Définition
The estimates provided by a sample survey deviate slightly from the results that would have been obtained by an exhaustive survey. If the sample is random, the notion of confidence interval can give an idea of that deviation. When a 95% confidence interval is given for a number, this means that the interval has a 95% chance of containing the value that would have been given by exhaustive questioning.
The confidence interval only takes account of the fact that the results come from a random sample survey, and not the other sources of errors: incorrect or misinterpreted responses, non-response bias, etc.