Free Online Standard Error and Confidence Interval Calculator
Work out the standard error of a mean or a proportion and build a confidence interval from it. On small samples the t distribution is used rather than z, which widens the interval correctly, and the difference between the two is shown so you can see what using the wrong one would have cost.
The sum runs in your browser. Nothing you type is sent anywhere.
📐 The sample
The mean, the standard deviation and how many there were. Or paste the numbers and have all three worked out.
📐 The counts
How many had the thing, out of how many you looked at.
| Step | What happens |
|---|
What a different sample size would have done, everything else unchanged. The interval narrows with the square root, so four times the data halves the margin.
| Sample size | Standard error | Multiplier | Margin of error |
|---|
| When | Calculation | Actions |
|---|
Number formatting
Rounding only changes what you see. The sum itself always runs at full precision, so a rounded number never feeds into the next step.
Display
Full screen hides the page around the tool. Press Escape, or the button in the bar, to come back.
History
Your data
Settings, history and saved setups live in this browser and nowhere else. There is no account and no server, which also means clearing them here is final.
Enjoying WizTools123? Help keep our server infrastructure 100% free and open for everyone.
How to Use Standard Error and Confidence Interval Calculator
Step by step, start to finish.
The mean, the standard deviation and the count, or paste the numbers and have them worked out. For a proportion, how many had the thing and out of how many.
Ninety five per cent is conventional rather than special. A higher level gives a wider interval, because being more certain of containing the answer means claiming less about where it is.
The sentence is the one you can honestly say out loud. The common phrasing, that there is a 95 per cent chance the true value is in this range, is shown next to it as the thing not to say.
Key Features
What this tool does, and what it does not do.
The multiplier comes from the t distribution with the right degrees of freedom, and the z value is shown beside it for comparison.
Two genuinely different standard errors, with the proportion one guarded against the cases where it fails.
Which stays within zero and one and works with small counts, including no successes at all.
Written out properly, with the common wrong version beside it so the difference is visible.
The square root law made concrete, which is what you want before deciding to collect more data.
It runs in this tab and keeps working offline.
About Standard Error and Confidence Interval Calculator
A sample mean is an estimate, and an estimate without a margin is half a result. The standard error says how much that estimate would bounce around if you drew a different sample, and the confidence interval turns it into a range. The arithmetic is short: standard error is the standard deviation divided by the square root of the sample size, and the interval is the mean plus or minus a multiplier times that.
The multiplier is where tools go wrong. If you do not know the population standard deviation, and almost nobody does, the right multiplier comes from the t distribution and depends on how much data you have. With five observations the 95 per cent multiplier is 2.776, not 1.96, which makes the interval 42 per cent wider. A calculator that uses 1.96 regardless hands you a tighter interval than your data supports, and tighter intervals are exactly what people want to believe. By the time you have a thousand observations the two agree to three decimal places, so the error only bites on small samples, which is where most real work happens.
The second thing worth getting right is what the interval means, because the usual phrasing is wrong in a way that matters. Saying there is a 95 per cent chance the true value lies in this range treats the true value as something that moves. It does not; it is a fixed number you do not know. What moves is the interval, which depends on the sample you drew. The honest statement is that if you repeated the whole exercise many times, 95 per cent of the intervals constructed this way would contain the true value. The page writes that sentence out with your own numbers in it, and shows the wrong version alongside, because seeing them together is the only way the distinction sticks.
Proportions get their own treatment because the standard formula misbehaves at the edges. One success out of fifty produces an interval reaching below zero; none at all produces an interval of zero width, which claims perfect certainty from the weakest possible evidence. Both come from a normal approximation that fails when counts are small. The Wilson interval is shown beside it: it never leaves the range zero to one, and with no successes out of thirty it gives nought to about eleven per cent, which is a real answer to a real question.
Frequently Asked Questions
Common questions about this tool.