Chunky

Every election year, the surveys done on voter preference somehow provokes polarized argument.

Those who feel disadvantaged by it call the exercise a fraud and demand that the thing be banned. Those who think there are favored by the surveys worship it as gospel truth and attempt to build a bandwagon effect from the most preliminary results.

The same unenlightened arguments are repeated all the time — although the parties in the argument somehow change sides, depending on whether one particular survey makes them look good and another is adverse.

The consistent loser in this polarized debate is the survey method itself. Although it is a highly developed statistical device that, when done properly, has great predictive value, polling is often vaguely understood by the larger public. Indeed, as the critics of opinion polling often ask, how could a sample of 1,200 respondents speak for 35 million voters?

Indeed. Unless we fully grasp the laws of statistical probability, that might seem like a fatal question.

Those who want to undermine the credibility of the surveys done on the senatorial race point out the case of John Osmeña. In the last election, running with the pro-administration coalition, he ranked near the top in the preliminary surveys. In the end, he lost the election, narrowly missing the win column.

Those who sell polling services, on the other hand, point out that most of the time the trend indicated by the surveys are validated by the outcomes. The accuracy is greater in one-on-one contests. Exit polls are generally more predictive because it take a sample of respondents after they had voted.

The recent surveys show candidates of the Genuine Opposition outnumbering candidates of Team Unity in the win column for the 12 Senate seats at stake in this elections. Spokesmen of the opposition understandably trumpet those results, predicting an opposition landslide on election day. These are the same guys who, the past three years, totally ignored the fact that all the surveys showed Gloria Macapagal Arroyo leading Fernando Poe Jr. in the week prior to election day.

Their view of surveys is completely colored by their partisanship.

It is true that the surveys on senatorial candidates are burdened by many factors. But they are generally good measures of name recall as well as voter inclination towards specific candidates.

One major vagary in the case of the senatorial contest is the fill-up rate — the number of names voters actually write in their long ballots. We have, as we all know, one of the most antiquated voting systems in the world, requiring voters to actually write down the names of nearly 40 candidates in a long ballot.

To minimize the effect of the variable fill-up rates, our major polling organizations try to simulate actual voting by asking respondents to write down the names of their preferences in a mock ballot. They have figured out that the average fill-up rate is 7 names out of the possible 12.

Pollsters love to compare the sample size of surveys to taking a teaspoonful of broth out of a large bin to see how the brew tastes like. That is a good metaphor — except that in the case of Philippine-style senatorial contests, we are speaking not of clear soup but something more like chunky stew.

Candidates will have variable voting strengths across regions because of the favorite son phenomenon. For instance, Bicolano voters are notorious for voting for Bicolano senatorial candidates exclusively and across party lines. It is a phenomenon that benefited Kit Tatad, Raul Roco, Joker Arroyo and Greg Honasan in previous elections. The Osmenas have traditionally benefited from strong Cebuano voter support.

Distinct regional preferences are the "chunks" in this stew. They are not very accurately captured by the survey method available.

Too, because of the laborious process of remembering the names of the candidates they should write down on the ballot, the senatorial contest is most amenable to sample ballot operations on election day. Sample ballots are distributed by the campaigners of local candidates. They usually contain some names of senatorial candidates endorsed by local contenders.

Therefore, the coalition with the greater number of local candidates (referred to as "candidate-presence" in the jargon of campaign tacticians) have a distinct advantage of getting their senatorial candidates on the sample ballots of local political lords. It is in the sample ballot operations where the phenomenon of "junking" happens: that is, when local brokers drop the names of some candidates in a party ticket from their sample ballots.

The ability to deliver the sample ballots on election day (which is a massive operation) has significant effect on the outcomes of the senatorial contest. This is what "campaign machinery" is all about. True believers in the efficacy of sample ballot operations say that this could sway as much as 10% of the vote.

The most important sample ballots are those distributed by religious sects that enjoy "command votes" as well as local political leaders who enjoy great prestige in their localities.

It is on this aspect that the superior candidate-presence of the pro-administration coalition will be most telling. In many localities, the contest will be between parties belonging to the same coalition — most likely Lakas against Kampi candidates. The opposition coalition has meager local candidate presence nationwide.

Voter affiliation with senatorial candidates is weaker than with presidential candidates. Here the superior campaign machinery plays a vastly more significant role in the actual voting that happens.