In the last few weeks, the Government has been promoting and distributing a bunch of polls that pretend to show that Hugo Chavez enjoys a huge lead over opposition candidate Henrique Capriles. The polls can be categorized in two groups: One, flight-by-night operations which are clearly funded by the Government to promote favorable numbers, such as GIS XXI, Consultores 30.11 and International Consulting Services. These three have been showing a 30% lead for Hugo Chavez, with the President holding between 55% and 58.7% of the vote, versus 22%-25.7% for Henrique Capriles. A second group is better “known” pollsters such as IVAD and Hinterlaces, the first one showing a 30% lead by the Venezuelan President and the second one an 18% lead.
These polls have thrown a lot of confusion among people, who precisely because this has happened in the past, are skeptical of polls, but nevertheless worry about them.
The flight by night operations clearly have little credibility. Jesse Chacon, the former military officer/telecom expert/security and police expert has now been turned into a poll expert at GIS XXI. But his track record is quite dismal, predicting, for example, an overwhelming victory by Chavismo in the National Assembly vote. The other two pollsters in this category only surfaced recently, have no track record and not even a webpage. Given that polling companies tend to make their money doing consumer, not political polls, this alone makes them suspect.
The other two are pollsters with longer “track records”, but not necessarily distinguished ones. IVAD for example, has as its main claim to fame predicting a Chavez victory in 1998 and 2000, but the details of its polls and its techniques are seldom published and somehow, the data is always “leaked” by pro-Government press and its owner Seijas, evades public presentations and speaking.
The other one, Hinterlaces, has a spotty and inconsistent track record, with its latest “miss” saying a week before the opposition primary that the race had become a two man (well, one man, one woman) race between Maria Corina Machado and Henrique Capriles. We all know how that turned out and you would think Hinterlaces would go back and review its methodology before speaking in public after that, but no sooner had Capriles been declared the winner, when Hinterlaces was talking up its poll on the Presidential race.
There are other pollsters like Datanalisis, Datos, Keller and Consultores 21, with a variety of reputations. I tend to follow Datanalisis and Consultores 21, simply because they have the better and more consistent track record. However, of this latter group, only Consultores 21 has a post-primary poll.
This poll says the following: Of the 2000 or so people polled, Chavez has a 46% to 45% lead over Capriles. However, if only those polled that say they intent to vote are included, Chavez leads becomes 51% o 46%.
Who should one believe?
Well, let’s look first at whether it is logical that Chavez has such a huge lead in the mid fifties, with the opposition only garnering 25% or so:
If there are approximately 18.5 million voters and abstention is 25%, then the total numbers of votes will be around 13.88 million. If the opposition gets 25%, it would get 3.47 million votes, barely above what the opposition got in the primary. More importantly, Rosales got 4.2 million votes in the 2006 Presidential election and the opposition got 5.7 million votes in the National Assembly election. Thus, this 25% seems non-sensical and suspect.
We could, as an alternative, look at the 2006 election, but almost everyone got that one right.
I prefer to look at a tougher one: The 2007 referendum. IVAD was saying then (Trust me, I can’t find a link) that the referendum would pass by 64% to 25% or 26%.
In contrast, in the last Consultores 21 poll right before the referendum, they published this chart:
On the left was the percentage of votes in favor (Red) and against the referendum (blue) if only those that say they will vote are counted. The result was a small edge for rejection of the referendum. Recall we never knew the final result, only a partial one, it was No 50.7% Si 49.3% on question A and No 51% Si 48.9% on question B. This is quite a good prediction. The right hand side was simply a hypothesis if voters turned out in larger numbers.
Thus, who do you trust more? Clearly Consultores 21, even if you may not like the result. But there is plenty of time, somehow the Government is promoting Capriles a lot by attacking him, Chavez has a long road with his health issue and there are seven months of campaigning to go.
A five point margin seems doable, even if the opposition has to win by a few percentage points.
The Devil's Excrement 
