besides Marcos going up by 7% since Pulse Asia’s previous survey from December 2021, the other candidates only moved up or down by 2%, well within the survey’s margin for error, which is statistically questionable, to say the least.
Survey says: Google Trends might be more reliable
Surveys have always been the go-to method for gauging public opinion, most especially during election season, but they’ve always been all kinds of sus. I, and everyone I’ve ever asked, have never met a single person who’s been surveyed by SWS or Pulse Asia—I’m willing to bet you haven’t, either.
On March 9th Google launched an official search tracker for the Philippine Elections, revealing some interesting facts about the search trends for the Presidential Candidates.
Google’s Philippine Elections Search Tracker is updated based on real-time searches. It basically shows the top queries, topics, and notably, the most highly searched presidential and vice-presidential candidates. The page also provides information on top policy issues in various parts of the country, covering topics such as education, drugs, and poverty, among many others.
What’s interesting is that the search tracker, ever since it was launched on March 9th, has shown that Robredo is the most Googled candidate by far. At the time of writing this (March 15, 2022), she makes up 47% of Google searches, with Marcos coming in second at 26%.
But these are just Google searches. Do they have any bearing on a candidate’s actual chances at winning come election day?
Yes, they almost certainly do
Data scientists and behavioral science scholars are now saying that monitoring Google Trends is an almost-too-reliable means of predicting presidential elections, because unlike traditional survey methods, it’s free from posturing, and looks at people’s true intentions based on their online behavior. Local political columnists are also backing their viability for predicting election results.
One study tried to test this by taking the results from the past four elections in the United States and the past five in Canada, and comparing the winners of the elections to the most highly searched on Google. This method accurately predicted the winners in all nine elections. This includes election winners from highly volatile campaign periods where trolls and misinformation ran rampant—such as Trump’s win in 2016.
Interestingly, Gallup Polls predicted that Romney would win over Obama in 2012. In 2016, Reuters predicted with 90% certainty that Clinton would win over Trump. Both public opinion surveys ended up being wrong, but predictions based on Google Trends were right.
To cite a local example, Google Trends also accurately predicted Duterte’s win in 2016. So, based on data taken from the last five weeks, Google Trends predicts that VP Robredo will take the Presidential seat.
The caveat: this method has been fool-proof so far, but researchers say that it still needs to be tested in more elections before we can conclusively call it accurate. It’s also worth noting that in the coming weeks, the trend could very easily shift.
It’s too early to celebrate or mourn (whichever side you’re on). For now, we wait and see—and actually show up on election day to vote for our candidates, and hopefully, get the results we want.
