Google Trends gets tested in 2022 Philippine Elections

In the last few weeks leading to the May 9 national elections, Google Trends has been consistently in front of the news as a more accurate predictor of election results, especially for top national positions such as President and Vice President.

Google Trends is a website by Google that analyzes the popularity of top search queries in Google Search across various regions and languages. The website uses graphs to compare the search volume of different queries over time.

Many digital experts point out that big data is a much better predictor compared to traditional pollsters due to the massive amount of data that it processes in real-time. Will millions of searches for presidential and vice-presidential candidates, the numbers are far more representative of the voting population compared to the random sampling of 1,200 to 3,000 registered voters.

Google Trends as of May 9, 2022 11:30pmMany experts claim that Google Trends is a better predictor of election results since it measures interest with a likely intent to vote for the subject of the search. It is also hard to manipulate as Google itself sanitizes the data from possible spamming.

There were also claims that survey firms can be bought or are aligned with certain political personalities, tainting the credibility of their results. The sampling methodology of the surveys was likewise scrutinized and do not represent the actual social demographics of the population.

Google Trends as of May 9, 2022 11:30pmResearchers even pointed out that Google Trends have accurately predicted presidential elections in other countries such as the United States, Brazil, Spain, Canada and, more recently, France.

However, Google states the contrary to this assertion:

Is Google Trends the same as polling data?

Google Trends is not a scientific poll and shouldn’t be confused with polling data. It merely reflects the search interest in particular topics. A spike in a particular topic does not reflect that a topic is somehow “popular” or “winning,” only that for some unspecified reason, there appear to be many users performing a search about a topic. Google Trends data should always be considered as one data point among others before drawing conclusions.

The website, FactRakers.org, also corrected claims that Google Trends predicted Noynoy and Binay winning the 2010 elections as well as Duterte and Robredo winning the 2016 national elections.

In 2010, Binay was the 2nd most searched vice-presidential candidate, and Mar Roxas was the top search VP candidate. Binay won in 2010. In 2016, Robredo was only the 3rd most searched vice-presidential candidate whereas Sen. Trillanes was the 2nd most popular and Sen. Bongbong Marcos led the popular search.

Ultimately, the 2022 election results will validate the theories in favor of Google Trends as a more accurate predictor. As of 2:55am, May 10, 2020, the partial/unofficial tally by the COMELEC transparency servers have processed 90.91% of the precincts.

That’s about 49.17 million votes out of the estimated 65 million voters, and here are the partial results compared to the numbers predicted by survey firms and Google Trends:

We also averaged the numbers of the 4 survey firms to give us a better picture of how they fared with their predictions. Google Trends was completely off-mark for the percentages between Marcos and Robredo for the presidency, and while it correctly predicted that Sara will lead the polls, the percentages between Duterte and Pangilinan were also off.

Meanwhile, all 4 pollsters correctly predicted Marcos and Duterte to lead the race, mostly within an acceptable margin of error.

Conclusion

The 2022 Philippine elections has shown that Google Trends is not an accurate predictor of presidential and vice-presidential results. The discrepancies are way off the mark. Well, at least in the Philippines.

In contrast, survey firms have once again proven their time-tested credibility to predict election results for many years. Random-sampling 1,200 registered voters, when done right, are a much better barometer to check the pulse of the voting population at a given time frame.

Again, we quote Google when it said “Google Trends is not a scientific poll and shouldn’t be confused with polling data.“.

Author’s Note: Since the figures of election results used in this comparison are based on COMELEC’s partial/unofficial tally from their transparency servers, we will be updating the figures once the official count is available.

• FactRakers is a Philippines-based fact-checking initiative of journalism majors at the University of the Philippines-Diliman.