Facebook data shows who is leading presidential candidates


Full disclosure: I’m one of the analyst of this report using Facebook’s full data feed from Oct 2 to 15, 2021. The analytics was done by ADDS (Alpha Distinct Development Solutions Inc).

The graphic below show how Facebook users regard the various presidential aspirants. It seems to show that Leni Robredo has a distinct lead over Bong Bong Marcos.

Among FB users, Leni Robredo is more popular. She has more mentions, more shares, and more engagement than Bong Bong Marcos. But as political strategists will tell you: converting popularity to votes is another matter altogether. Indeed Reddit users point out that in the last election, the Otso Direcho also had the online numbers, but lost in the polls.

What I see is the elections shaping up to be a two horse race between Leni Robredo and Bong Bong Marcos (at this point). The other aspirants may most likely form alliances with one of these two.

But before the Leni camp celebrates, we show the second chart:

How does one calculate Positive/Negative or Swing reaction? The report doesn’t use NLP (Natural Language Processing) to track users emotions. Instead it is based on the 6 emotive indices on Facebook to gauge user sentiment. The are the “haha” emoticons. Each user can only use ONE emotion therefore it is harder to skew the index since it is not possible to pick multiple emoticons.

The key variable to note is the “Swing Reaction”. According to Roger Do, Qsearch founder, these are the “interested” but “not yet committed” voters. In other words, there are still a lot of undecided voters. There is still a long way to go before the May 2022 elections. Each of them could change their votes on election day.

As a sort of simple peer review, Anton Sheker shared this other analytics that uses other online sources, and also confirms the same trend:

Leni still leads

Please note: The FB users data also include non-voters. These are those that are underaged, and/or not a citizen, or did not register to vote. There is no way our platform can infer and filter these out.

Please like/follow/subscribe this to be notified of updates in the succeeding months.