Venezuelan opposition's bogus poll plot

Manuel Rosales faces almost certain defeat in Venezuela's elections but he has told his supporters: "It is true, we are winning."

With only days to go before Venezuela's presidential elections, President Hugo Chavez has a massive lead over his US-backed rival, Manuel Rosales. The six most recent polls conducted by recognised firms put the gap between the two candidates as follows: Zogby — University of Miami: 29%, Associated Press — IPSOS: 32%, Datanalysis: 27%, Datos: 27%, Consultores 21: 17%, Evans McDonough: 22%.

Faced with almost certain defeat, one might have expected Rosales to rally his supporters with typical politician's bluster about how victory was still possible. Instead, he issued a very curious statement: "Every single important pollster has reported we have an apparent lead over the other candidate. It is true, we are winning."

One such "important pollster" is Alfredo Keller of Keller and Associates. On November 2, Keller released a poll that purported to show that the candidates were neck and neck. Later, it emerged that the interviewees had never actually been asked how they would vote. Keller simply took a guess based on their answers to other questions. Keller is an opposition supporter.

Another "important pollster" is Survey Fast Venezuela. They published a poll on November 21 that predicted a statistical dead heat. I had never heard of this so-called pollster, so I decided to type their name into the Google search engine. They attracted a mere 35 references (all of which related to their November 21 poll) and did not appear to have either a website or an identifiable track record or methodology. In other words, Survey Fast is a bogus firm and a front for the Rosales campaign. By contrast, when I googled the unimportant pollsters, i.e. the major firms that all show Chavez way out in front, I was rewarded with tens or hundreds of thousands of references.

And then there are the really, really important pollsters. Pollsters like Victor Manuel Garcia, director of Ceca, who released a poll showing that Rosales was a full 10 points ahead. Garcia, like his presidential hopeful Rosales, was a key participant in the failed 2002 coup that briefly overthrew Chavez and abolished parliament.

So why do a handful of dodgy polls matter?

To answer that question, we must go back to August 15, 2004. That was the day of the referendum to decide whether Chávez should be thrown out of office midway through his presidential term. Whilst voting was still taking place, a New York-based firm called Penn, Schoen & Berland produced an exit poll which claimed that Chavez had lost by 18 points. The opposition went wild and declared victory. When the actual result was announced — Chavez had won by 18 points in an internationally certified free and fair election — the opposition claimed fraud, citing as evidence this exit poll.

Penn, Schoen & Berland say that their exit polls have a margin of error of under +/- 1%. So how could they have been off by a margin of 36 percentage points? Simple. They had subcontracted the conduct of the poll to a US government funded, anti-Chavez group called Sumate, whose leader participated in the 2002 coup. The poll was entirely bogus, but it served the purpose of casting a shadow over the democratic credentials of the Chavez presidency and the left in general.

Past performance is the best guide to future performance, so I'll end this piece with a prediction. Chavez will win Sunday's election by a landslide and the result will be declared free and fair by election observers from the EU, the OAS and the Carter Centre. The following day, opposition extremists will pour onto the streets screaming fraud in a vain attempt to stage a Ukrainian-style "orange revolution" and unseat the man they couldn't remove through the ballot box.

And how will all this be justified?

"We were robbed by the computerised voting machines", they will tell us.

And the proof?

"Every single important pollster had us ahead."

[Calvin Tucker is co-editor of . Originally printed at .]