More election predictions

11.16.2005

Technorati tag:

It's been a few days since the most recent Bolivian pre-electoral poll came out in La Prensa (sorry, no graphic). Already MAAB & Boz have some great analysis. So I figured I'd toss in some thoughts as well. Mostly, I want to point out that this poll had some rather sophisticated questions that allow for deeper interpretations.

First, the polls suggest that Tuto Quiroga, not Evo Morales, will likely win the presidency after the 18 December elections. While Evo has a 2% lead (30.7% over 28.7%) in the national poll, he's losing in six of nine departments. This despite Evo seen by most Bolivians as the potential winner (when asked "who do you think will win, regardless of your vote?").

Second, the polls demonstrate that while Tuto is the Condorcet winner, Evo is the Condorcet loser. (A Condorcet winner is the candidate that is most preferred in a two-person race against all other candidates; a Condorcet loser is the candidate least preferred in a two-person race against all others.) When asked which candidate was least preferred, most respondents chose Evo (31.2%); Tuto ranked fourth (11%). So that makes Evo a Condorcet loser.

But why is Tuto a Condorcet winner? Because the poll also asked an interesting question: "If a candidate withdrew from the race, how would that change your vote?" If Evo resigned, 38.7% of his voters wouldn't vote for another candidate — but a significant number of those who would (18.7%) would vote for Tuto (19% would vote for Samuel Doria Medina). In contrast, if Tuto resigned, 46.7% of his voters would go to Doria Medina. And if Doria Medina resigned, 44.8% would switch to Tuto. In both cases, fewer voters (12.9% & 13.9% respectively) would move towards Evo.

This suggests three things: First, the electorate is polarized into Evo vs. Tuto & Doria Medina blocks. This could end up producing a Tuto presidency if Doria Medina comes in third, but votes in the legislature to support Tuto, which seems more likely w/ each passing day. Second, there are enough voters in both blocks that can act as "swing voters" between the two ideological camps, suggesting a sizable block of "moderate" voters. Third, it seems that Evo will have great difficulty gaining new voters, but Tuto has more room to expand.

But the key difference is the percentages in the nine departments. Since no candidate will win 50%+1 of the vote, parliament will choose the president. Here, the seats won in the nine departmental legislative districts is critical. Sadly, the figures reported aren't complete, they mostly focus on who's winning, where.

Evo places first in three departments: La Paz (48.5%), Cochabamba (33%), and Oruro (36.5%) That would give him approximately 6 senators & 23 deputies.

Tuto places first in six departments: Santa Cruz (33.8%), Tarija (51.6%), Potosí (36.3%), Beni (37.5%), and Pando (60.7%). That would give him approximately 18 senators & 25 deputies. The senators are key, 18 of 27 senate seats is a majority of the important upper house.

But it seems Evo can win few other seats, since in the other six departments he's at best a very distant second. Tuto is also a distant second in La Paz & Oruro. So both candidates can pick up one or more of the three up-for-grabs senate seats, though Doria Medina might pick up one or two (by coming in second in one or two departments). Evo still suffers from a low voter support in the departments where he's not a strong candidate, while Tuto has enough presence in La Paz, Cochabamba, and Oruro to perhaps win a third of the remaining lower house seats there.

In the end, w/ a month left to go in the campaign, it seems that Evo's support has peaked. But it seems Tuto still has room to expand his base (as do others) by chipping away at Evo's weaknesses (especially his unwillingness to debate). In the end, I'm ready to stand by my previous prediction.

----
NOTE: The poll in question was conducted by the Ipso-Captura polling firm, commisioned by Grupo Líder (an alliance of Bolivian newspapers). It surveyed 2,400 peopld in 46 cities & towns across Bolivia during the first week of November (thus, it might not include any possible "bump" for Evo after his Argentine trip). The margin of error isn't reported, though it's probably around 2%, which means nationally Tuto & Evo are in a statistical dead heat.

Posted by Miguel at 11:20 AM