Confidence Intervals and Margin of Error, Explained Simply
What a survey result of "55 percent ± 3" actually means, and how sample size changes it.
News channels love poll results like "55 percent support, margin of error 3 points". Many readers nod without knowing what it says. It is simpler than it sounds.
What the margin of error means
A poll asks a sample, not everyone. If you ran the same poll again with different people, you would get slightly different numbers. The margin of error tells you how much they could plausibly differ. If the result is 55 percent with a 3-point margin, the true value is probably between 52 and 58 percent.
Sample size is the lever
For a 95 percent confidence level and a result near 50 percent, 400 people give a margin of about 5 points and 1,000 people about 3 points. Quadrupling the sample only halves the margin, so big gains get expensive. Check it with the margin of error calculator and plan your own survey with the sample size calculator.
Confidence intervals for averages
The same idea works for averages. Give the sample mean, the standard deviation and the size, and the confidence interval calculator gives the range. A narrow range means you can trust the average more.
What it does not cover
The margin of error covers random sampling error only. It does not fix a biased sample, confusing questions or people who do not answer honestly. A poll of 10,000 people from one website can still be badly wrong.
A quick check
If two parties are 2 points apart and the margin is 3 points, the race is too close to call from that poll. For the maths behind the bell curve, the normal distribution calculator shows how values spread around the mean.