As of this writing, there’s more than two months of campaigning left, and recent polls will have you believe that Ferdinand Marcos Jr. (BBM) will be the landslide winner. We disagree.
Source: https://www.facebook.com/isupportbbm/posts/349456927026748
Surveys can be wrong at predicting the true voter preference. We have seen too many upsets, especially in recent times, where the electoral outcomes defied the industry’s prediction.
Remember when surveys predicted a 90–95% confidence of a Hillary Clinton win against Donald Trump in the US elections? The survey results were flawed by selection bias, framing and the “shy Trump” effects. An election is not a statistical model but a live thing and can only be determined after the last vote is cast.
At least 6 surveys/polls (see above) point to a BBM victory by a large margin. Each of these surveys is based on a sample size of between 1500 to 2400 respondents only. Statisticians generally concur that this sample size is enough to represent the voting population. Except for one of the polls, most were done in January. Thus, these polls may have largely failed to capture changing preferences in February.
We believe it is possible to get a better “pulse”. Using more than 50 million social data points, our analysis reveals a much tighter race. A Leni victory over BBM is very possible, with only minor tactical changes needed from Leni’s campaign team.
Here is our reasoning and data for you to decide. We measure campaign progress on two fronts: Engagement and Net sentiment.
Engagement measures a candidate’s potential voter base, those voters who have an emotional reaction to the candidate, or the range of people having enough interest to be voters. Higher engagement scores mean higher name recognition, and consequently a well-informed opinion of the candidate.
This shows that BBM has a higher recall. Indeed, in elections, familiarity does play an important part in winning. This Publicity chart further confirm the numbers. It tracks the number of posts made in favour of each candidate where BBM has the commanding lead.
We now drill down to the sentiments of the posts. More positive sentiments bode well for the candidate. Conversely, more negative sentiments spell doom for them as well. Here, we track users’ reactions to the candidates’ posts as well. We use the emoticons LOVE, SAD and Angry. Interestingly, this paints a much closer battle.
Net sentiment tracks the cumulative social sentiment change (positive-negative social sentiment) over time. Higher net sentiment indicates that potential voters would likely vote for the candidate.
This chart favors Leni in this “two horse race”. For the period between Jan 1 to Feb 27, 2022, the Leni camp widened her lead over BBM. Barring unforeseen events, the current trajectory indicates the Leni camp is more effective at their online ground war.
Two additional things to highlight.
First, we note that BBM’s net sentiment scores slowed after his no-show at the Jessica Soho Presidential Interview. While BBM was more talked about during this time, the net sentiment did not go in his favour.
Second, Isko Moreno’s increase in Net Sentiment scores came at the expense of BBM. Political strategists might use this insight to recalibrate their strategies.
Roger Do, Autopolitic CEO shares his analysis of the campaign space as a 2X2 grid. When both sentiment and engagement are high, candidates are in a good place. When they are in the same space, the race is contentious. Campaign space positioning of the candidates are illustrative, not computed or to scale.
When candidates’ metrics are both low, they have NO chance. Leni and BBM are both in the same competitive space. While BBM has a slight edge over the potential voter base, Leni has a higher net sentiment change, indicating her potential voters are converting at a higher percentage than BBM, while BBM enjoys higher name recognition.
However, Roger Do, does have a fearless forecast. While the winner is too close to call, he CAN SAY who WON’T win: Senator Lacson and Senator Pacquiao. The following are Roger’s insights into the current campaign:
Senator Lacson is the most qualified candidate in this year’s race and with the most qualified running-mate, has committed the unforced error of being unprepared. The campaign operation started very late and was without a coherent strategic plan. The policy-heavy orientation failed to connect with the voters, and his coalition of responsible politics was not put to good use. A very disheartening campaign performance by an intellectual and effective manager of national affairs.
Senator Pacquiao has been knocked out before the bell was rung. It was obvious from the beginning that his ambition to serve the country as the president is ill-serviced by his campaign even a year ago when the expected rebranding of him from a fighter to a political leader failed to appear. This self-inflicted wound was further compounded by the party’s inner-scheming and reliance on relief goods strategy that would have worked on the regional election but not at the national level.
With Mayor Isko’s social media presence and performance, the model predicted a vastly larger base by now. But the mayor’s team leadership problems delayed any serious campaigning. Along with BBM and Leni, his campaigning has an impact on the voters. We see his performance in both engagement and net sentiment increased, moving him slowly out of the also-ran and towards contention, but with only two months left, normal campaign modes won’t be enough to move him into the contender’s zone.
Any enterprise that deals with divining the future is an exercise in hubris, and fate loves to upset the hubris of men and women. Or do the surveys serve another purpose? One of creating a “Bandwagon” effect for their candidate? This is cycle of getting more people to support the “likely winner”? Check out the Barabasi-Albert Model using a preferential attachment mechanism.
The real contribution of a survey is to help the campaign adjust its tactics. It was never intended to be consumed by the public. The work of data-driven prediction offers more nuanced hedges than assured certainty, but reading the press release, the reader would be forgiven to think tomorrow is known and the past understood.
There are limitations that readers should know about the use of Data Analytics and Facebook Social Media Sentiments. (With our thanks to points raised by Coordinator of Action for Economic Reforms Men Sta Ana):
1. Not all voters are users of Facebook
2. Not all Facebook users are registered Philippine voters.
3. Not all positive sentiments will translate to votes
4. Not all reactions are authentic. Data privacy restrictions prevent verification of Facebook accounts.
Running Totals: Social Sentiment — Net is a cumulative report of the net positive Facebook reaction since Oct 1, 2021. IT shows Leni having a substantial lead ever since November.