Sampling error margin
There is a sampling error in any survey, since not all the universe is being interviewed. As we work with samples, there is a known and calculated sampling error. Such error is calculated by taking into account sample size and survey results.
In fact, error margin that is usually disclosed refers to a maximum error estimate, considering a simple random sampling model.
For the same size of sample, the greater the homogeneity of surveyed population, the lesser shall be the sampling error and vice-versa.

Contrary to information usually disclosed, there is not a unique sampling error for the survey as a whole, since every information provided  by the survey there is a corresponding sampling error (depending on the result of each question, for example, it shall have a specific error margin).

Regarding political surveys, such errors are generally unequal for several candidates, due to voters’ geographical distribution of each of them, although the survey companies in general disclose a sole error estimate for all survey results.
Thus, when we refer to an error estimate of a certain survey as being 2 percentage points, we understand that a 25% result may have a two-point-variation, to more or less, which should be read as 23% to 27% interval.


 Related News

IBOPE Intelligence points correctly 95% of votes in first round
In all, were voting intention estimates 400 to prefects in 52 municipalities in the country.
Surprises are inevitable
Surprises are inevitable.


 Related Products

Measures the perception people have of a candidate.
Directs and appropriate communication strategies and direction of the campaign.
Election results and daily moving averages, delivered in 24 hours.
Monitor user satisfaction of public and private services.
Online Purchase
In the online store, it is possible to more easily purchase research reports regarding the use of internet in Brasil, or even request a customized study to the customer assistance team on different topics.