Among all the surprises that for many brought the 2016 has appeared on a regular basis: The failure predictive surveys. Or at least that it has been said around the Brexit, the plebiscite of colombia, also around the surveys of the city, and finally, it has been repeated with regard to the recent presidential election in the U.S. it might Be that such a failure has been exaggerated: The surveys Brexit indicated an even result, in the case of EE.U.S. national opinion polls predicted that Clinton would get more votes than Trump, and this is what happened (being the difference in the margin of error). However, there would be defenses weak.
Because the case is that from the surveys were constructed expectations, and do so without misrepresenting the data. Aggregators of polls, and remember that the mere exercise of adding in principle it should increase accuracy, indicating certain results. There is something so much more complex background.
If we refer to the chilean reality, we find ourselves with a number of issues in the research of surveys that require, at least, a test. On the one hand, the surveys face-to-face each day more are found with problems of response rate, and not even the best selection procedure ensures reliable results if the response rate is low. On the other hand, telephone companies are faced with the fact that each year decrease in homes with landline phone, and cell phones, there’s no way of controlling those who are behind the numbers (i.and the relationship between numbers and people is 1-to-1). In the case of the Online, is has expanded the idea of having a pool of interviewees from which to select (more or less randomly), although I don’t know very much about the adequacy of these pools is very guaranteed.
In other words, the social context of the surveys has changed, and the same procedure that years ago allowed to know the reality does not allow at present. A reminder of simple methods and techniques do not fall from the sky nor are they pure things, but depend on the reality that they examine. And when that reality changed, then you have to do is to change the ways of approaching it.
What? There is the temptation, given the huge expansion of data on the social reality, then forget about this strange issue to ask people what they think and what they do, and to refer to what they actually do. Two problems. One is relatively minor -that not everyone is part of those instruments that generate data of conduct. But that is a thing not to forget to do this, but of those universes bounded of whom can speak effectively, they do so with great reliability. The second is really the key: Because it is easy to forget that behavior cannot be derived opinions. The idea that the behavior is revealed preference illuminates things, provided you do not forget what should not be forgotten: That it is not always approve of our behavior, that the level of comfort of our selection is variable; that to repent, that to protect one’s self, are real situations that have consequences, and if someone were to forget it would not hurt to review Ulysses and the Sirens, Elster.
In the period of the zenith of the surveys, many times they seemed to be the paradigm of good knowledge. Now it is clear that they are not. It will be forgivable to the hope that, learning from our mistakes, let us not fall back on the idea that there is a certain technique infallible to be sure that we know the world well. Learn the world is not something that can be done by repeating recipes, and it is the first lesson of methodology that a sociologist should learn.