The election polling stages

Election polling has three stages: the first one is to help prospective candidates and their parties decide who should run for office, the second after the candidates and their affiliations are already known, and the final one before election day. In the first stage, the polls will work with alternative scenarios about the competition. Might it be: (1) A versus B versus C, or (2) B versus D versus E, or (3) C versus E versus F? Who might run for senator? Would there be 25 or 50 candidates?

The first stage is over, since the period for filing certificates of candidacy for the May 12, 2025 election ends next Tuesday, Oct. 8. Political parties have started revealing their candidates. Candidates are deciding on their parties.

The second stage is during the campaign. With the candidates already identified, the polls go into a second stage of tracking the actual horse race, rather than alternative hypothetical ones. The earlier this is done, the better for the campaigners.

For the May 1987 election for 24 senatorial seats (the first under the new Constitution), the SWS-Ateneo sociopolitical poll found, as of March 1987, a tie of 12-12 between the candidates of President Corazon Aquino and their opponents. Alarmed by the finding, campaign manager Paul Aquino told his people that they could not sleep anymore. The party buckled down to intense campaigning—calling on President Aquino herself to join the rallies—and managed to improve its score to 22-2 when the election was done, which was called “Cory magic.” (While our critics gleefully claimed that we had predicted 12-12.)

In the course of a campaign, the standings of the candidates will be known to parties that have their own private polls, but not to the general public, unless some media companies sponsor polls for broadcast and/or publication. This is a fine public service, the cost of which is potentially recoverable through advertising income. Social Weather Stations (SWS) was fortunate to have media sponsors in past elections, and thus have horse race records now; what 2025 will bring is unknown.

In my experience, there have not been radical changes in senatorial candidates’ standings during the campaign period—the surge of Antonio “Sonny” Trillanes IV in early May 2007 was exceptional. Opinion polls hardly affect the preferences of voters. Data from exit polls—which are done after the voting but before the official announcement of the results—show that most voters aren’t aware of the opinion polls. Among those that are, almost as many (and sometimes more) like to go for the underdogs rather than the favorites.

The ones mainly affected by the horse race polls are the political funders. One highly favored Filipino candidate once bragged to me that so many VIPs were offering him campaign funds that he decided to give them their money back. The late political guru Antonio P. Gatmaitan (former campaign manager of Danding Cojuangco), knew that American campaigners counted X thousand dollars of contributions for every percentage in a candidate’s voting rate.

SWS battled in 2001 against legislation that suppressed pre-election polling and quickly won in the Supreme Court (GR 147571). The true inspiration for the unlamented portion of the Fair Election Act was politicians’ greedy struggle over home-stretch contributions to their campaigns.

The final stage is the prediction. Whatever is a poll’s final reading before election day is its de facto prediction. The pollster’s challenge is to field the interviews as late as possible—to catch the voters’ final choices—and yet process the responses in time to publish the findings before election day.

The final polling result, compared to the official vote count, is what goes into the pollster’s track record. For the May 14, 2001 senatorial election, the SWS May 4-7, 2001 survey correctly identified 10 of the 12 winners, with a rank correlation of 98 percent. For the May 10, 2004 senatorial election, the SWS May 1-4, 2004 survey correctly identified 12 of the 12 winners, with a rank-correlation of 98 percent also. For the May 14, 2007 senatorial election, the SWS 2-4, 2007 survey correctly identified 11 of the 12 winners, with a rank correlation of 97 percent.

Reference: M. Mangahas, “The challenge of election surveys: SWS experience in the Philippines,” Second Cairo International Conference on Public Opinion, 11/8-10/2009 (https://rb.gy/mx359w).

————–

Contact: [email protected].