On Wednesday, the 13th of this month, the Association of Sociologists had to make a Panel about Political Polls. In the discussion, regarding the difference between opinion polls and election polls, I mentioned as a comment to an old idea of Jesus Ibáñez: the poll and the election are homologous activities: In both cases it performs the same operation (a selection is made between several alternatives pre-determined); and that this constitutes, at the end of the day, one advantage of the electoral survey on other -where this does not occur.
Incidentally, Bernardo Mackenna -exposing as a panelist – he made the comment that the electoral survey was not the only one who had a schedule external. You can compare the income that you get in MATCH with the data of the Central Bank, we can compare the data of the National Survey of Citizen Security with the crime statistics. And yet, it seems that only the electoral surveys are asked for a correction so precise.
I think that from this discussion you can refine the peculiarity of the electoral survey -and why finally, it is very reasonable that they will ask for a requirement of correction is so high. Let us return to the initial observation. It is not that there are scale external, is that the operation to respond to a survey and vote in an election are the same operation. Then, it is clear that this does not occur in the other cases.
In the surveys of attitude simply because clearly the attitude or belief that one has is not -for nothing – this is equivalent to answering a survey. The response of the survey is always away, may not correspond exactly to what one wants to measure. There is an insurmountable distance. Not for that reason cease to be useful to opinion polls, but that utility is built through that distance.
In the case of other surveys of behaviors it is true that, in principle, there is a ‘correct answer’ that it is possible to determine. There is such a thing as a person’s income, and this can be said. However, the fact remains that the operation of receiving an income (or the situation of having a crime) they are not the equivalent of a vote: to Receive a salary is not to select an alternative from a list, nor is the performing a job or be assaulted. The distance here is not insurmountable, but depending on the situation the question and reality are distance. It is a matter of thinking in all the difficulties that exist in regard to be able to determine the income of the household (it is a matter of reviewing the questionnaire of the MATCH and observe all of the operations that requires a person to select a response). The mapping of the real and the mapping of the survey are not identical. Further, the scale is external to your time is different from the reality that we want to investigate: The income that individuals receive are not the national accounts (there are income that does not appear in them), the crimes that actually happen are not equivalent to the allegations (there are crimes that are not reported, there are allegations that do not always correspond to a crime). While in principle it is possible to perform a test of reality, it is manifestly complex.
The case of the electoral survey, then, is clearly more simple. Select the candidate X for a list of possible candidates is the same operation that I perform to the vote. By the way, the measurement is not trivial, nor transparent, but the closeness of the structural operations facilitates the resolution of these issues (in the last instance, is so equivalent that the CEP can simply do the survey with vote in the urn). Moreover, here the test reality is perfect: The counting of votes is the result of the vote (while the national accounts are not the national income). The electoral survey, which has its own problems, has far less problems of validity of measurement and proof of correction, to be more precise, it allows for a better fit.
Now, this allows for a better adjustment, translates into a ‘demand best fit’. It is reasonable to assume that the MARRIAGE is not the same as the national accounts (and in the times in compliance with the declared income of the couple with these accounts, there were relevant changes). The difficulty of measuring the income, and the fact that the national accounts are not the reality to be measured, has that consequence. But given that the electoral survey there are minor difficulties, and the reality is perfectly known, it is reasonable to ask more precise and have higher demands on the measurement.
In some sense, until a few years ago the electoral survey in Chile was a matter of relatively simple. Now, due to various changes (which, by the way, it is also not clear, in particular, that is behind the recent problems), that measurement has become more complex. And, however, it is still something less complex than what the INE should resolve with employment, the MDS for measuring poverty, and one can continue with other cases. And if, ultimately, those other problems have been resolved (at least, he can deliver a measure that has some meaning and validity), then one should expect that this will also happen with the election polls.