How SWS and Pulse Asia have been fooling us, and undermining democracy

First of two parts

Pollsters Social Weather Stations and Pulse Asia have been fooling us with their pseudo-scientific opinion surveys.

Their polls distort the real state of public opinion on urgent issues and people’s current preferences for candidates in an electoral contest. This is because they conceal the fact that there is a huge percentage of respondents who do not really have opinions on topics they are asked about, or in the case of their presidential voter preference polls, have not decided yet on which candidate to vote for many months before Election Day.

A quintessential banner-headline, horse-race treatment by a newspaper on false polls. (Candidates’ names and photos blocked to foil the real aim of such polls.)
A quintessential banner-headline, horse-race treatment by a newspaper on false polls. (Candidates’ names and photos blocked to foil the real aim of such polls.)

Consider how SWS and PulseAsia have fooled us with their voter-preference polls in the 2010 presidential elections and for the coming 2016 balloting:
horse-race220151019

Focus on the “undecided” figure: the three different surveys report that only 2 to 3 percent of Filipinos about six months before Election Day had not decided who to vote for.

That’s really preposterous: 97 to 98 out of 100 Filipinos had already decided who they will vote for, many months before the election? This even goes against your common sense and experience: Ask your staff, boss, or friends if they have made their choices for the top posts, and I’ll bet you dinner four out of five will tell you they haven’t decided. Have you, dear reader, decided and won’t change your mind?

One pollster kindly responded to my queries on this flaw of opinion polling when he disclosed that his firm, per request of their clients and on a confidential basis, had asked the question whether the respondents would still change their minds on their voting preferences. The result was that 35 to 40 percent said they might still change their preferences.

That question is really a proxy for the more proper question respondents should have answered first: Have you already decided on which candidates to vote for on Election Day?

This basic flaw of voter-preference polls have long been known in the US and elsewhere, so that in many countries such polls are totally banned (as in Singapore) or banned for a certain period of time before elections (France and Italy). Few broadsheets in the West, in fact, report such polls on their front pages, much less – as is the prevalent practice here – make them the banner headline for the day.

The founder of modern opinion polling, George Gallup, himself pointed out this flaw in 1947, early on in the development of that business. He proposed that this flaw be corrected by requiring as a first step (in what he dramatically called a “quintamensional plan”) any poll, for the respondent to be asked if he has heard of the issue, or has an opinion on it, before he is asked to respond to a question on it. If the respondent says no, he is automatically excluded from the pollsters’ sample.

Gallup Poll itself undertook in 2007 a variation of this procedure when, in conjunction with the usual poll, it had a special one in which it asked first if the respondent had decided or was still unsure of which candidate to vote for as US President. The results of the special poll changed drastically the portrayal in the usual poll that Giuliani and Romney were way ahead:

From: Moore, David, “The Opinion Makers: An Insider Exposes the Truth behind the Polls,” Boston: Beacon Press, 2012.
From: Moore, David, “The Opinion Makers: An Insider Exposes the Truth behind the Polls,” Boston: Beacon Press, 2012.

This is what has been called by political scientists as the problem of “nonattitudes” in all opinion polling. Even an American textbook on polling (Asher, Herbert, “Polling and the Public: What Every Citizen Should Know,” 2012) pointed out: “The presence of nonattitudes is one of the simplest, yet most perplexing problems in public opinion polling … People respond to questions about which they have no genuine attitudes or opinions.”

“No distinction is made between people with real views on an issue and those whose responses are simply artifacts of being asked a question in the first place,” the book explained. “Few people in such circumstances want to admit they are uninformed, particularly on a popular or timely issue.”

Neither SWS nor PulseAsia asks that crucial question in voter-preference polls – whether the respondent has decided or not – which in effect forces them to choose, even if they haven’t really yet made a choice.

Both SWS and PulseAsia’s voter-preference questionnaires ask: “Of the people on this list, whom would you vote for as President (or Vice President) if the 2016 elections were held today.” (Emphasis mine).

But the elections are not being held on that day when they may have not yet made up their mind. Crucially, the respondents are not given this choice: “You may answer that you haven’t decided yet.” Respondents would, therefore, give off-the-top-of-their-head names, which might not necessarily be their considered choices and could change their minds about at the flip of a coin.

In such a forced-choice survey, and especially since the question is one of the 150 or so questions SWS and PulseAsia typically ask a respondent, the respondent is likely to choose a candidate he had recently seen on TV or heard over the radio.

That is why the newspaper Manila Standard recently reported that Grace Poe-Llamanzares and her running mate, Francis Escudero, were “miles ahead” in a poll it commissioned. The survey was done Sept. 21 to Oct. 1: Llamanzares and Escudero have been hogging the headlines since Sept. 16 when they launched their candidacies showbiz style. Similarly, Manuel Roxas 2nd spent P258 million on a TV ad blitz from Aug. 15 to Sept. 15, which pushed up his rating in the poll undertaken by PulseAsia from Sept. 8 to 14.

SWS and PulseAsia have been fooling us by portraying in their reports that the percentage of people who have no opinion on a topic, or who haven’t decided yet (in voter preference polls), is in their “undecided” (in SWS reports) or “Don’t know/Refused/None” (for PulseAsia) items.

This is a lie. Responses placed in that category are those who are adamant in cooperating in any way with a pollster. It also simply represents human errors either in data gathering or in the tabulation of results, typical in any scientific inquiry gathering data and collating them.

This is the reason why in all their polls, whether they be about hunger, satisfaction with government, love, or voting preference, SWS and PulseAsia always report an insignificant percentage of respondents – typically from 2 to 4 percent — as being “undecided” or have “no response.” This is the percentage in the “undecided” or “don’t know” response in any kind of poll here or elsewhere.

Other than the 35 to 40 percent mentioned above as reported by a pollster as those who would change their minds, do we have an idea how many really haven’t decided in such voter-preference polls?

For the vote for Parliament scheduled next week in the United Kingdom, a poll found that 34 percent of voters haven’t yet decided. In Canada, which is holding its federal election on October 19, 50 percent of voters haven’t yet decided, according to a reputable poll. Despite the prime-time TV Republican and Democratic debates, only 27 percent of Americans in a poll say they are “paying attention” to the elections in November.

If we assume that 35 percent of Filipinos haven’t decided on whom to vote for, and assume that the percentages reported in the three polls above represent only those who have decided, the landscape of the 2010 contest and the 2016 poll should be:
horse-race420151019

 

 

The landscapes if one uses a more accurate 35 percent undecided, are so different from the table at the start of this column, which used an insignificant and inaccurate 2 percent undecided.
horse-race520151019

 

More importantly, the fact that 35 percent are undecided could more than eat up the survey’s 3 percentage points plus/minus margin of error. This makes the poll actually nearly useless, which would mean big trouble for someone like SWS President Mahar Mangahas, whose entire professional income has been based on the idea that polls really reflect public opinion.

SWS in 2010 undermined democracy by portraying that a huge number of Filipinos — 42 percent — at that early date (December 2010 ) already had chosen Benigno Aquino 3rd, when the reality was, if those undecided were properly reported, it would turn out only 27 percent did, which was within Villar’s striking distance.
horse-race620151019

 

Moore (2012), who was a former vice president of Gallup Poll, lamented in his book that bogus pre-election polls, which do not accurately reflect the extent of the undecided, falsely create the front-runners. These consequently attract more volunteers to their campaign, stimulate contributions to their kitty, and increase their media coverage – which, in turn, raise their name-recall.

The effects of such false polls are magnified in the Philippine setting, as campaign finances are mostly donations from big businessmen, who allocate their contributions depending on candidates’ ranks in the polls. They were fooled by Aquino’s fake lead in December, which SWS and PulseAsia maintained at the 40 percent levels in the succeeding months. Being opportunists the businessmen were, they threw their money behind Aquino, making the initial bogus poll findings of 40 percent preference a self-fulfilling prophecy.

A tool for democracy, to crystallize public opinion? Hardly. And I understand now why the late Marcos, through the Development Academy of the Philippines, had Mangahas and Felipe Miranda (who founded PulseAsia) develop the method starting in 1982.

tiglao.manilatimes@gmail.com
www.rigobertotiglao.com (for archives)