Beating Duterte: What the polls indicate

With the May 9, 2016 elections less than a week away, can Mayor Rodrigo Duterte’s rivals catch up and steal the presidency from him? The answer is No.

The reason is simple arithmetic. In the 1998, 2004 and 2010 presidential elections, the major pollsters always got the winner right though not the winning percentages (which invariably were higher than the ratios monitored by pre-election surveys).

Digong Duterte’s possible margin of victory is so huge (between four and over five million votes) the only way to defeat him is to cheat him massively. Cheating, however, would trigger violence, on the scale of a mass unrest, erupting not just in Metro Manila and Davao but perhaps in every major city nationwide.

In the Pulse Asia poll of April 19-24, 2016 paid for by ABS-CBN, Duterte garnered 33 percent, 11 percentage points ahead of his nearest rival, Senator Grace Poe, with 22 percent. Assuming a voter turnout of 43.49 million, 11 percent is equivalent to 4.78 million votes. The Pulse Asia survey has a margin of error of +/-1.5 percentage points (between 650,000 and 1.3 million votes), with 4,000 respondents.

In the Social Weather Stations survey of April 18-20, 2016, Duterte is ahead of Poe by nine percentage points, 33 percent vs. 24 percent. Nine points is equivalent to 3.9 million votes. The SWS poll has a margin of error of +/-2 percentage points (equivalent to 870,000 to 1.73 million votes), with 1,800 respondents.

In polling, the more respondents, the better or more accurate the results. With its smaller error margin, Pulse Asia, it seems, is the better and more reliable pollster.

In the 1998 presidential election, SWS missed the percentage winning ratio of the president-elect but got the rankings right. Joseph Estrada won the presidency with 39.86 percent (10.722 million votes). He defeated Speaker Jose de Venecia who had 15.87 percent or 4.268 million votes.

In its April 16, 1998 survey, SWS predicted a 30 percent share for Erap and 12

percent for JdV. It missed Erap’s actual share of the vote by 9.86 percentage points (39.86 minus 30). Thankfully, Erap’s margin of victory over second placer Jose de Venecia was so large (6.45 million votes), the error did not matter.

In its May  1 to 4, 2004 preelection survey, SWS predicted a win for Gloria Macapagal Arroyo’s 37 percent to challenger Fernando Poe Jr.’s 30 percent, and Panfilo Lacson’s 11 percent. Pulse Asia, in its April 26-29, 2004 survey, showed Arroyo winning with 37 percent—by six points over FPJ’s 31 percent and by 26 points over Lacson’s 11 percent.

The actual results in 2004: Arroyo 12.905 million (39.99 percent); Poe 11.782 million (36.51 percent); and Panfilo Lacson 3.51 million (10.88 percent). Both SWS and Pulse Asia got the 1-2-3 rankings right but got the ratios or percentages grossly wrong.

Both SWS and Pulse Asia missed GMA’s winning margin by 2.99 percentage points or 964,899 votes. SWS missed FPJ’s actual ratio of 36.51 percent by 6.51 points or 2.1 million votes, and Pulse Asia missed it by 5.51 points or 1.778 million votes.

In 2010, both SWS and Pulse Asia pre-election surveys got the winner right but again, their percentages were wrong. The SWS survey of April 28, 2010 predicted 38 percent would vote for Benigno S. Aquino III, 26 percent for Manny Villar, and only 17 percent for Joseph Estrada. The Pulse Asia poll of April 23-25, 2010 had it 39 percent for BS Aquino III, and 20 percent each for Estrada and Villar (a tie). In its preelection survey on April 28, 2010, SWS got the second placer wrong (Villar, instead of Estrada).

The actual numbers: Aquino 15.2 million votes (42.08 percent); Estrada 9.487 million (26.25 percent), and Villar 5.573 million (15.42 percent).

SWS missed Aquino’s ratio by four percentage points (38 vs 42.08), while Pulse Asia missed it by 3.08 points (39 vs 42.08). The error was equivalent to 1.445 million votes for SWS and 1.11 million votes for Pulse Asia.

SWS missed the ratio for the winner by 9.8 points in 1998, by 2.99 points in 2004, and by 4.08 points in 2010—an average error of 5.62 points. Pulse Asia missed the ratio for the winner in 2004 by 2.99 points and in 2010 by 3.08 points—an average error 3.03 points. This implies that Pulse Asia is the more reliable pollster.

In three elections, SWS got the winner right but was wrong on the winning ratio by an average of 5.6 percentage points.

In two elections, Pulse Asia predicted the winner but missed the winning ratio by an average of 3.03 percentage points. In 2010, both pollsters failed to predict the sharp plunge in numbers of Villar who ended up a poor third, instead of a strong second as they predicted.

Fast forward to 2016. SWS predicts a 33-percent vote share for winner Duterte, nine points higher than Poe’s 24 percent. If SWS is wrong by 5.6 points (its average mistake in three presidential elections), Duterte’s winning margin is cut to 3.4 percent (from 9 percentage points). The 3.4 would be equivalent to 1.47 million votes. The mayor’s vote will only be 27.4 percent or 11.91 million votes, which is still higher than Poe’s expected 24 percent or 10.43 million votes.

But considering increasing sophistication in the science and magic of polling, will SWS be really that dumb to miss Duterte’s winning margin by a hefty 5.6 percentage points? If it does, it should get out of the survey business.

In its April 19-24, 2016 survey, Pulse Asia predicts a Duterte win of also 33 percent. If it is wrong by 3.03 points (its average error in two elections), Duterte gets 29.97 percent (33 less 3.03). Digong still upstages Poe’s 22 percent by about eight percentage points (or by 3.47 million votes). Note that Poe would capture only 20 votes or 20 percent of every 100 votes that Duterte loses so that Poe would gain only an insignificant 0.60 percent (20 percent of 3.03).

With SWS surveys, Poe has a better chance of winning than with Pulse Asia’s.

But then, Pulse Asia has proved the more reliable and credible pollster in the past.

Pulse Asia figures indicate a solid victory by Duterte who would take 33 to 35 percent (if not higher) of the May 9 votes. The mayor will win by five million votes (perhaps more) over the second placer.

[email protected]