The surveys are in a situation that is worthy of note: we Know that in recent years have encountered great difficulty in giving account of reality (Brexit, the plebiscite colombian, I do not have the choice of EE.UU, because in that case the national yes were close to the distribution of the vote). And, at the same time we all know that, we continue to use them to calibrate the electoral situation and even continue to be used for the definition of candidates. In the absence of other sources of information are used that are; even if we all think that are not good.
Which brings us to two points. The first is that the use of surveys is fragile. Simply that someone design another way of approaching reality that give quantitative results (and the survey is a way, but not necessarily the only one) and could certainly be done. The second is that, while performing surveys, then there is to worry about solving their problems.
In the chilean case in particular, there are two relevant phenomena:
(1) The first is that we are facing elections with votes volunteers. As far as I know, there is not a clear model of how to identify voters likely. Each who, I think, makes it as you can; but lacking a protocol to clear and public that there are things to do to identify them. And this lack clearly affects the usefulness of political polls.
(2) There is a growing refusal to answer political polls and questions of a political nature in surveys. And as everyone knows, a high rate of attrition makes it difficult to any operation: The value of having high n low, the possibility of solving problems of bias via weighting also becomes problematic.
One might think that these two problems could cancel each other out: That’s just the population with the greatest propensity to not answer is the one with a lower propensity to vote. And if this were so, then ‘doing nothing’ surveys would still be useful. Now, that would have to watch him.
The important thing in this discussion is to observe that, with regard to these surveys -which do not cease to be an important part of the public use of our discipline. there is enough methodological research to perform.