To state a confidence level of 95% (pollsters will say "accurate 19 times out of 20"), the margin of error is calculated from a simple equation if we assume a normal population distribution (we do). Margin of error at the 95% confidence interval is calculated like so 1.96((0.5(1-0.5))/n)^0.5 or 0.98 divided by the square-root of n where n is the number in the sample which is to be representative of the population. so, with a sample of 3,400 homes, that's a direct margin of error of (0.98 / (3400)^0.5) = 1.68%. This formula gives margins of error of 3.1% on polls involving 1000 people which is usually the standard sample size for political polls. When we wish a sample to be representative of a normally distributed population we state the margin of error. If we want a 19 times out of 20 confidence level we use the equation above. If we say 99 times out of 100 the margin of error equation becomes 1.29/n^0.5 and the margin of error goes up. The point is, just by stating the sample size and the confidence interval, the margin of error is implicitly stated.