Perception is king

Gov’t cuts ’08 export growth goal to 8% as exporters grumble, Gov’t moves ‘too little, too late’ (the data on Cebu validates the info I received and wrote up in my column, The future’s bright). This may have contributed to the President’s latest beating in the survey (see below). The Manila Times editorial today puts forward unencouraging statistics:

FROM 1997 to 2003, close to four out of 100 middle-income families slipped to the low-income group, according to secretary-general Romulo Virola of the National Statistical Coordinating Board.

While their numbers grew to 3.42 million in 2000 from 3.26 million in 1997, the study said the middle class dwindled to by 140,013 to 3.28 million from 2000 to 2003.

As a percentage of the total population, they accounted for 19.9 percent in 2003, down from 22.7 percent in 2000 and 23 percent in 1997, thus swelling the low-income group.

While Dr. Virola hinted of a continuing trend into 2006, he failed to elaborate. Does this imply that more Filipinos are closer to poverty than originally perceived?

In 1997, middle-class incomes ranged from P148,307 to P1,207,122 a year. Six years alter, annual incomes rose to P203,109 to P1,651,632. Dr. Virola defines middle-income families as those who own their own houses as well as electric appliances like refrigerators and radio sets.

The editorial goes on to say,

Now how to explain the downtrend? Dr. Virola said the lack of policies aimed at nurturing the middle class may be responsible for their diminishing numbers. While the government is focused on lifting “the poorest of the poor,” Dr. Virola said, a strategy that pays attention to the middle class may be more effective in sustaining economic expansion.

We offer another possible explanation: a tax system that has weighed down more on the middle class than on the rich and the poor. While the rich have ways to avoid or reduce their tax liabilities and the poor’s income is hardly taxed, the predominantly salaried middle-class can’t escape the taxman.

This may explain why our “best and the brightest” are leaving to work in foreign shores.

A note on this -I’ve written elsewhere that it seems to me that since 1986, the middle class has essentially decided to leave the country, and that the trend’s accelerating. This is creating a power vacuum of sorts, one not being filled quickly enough by the new middle class, which isn’t built in the image of the old upper class the way our traditional middle class is. A reason for the exodus is that aside from toppling presidents, the middle remains too small to fundamentally matter during elections, and, because of this, it knows it is seriously unrepresented, even victimized, by elected leaders who pander to other, better-organized and more numerous (or better-funded) constituencies. For example, the upper class will invest in elections; the masses swarm out to vote. The middle, on the other hand, pretends that a lack of interest and contempt for politics is some kind of virtue. An attitude they can maintain, of course, as immigrants in other countries.

Overseas, in Korea: School for Scandal, a look at the dynamics of South Korean politics, as the formerly vibrant East Asian success story grapples with problems:

Employment growth has been slowing since the second quarter as a strong won cuts into exports. Korean companies are slowing hiring, mortgage borrowing is falling slightly, which means consumer spending will probably weaken somewhat before a rebound expected towards the end of 2009. Housing, now among the world’s most expensive, is an issue, along with education. There is fear over the future, especially given China’s rapid growth.

The other elephant in the room is the 200 billion won (US$216.7 million) Samsung slush fund, which could well live up to its advance billing as one of the messiest scandals in Korean history, although it is hard to tell from the outside given the Korean media’s seeming conspiracy of virtual silence on the issue.

In early December, Newsweek Magazine predicted that the scandal “is about to not only break up the “Republic of Samsung,” but also “re-contour Korea Inc.” But that seems to be a remarkable bit of hyperbole. The Republic of Samsung is expected to endure and prosper as Korean society pulls together to protect its biggest company from too much pain, despite the pending appointment of a special prosecutor mandated by the National Assembly…

Korea’s judicial, social and governmental systems have generally given the chaebol a pass. Samsung officials have repeatedly denied everything and warned that the investigation could damage the company and the country’s overall economy, a tactic frequently used by huge corporations in Korea to try and deflect legal action by cloaking themselves in the mantle of the nation.

The next chapter? Lee will be investigated. Samsung will be investigated. But not very much is likely to change. “If the Koreans elect Lee president knowing he is a crook, they will have simply accelerated their normal process,” said one cynical observer in Seoul. “This way they can get right to investigating him while he is in office instead of waiting for the end of his term.”

The latest survey says: Satisfaction with Arroyo drops. I’m not keen on “net satisfaction” but the report lists the numbers that should count: the separate satisfied and dissatisfied figures. They’re down, at times significantly so, across the board, nationwide:

The President’s net ratings were close to neutral in the first and second quarters (-4 and -3, respectively) before plunging to -11 in the third quarter.

The question posed to 1,200 adults nationwide was “Please tell me how satisfied or dissatisfied you are in the performance of Gloria Macapagal-Arroyo as President of the Philippines. Are you very satisfied, somewhat satisfied, undecided if satisfied ordissatisfied, somewhat dissatisfied, very dissatisfied?”

The independent survey research institution said loss of support in Mindanao and the Visayas, and worsening dissatisfaction in Metro Manila, resulted in the fourth-quarter drop.

In Mindanao, Mrs. Arroyo’s net ratings plunged to -20 (30% satisfied, 49% dissatisfied) from -1 (38% satisfied/dissatisfied, rounded) just a quarter ago. In the Visayas, it turned negative, at -9 (35% satisfied, 44% dissatisfied), from the neutral +1 (40% satisfied, 38% dissatisfied, rounded) in the third quarter.

In Metro Manila, it was -34 (25% satisfied, 59% dissatisfied), down from September’s -24 (28% satisfied, 52% dissatisfied).

She gained somewhat in the Balance of Luzon, where her net rating was -12 (34% satisfied, 45% dissatisfied) from -18 (32% satisfied, 50% dissatisfied) previously.

Rural net satisfaction, however, fell to -8 from -3, while in urban areas it dropped to -23 from -18.

The margin of error, the SWS said, was plus or minus 3% for the national average, and plus or minus 6% for area results.

By classes, net satisfaction with the President turned negative among the ABC sector, at a net -16 (35% satisfied, 51% dissatisfied), from +8 (46% satisfied, 38% dissatisfied) in the previous quarter.

It remained negative among the class D or masa, at -13 (33% satisfied, 46% dissatisfied) from -12 (34% satisfied, 46% dissatisfied) previously.

Satisfaction went down among the class E, to -25 (26% satisfied, 51% dissatisfied), compared to -13 (33% satisfied, 46% dissatisfied) in September.

See Randy David’s The reality of surveys:

Perception is our only access, as human beings, to the reality of the social and natural worlds. We have no direct knowledge of the reality of the world “as it really is” against which we might compare our perceptions. We can only compare perception with perception, because what we call reality is indeed just another perception.

In a Philippines Free Press article I tried to come up with a rule of thumb:

IN the Philippines, perception is king, and plausibility is queen. If things are perceived to be, then they are, regardless of reality. At the same time, if something is plausible, it is viewed as probable –again, regardless of reality.

From the Lionel Giles translation of the Analects of Confucius:

Tzû Kung asked for a definition of good government. The Master replied: It consists in providing enough food to eat, in keeping enough soldiers to guard the State, and in winning the confidence of the people.–And if one of these three things had to be sacrificed, which should go first?–The Master replied: Sacrifice the soldiers.–And if of the two remaining things one had to be sacrificed, which should it be?–The master said: Let it be the food. From the beginning, men have always had to die. But without the confidence of the people no government can stand at all.

It’s this obsession with public confidence -identifying if it’s there, and maintaining it if it does, and building it, if it doesn’t, that’s at the heart of surveys in between elections. In his column today, John Nery writes of The use of error–

It pained me to see Cabinet Secretary Ricardo Saludo, possibly Malacañang’s straightest thinker, adopt the small-sample defense (even if only by implication). Instead of, say, asking whether the sample of 1,200 voting-age Filipinos Pulse Asia canvassed was truly random or not, he echoed Press Secretary Ignacio Bunye’s innumerate demonization of the use of random samples in the first place.

Survey findings, of course, can be mistaken; that, in part, is what the margin of error is for. But think about it. The power of surveys — a power Malacañang itself put to good use in April and May 2004 and June and July 2005 — rests on both the recognition that errors are inevitable and on the possibility that they can be accounted for.

Here’s something intriguing in the same column:

Sometime in 2003 or 2004, someone who was partly responsible for the operation told me that Malacañang had some 2,000 telephone lines (he said lines, but he might have meant numbers or SIM cards) that it could use to mobilize text-message support on certain issues or in certain situations (such as live polling on TV talk shows). I have never verified this assertion for myself (to be sure, I trusted my confidant). But in the past couple of days I have been receiving text messages from unknown numbers — all from the same polluted source. How do I know? Because the errors and misspellings in the messages, claiming that the “most corrupt” survey findings are the dastardly work of two named individuals, are exactly the same. A word to the wise: Pulse Asia’s boss is Rapa Lopa, not “Rafa.”

In other words, the curiosity, the desire to know, that motivates surveys are at the heart of governance -as is the more contemporary urge to debunk them by means of black ops. Few remember that Arturo B. Rotor, physician, and among the most notable writers of his generation, also served as Executive Secretary in the Commonwealth Government-in-Exile. From that period of service, he developed a shrewd appreciation of leaders, followers, and what power does to both. Writing in The Bulletin of the American Historical Collection, Rotor explained how, even without surveys, leaders have to undertake a constant evaluation of the public mood:

Most important of all, in assessing the remarks and opinions of Quezon, one must always bear in mind the Quezon personality. A gifted writer and orator, possessed of a tremendous memory and the knack for split-second timing of repartee, Quezon could hold an audience spellbound in an auditorium, or keep a social group together until the small hours of the morning with spicy anecdotes and epigrams. He could quote from Rizal or Cervantes as easily as he could crush an opponent with a double-edged witticism in, English, Spanish, or Tagalog. Thus what he said at the moment may be nothing more than, an idle observation, designed to draw a laugh, or it may be the conclusion from months of pondering.

He could chastise a colleague or underling today and appoint him to an important post tomorrow. When Romulo’s book “I Saw the Fall of the Philippines,” came out, “Quezon had been so angry at Romulo that he had told him to get the hell out of here and never come, back,” and had deprived him of his uniform as a Lieutenant Colonel of the Philippine Army when he was on the lecture platform. Yet, later Romulo was to be promoted and given important assignments.

One reason for this of course was that Quezon had his own way of gauging public opinion, of taking a poll survey. He would say something preposterous or do the completely unexpected to find out what the people thought of a political leader, or to measure their opposition to religious instruction in schools. If the act aroused a bigger rumpus than he had calculated, he would institute an appropriate measure.

Thus to the uninformed, Quezon often appeared inconsistent, mercurial, unreliable, a man whose word could not be trusted. No greater mistake can be made. When Quezon had studied a problem and made up his mind, no earthly force could stop him. Had it not been for this trait, probably the Philippines would still be a Commonwealth like Puerto Rico, or a State like Hawaii. His vision was often prophetic; the conflicts that have taken place in American Army bases at Angeles and Subic were clearly foreseen by him in his opposition to the Hare-Hawes-Cutting bill. He had always warned his country against American imperialists deeply entrenched in State and Interior; if he had been alive, the “parity rights” of Americans would have never been forced on the country.

Here’s a page from the article, Public Opinion Polls and the 1961 Philippine Election by Martin Meadows, which gives you the background history of surveys in our country.

Now the article itself is mainly concerned with putting the 1961 elections in the context of surveys. A particularly interesting incident was that the firm that had the biggest number of respondents -no less than over five million- was the survey that got things the most wrong:

Writing of the same (1961) elections, in his post mortem titled The Winners, ’61, Nick Joaquin wrote that from the incumbent on down, people were surprised Macapagal defeated Garcia, and notes even the surveys had the two neck-and-neck:

President Garcia can hardly be blamed for not conceding defeat at once; he, too, just couldn’t believe that Macapagal was winning and, but not conceding, was merely expressing a general astonishment and incredulity. It seems now that everyone who voted for Macapagal did so with no great hope that he would win. Each pro-Macapagal voter must have felt solitary, one in a hundred. So many people who had expressed disgust of the Garcia regime had followed denunciation with despair: “But how can one vote for Macapagal?”

This is in sharp contrast to the atmosphere in 1953, when everyone who voted for Magsaysay felt quite sure that everybody else was doing the same.

The doubts about a Macapagal triumph were indicated by all the pre-election forecasts, even those that had him leading. The pollsters in general detected a trend in his favor but apparently questioned the strength of the trend. Those who gave him the lead carefully stressed that the lead was very small. In fact, the last poll survey to be made public just before the elections, the U.P poll, flatly declared that Garcia and Macapagal were running even, any edge in favor of the latter being so slight as to be “insignificant.”

When the returns started coming in, the public literally couldn’t believe its eyes.

Nick Joaquin, writing his post mortem of the 1963 elections in Ayos na ang buto-buto, delves into surveys that went wrong. And incidentally, it all sounds very, very familiar:

Robot’s revelations shook the local political earth. The Liberals would win the senatorial race by 5-3, or more likely by 6-2, with either Padilla or Roxas as topnotcher, followed by Tolentino, Diokno, Ziga, Climaco and Liwag. The eighth place would be contested by De la Rosa, Balao, Puyat and Cuenco, with the first two having “a slight edge over the others.”

As it turned out, the topnotcher berth was contested by Roxas and Tolentino, not Padilla and Roxas; Puyat, whom Robot placed almost outside the magic eight, landed in fifth place; De la Rosa, Balao and Cuenco ended way, way below eighth place; and the unmentioned Ganzon and Lim fought it out with Climaco for the tail end of the line.

The Robot findings, released to the press a week before election day, were published three days before the elections, and one day before the U.P. statistical center released its own poll survey, which also had the LPs leading, 6-2, with Padilla and Roxas in the first two places, followed by Tolentino, Diokno, Climaco, Ziga and Liwag, and the eighth place being contested by Balao and Puyat. As Robot, aggrieved, would later point out, the U.P. poll escaped the ire of the politicians, but Robot got it from both sides.

De la Rosa and Cuenco angrily questioned the accuracy of the poll. The NPs were, of course, even angrier. They denounced the poll as “part of the Liberal scheme to cheat” in the elections, “a smoke-screen to prepare the people’s minds to accept rigged election returns.” The Robot poll results had been “doctored” to produce a “bandwagon mentality” among voters, and their “premature publication” was an LP propaganda gimmick. The NPs insisted that they would either sweep the polls or get a clear majority.

The day after the elections, people were quipping that there was one sure loser: Robot. Its forecast had flopped.

Says Vice-President Francisco Lopez of Robot Statistics: “What we published was an estimate of the situation as of a given period of time: from late October to early November. It was not a forecast, it was not a prediction. If we had wanted to make a real forecast, we would have continued polling up to the eve of the elections.”

The trouble with this disclaimer is that Robot was using that very word, forecast, during the days it was frantically trying to decide whether or not to publish its poll findings ahead of the balloting, or wait, as it did in 1959, until the last ballot had been cast. One-upmanship finally prompted the “premature publication.” Robot feared to be one-upped by another poll organization, and decided to release its findings to the press a week before election day.

Joaquin continues by looking into Index, the rival of Robot:

The other poll organization was Index, which had, in late October, begun publishing a series of reports on voter attitudes based on a survey. Robot felt sure that the series would be climaxed by a forecast of election results. The fear was unfounded; but Robot not only didn’t want to be beaten to a forecast but was afraid the poll figures it had been gathering month after month since the campaign started might be stolen and used.

On November 2, Robot invited three distinguished citizens–Father Francisco Araneta, Professor Ariston Estrada and Judge Pastor Endencia–to read its latest survey on poll trends. Copies of the survey were read and signed by the three men, and then locked up in a vault, as proof that Robot already had those figures at that time. One-upmanship is a nervous way of life in every branch of Madison Avenue.

This was on a Saturday. The following Monday, November 4, Robot, apparently still jittery about being beaten to the draw, assembled representatives of the four leading Manila newspapers and provided them with copies of the latest Robot poll results.

Explains Robot’s Armando Baltazar: “That was for their guidance only. We wanted them to know the real score. Their columnists were making predictions and might go off on a wild tangent. The publishers could keep their columnists from going out on a limb if they knew what the figures were. But we made it clear that we did not want any publication.”

Robot’s George Cohen modifies this: the poll figures were released to the press; it was up to the press to decide whether to publish them or not, and when. On November 8, Cohen dispatched a letter to the publishers:

“You will recall last Monday that Robot wished to impose an embargo on the release of its election estimates until the closing of the polls on election day when survey results could not possibly be accused of influencing events. Robot in fact does not believe that at this stage of the election campaign a release of its survey results now would significantly affect its outcome–if at all. However, Robot does not wish to be the first polling group to be releasing pre-election forecasts–but as a public opinion/marketing research organization it feels obliged researchers do. Thus please feel free to publish the results enclosed within red quotation marks if other polling organizations or research groups (exclude informal newspaper or magazine surveys) such as the University of the Philippines, Index, et al. have or are in the act of publishing national senatorial election forecasts. If not, Robot respectfully requests that you withhold publication until the polls have closed on election day.

“Finally we wish to remind that some 15% of the voters still do not know whom they will select for their senatorial choices on November 12. This figure constitutes a 4% increase over the ‘don’t know’ers’ since September, thus indicating considerable uncertainty on the part of the voters. Thus last minute shifts of preferences are possible even on election day–which could upset the above forecast. What the forecast represents is the best estimate of the state of the public opinion at a given point of time, 26 October to 6 November.”

Through the tangle of language, the publishers presumably saw permission to publish, since the U.P. was “in the act of publishing” its own forecast. But why did Robot’s “best estimate of the state of public opinion” fail to tally with actual public opinion as expressed in the elections?

Cohen and his colleagues say that they went by trends. When they began polling in July, Puyat, for instance, was in fourth place but kept slipping, slipping, until he was in seventh or eighth place. The Puyat trend was, therefore, downward: “But he didn’t slip as much as we expected him to. He caught it in time, arrested his decline.” Robot failed to catch that stoppage and went by the general Puyat trend–which is why the forecast had him still slipping off the tail end.

Joaquin then returns to describing what the surveys got right and wrong and why:

Another candidate whose trend was a downward slide. De la Rosa, was popularly believed to be a sure winner. Robot was a bit more accurate here, and surprised everybody by having De la Rosa just hovering over the edge of the eighth place: “If we had surveyed more, up to a few days before the elections, we might have caught him on his way out.” The U.P. poll did find De la Rosa already out.

The three fastest risers, according to Robot, were Roxas, Diokno and Liwag. Diokno started at 13th or 14th place, rose steadily, suddenly shot straight up during the last phase of the campaign. If graphed, his progress would be a long slanting line that ends in a steep curve. Liwag started at 16th, worked his way up to 7th in a more even manner. Most spectacular of all was Roxas, who started below the eighth place and rocketed to the top. Robot’s data indicate how effective propaganda can be when skillfully used, for Roxas, Diokno and Liwag had the smartest publicity machines in this campaign.

The candidates that really got Robot into trouble were Climaco and Ganzon. Robot estimated that Climaco would outpoll Ganzon in Mindanao, 2-1. The elections proved they had about even strength there–which, says Cohen, is inexplicable, since Climaco, after all, is from Mindanao. Cohen hazards the guess that Climaco’s drive against smuggling while in Customs turned the Moro vote against him.

And the pollster’s explanations why this was so:

To people who say that Robot took a beating in these elections, Cohen points out that his organization had a near-perfect score in the gubernatorial races, pinpointing the winners in 21 out of the 22 provinces it polled. (Robot, like everybody else, guessed wrong in Bulacan.) Cohen also claims that Robot scored almost 100% in its forecast of election results in the Manila area; it missed only one winner: the vice-mayor of Quezon City. But Robot saw the Manila vote as 4-4 in the senatorial election (the actual ratio was 6-2 in favor of the NPs) and 4-1 in the mayoralty contest (Villegas actually had only about a 2-1 lead over Oca). Cohen has two explanations for the increased figures in favor of the NPs: their miting de avance on Plaza Miranda was a major event of the campaign, giving the NP senatorial candidates, and Oca along with them, the benefit of maximum public exposure, and exerting a terrific influence on the undecided vote. Cohen’s other explanation is that Manila has a large floating vote: the squatters but still vote in the city. Because it polled only actual residents, Robot failed to get a picture of the total Manila vote.

And he then asked something repeatedly discussed since:

Just how much do these forecasts affect voters’ decisions? In the U.S. not at all–or so they say. In the Philippines, such forecasts, Cohen admits, may sway votes, but only if published, say, ten days or two weeks before the elections. But a forecast published practically on the eve of the polls can have little effect on them. Cohen cites an instance. In 1961, just two days before the elections, Mayor Lacson, against Robot’s wishes, published the Robot poll survey that showed Garcia was losing. The forecast, according to Cohen, did not appreciably alter voting trends. But it did have one unexpected result that has passed into political legend. The story goes that money given to the leaders to distribute on election day was not handed out because, the leaders told themselves, Garcia was going to lose anyway. Failure to flood the polls with handouts may have helped Garcia lose.

And how the candidates and leaders, then as now, have a love-hate relationship with surveys:

The NPs, who are usually so zealous for freedom of expression, are currently up in arms against public opinion polls. Senator Primicias threatened to sue Robot for multimillion-peso damages and to have it investigated as a foreign agency interfering with Philippine elections. Robot says its capital is 90% Filipino, that the company is run by Filipinos, and that it is in no way subsidized by World Gallup Polls. One NP who doesn’t believe the Robot forecast was “rigged to please its client” (Robot says it had clients from both parties in this campaign) is Diokno. Robot tried to assess the situation as best it could, but, says Diokno, it failed to take into account an important “x-factor”: people’s fear of the administration. As Robot was not really undecided, was already for the NP, but preferred to keep mum and express itself only at the polling booths, for fear of reprisals.

The Robot forecast appeared the Saturday before election day. The NPs had their miting de avance on Plaza Miranda that Monday night; and Robot, the second favorite target, suffered the slings and arrows for outrageous fortune-telling. The crowd the NPs drew that night was unquestionably the hugest to assemble on Plaza Miranda since the time of Magsaysay.

The main point, though, is simple: even before surveys existed, politicians had to devise ways to probe public opinion. Surveys, once they entered the scene, enable leaders and followers to more accurately, because scientifically, gauge public opinion. It’s a doomed leader, a decaying leadership, and a deluded public, that wonders whether surveys have a role to play in governance. They always have -even before they were invented!