Surveys of elections past

I REMEMBER an old cartoon by Jules Feiffer in which a couch potato, in front of a TV, says, “I like to watch the opinion polls so that I’ll know how to think.” It’s just as weird if the potato says, “I like to watch the opinion polls so that I’ll know for whom to vote.”

Actually, opinion polls have almost no effect on how people vote. We know this from having surveyed it. A few say surveys influence their vote, but that’s not undemocratic. It’s their free choice. By the way, many shifters go for the underdog, not the leader, so the net bandwagon effect can be zero, or even negative.

Eighty-five percent or more say they vote according to their conscience, whoever may win. Only 15 percent or less say they waste their vote by not giving it to the probable winner.

Surveys do have an effect on the support base—money-wise and people-wise—of the candidates. This is not undemocratic. But only unbiased surveys can have this effect. Political backers are smart; they know which surveys are genuine.

Biased surveys die a natural death. Within six months of the 1998 presidential election, the Facts Base company—which had insisted that Joe de Venecia had a good chance of defeating Erap Estrada—could no longer be found.

In mid-1985, the Bishops-Businessmen’s Conference (BBC) survey found that two-thirds of Filipinos disliked Ferdinand Marcos’ power to legislate by decree, and two-thirds also disliked his power to detain by fiat. When he called for a snap election, Marcos bluffed that the BBC survey had an item predicting he would win.

I wasn’t able to do an opinion poll before the February 1986 snap election, despite finding a sponsor whose only condition was that the poll be identified with BBC. The BBC chair, the late Vicente “Ting” Jayme, endorsed it to the BBC board, but it declined to accept the responsibility.

Ting explained to me that BBC would have to publicly release the survey, regardless of findings. Even if BBC tried to keep it secret, it was bound to be leaked. If it showed Marcos in the lead, it would discourage Cory Aquino’s camp. On the other hand, if it showed Cory in the lead, then Marcos could even cancel the election altogether. Better, Ting said, not to know what to expect.

The official count of the snap election said that Marcos won, whereas Namfrel’s parallel count said that Cory won. SWS’s first opinion poll, done in May 1986 in partnership with Ateneo de Manila University, found 64 percent of Filipinos saying they had voted for Cory.

A count of the votes of the full population validates (or else repudiates) the count in a sample survey, not the other way around. Provided, of course, that the full count is honest. Therefore the honesty of the official counting system should be constantly inspected. Parallel counting, Namfrel-style, is still the most reliable way to do it.

The main value of an exit poll is not in beating the official count to the media but in learning more things about the vote, by asking, for instance: “When did you make your final choice on whom to vote for?”

In the historic SWS-TV5 exit poll of May 10, 2010, 27 percent said they made their choice for president only in May itself, 16 percent decided in March or April, 51 percent decided in February or earlier, and 6 percent didn’t answer.

Thus, from now up to February, I think only half of the electorate will have chosen their candidate. By the end of April, perhaps it will be three-fourths. Perhaps one-fourth of the electorate will decide only in the last 10 days.

In preelection surveys, almost every respondent can name a choice of candidate, but it is obviously a tentative choice. Voters have the right to be tentative. The so-called “undecided” in these surveys are merely those without any candidate yet.

Here is the survey leaderboard in past election campaigns:

In 1992, Fidel V. Ramos was usually slightly ahead of Miriam Santiago, but they got tied at the time of FVR’s mysterious trip abroad; later he resumed his slight lead. What I found remarkable was the rise of Danding Cojuangco, a Marcos crony, from last at the start to third at the end.

In the vice-presidential race, Erap Estrada always had a big lead over Marcelo Fernan.

In 1998, Erap started way ahead, and even widened his lead. Joe de Venecia started out in fourth place, and only managed a distant second at the end.

In the VP race, Gloria Macapagal Arroyo always had a big lead. Serge Osmeña was second place up to March, and was overtaken by Edgardo Angara in April.

In 2004, Fernando Poe Jr. started out ahead, but the citizenship issue allowed GMA to catch up. After his citizenship was settled, FPJ got the lead again, until GMA’s big boost from the revelation that Raul Roco had cancer. It was the tightest presidential race.

In the VP race, Noli de Castro had a very large initial lead over Loren Legarda, and nearly dissipated it.

In 2010, after Cory Aquino died in August 2009, Noynoy Aquino came literally from nowhere to take the lead (away from Manny Villar) by September. Noynoy’s lead narrowed in February, but widened again afterward. It was only in early May that Erap tied for second place; he was never a serious threat.

Much more interesting was the VP race, where Jojo Binay improved steadily from third place, matched second-placer Legarda by mid-April, and then tied the leader Mar Roxas in early May. It was neck and neck at the very end.

For a better view of the future: Study the past first.

* * *

(This is a short version of my talk at Rotary Club Manila on 1/21/2016.) Contact [email protected].