boz offers some reflections on why several pre-election polls in Bolivia said that Evo Morales would not break 35% when in fact he broke 50% (and, apparently, with plenty of room to spare). He finds (as do I) that the most plausible explanations lie in pollsters’ reliance on models that failed to take account of how much more motivated to turn out were Morales’s supporters compared to those of other candidates, and the likelihood that a lot of voters decided very late in the campaign.
Yes, there are a lot of problems w/ Bolivian pre election polls. All of which others have mentioned, but also the fact that they’re not allowed (by law) to poll a week before the election (which is when many people switch, when undecideds break, etc). The last few days of any competitive campaign are crucial, and w/o last-minute information, you really can’t predict more than trends (who’s up, who’s down). Part of why I’m not a fan of using cross-national opinions polls (a la CSES studies).