Math calculators

Standard Error Calculator

Updated Jul 27, 2026 By Jehan Wadia
Rate Formulas
Computation Mode
Quick-pick examples

SE of the Mean — Summary Inputs

Spread of individual observations.
Whole number of observations, at least 1.
Optional — only needed to generate a confidence interval preview.

Display Settings

Output sections
Rounding affects display only — all internal calculations use full precision.

Results

SE of the Mean

± 2

Intermediate values

MetricSymbolValue

Population spread vs. sampling distribution

Confidence interval preview

Distribution is chosen automatically: t when n < 30 (df = n − 1), z when n ≥ 30. Proportions always use z.

What this means

Sampling distribution

Step-by-Step Solution


Introduction

The standard error tells you how close your sample result is likely to be to the true answer for the whole group. A small standard error means your estimate is sharp. A big one means your estimate could be off by a lot.

This standard error calculator works three ways:

  • SE of the mean — type in the standard deviation (s) and sample size (n).
  • SE of a proportion — type in p̂ and n, or just the number of successes and trials.
  • SE from raw data — paste your list of numbers and let the tool find s for you.

You get the answer right away, plus the steps that got you there. The tool also shows a confidence interval (90%, 95%, or 99%), a bell curve of the sampling distribution, and a plain-English note on what your number means. It picks the t-distribution when n is under 30 and the z-distribution when n is 30 or more, so you don't have to guess.

Try a quick-pick example to see how it works, then swap in your own numbers.

How to use our Standard Error Calculator

Pick a mode, type in your sample numbers, and the standard error calculator shows the standard error, the key middle steps, a confidence interval, and a bell curve of the sampling distribution.

Computation Mode: Choose SE of the Mean if you already know the standard deviation and sample size, SE of a Proportion if you have a percent or a success count, or SE from Raw Data if you have a list of numbers.

Quick-pick examples: Click any chip to load a ready-made example. It fills the fields and runs the math right away.

Sample standard deviation (s): Type how spread out your single values are. It must be 0 or more.

Sample size (n): Type how many observations you have. Use a whole number, at least 1. Bigger n gives a smaller standard error. If you are still planning your study, our sample size calculator can help you pick n before you collect data.

Sample mean (x̄) — optional: Type your average if you want a confidence interval and a labeled bell curve. Leave it blank to get the standard error only. Not sure of your average yet? The mean, median, mode calculator will work it out.

How do you want to supply p̂: Pick Direct entry to type the proportion yourself, or Count entry to let the tool divide successes by trials.

Sample proportion (p̂): Type a decimal from 0 to 1, like 0.42 for 42%. For setting up ratios of successes to trials, see the proportion calculator.

Number of successes (x): Type how many times the thing you counted happened. It cannot be more than n.

Sample size / trials (n): Type the total number of people or trials in your survey or test.

Data values: Paste your numbers with commas, spaces, tabs, or new lines. You need at least 2 values, and the tool finds the mean, variance, standard deviation, and standard error for you.

Output sections: Check the boxes to show or hide the step-by-step solution, the plain-language meaning, and the sampling distribution chart.

Rounding / precision: Choose how many decimal places to display. This changes the look only, not the math. For separate work on rounding or significant figures, we have dedicated tools.

Confidence level: Pick 90%, 95%, or 99% to build a confidence interval. The tool uses t when n is under 30 and z when n is 30 or more, and always z for proportions. You can look up any cutoff with the critical value calculator.

Buttons: Press Calculate to run the numbers, Reset to go back to the start values, or Download results (PNG) to save your answer as an image.

What Is Standard Error?

The standard error (SE) tells you how much a sample result is likely to miss the true value for the whole group. When you take a sample, your sample mean will almost never match the real population mean exactly. The standard error puts a number on that normal, expected gap.

The Standard Error Formulas

There are two common versions, and both are simple:

  • Standard error of the mean: SE = s ÷ √n, where s is the sample standard deviation and n is the sample size. You can check s with the standard deviation calculator and the √ step with the square root calculator.
  • Standard error of a proportion: SE = √(p̂(1 − p̂) ÷ n), where is the share of "yes" answers in your sample.

Standard Error vs. Standard Deviation

People mix these up often, but they measure different things:

  • Standard deviation (s) shows how spread out the individual data points are.
  • Standard error (SE) shows how spread out the sample means would be if you repeated the study many times.

Because you divide by √n, the standard error is always smaller than the standard deviation (when n is more than 1). With n = 36, the SE is 6 times smaller than s. If you want the full spread picture of your data, the range calculator and IQR calculator describe it in other ways.

Why Sample Size Matters

A bigger sample gives a smaller standard error, which means a more exact estimate. But the gain slows down, because you divide by the square root of n. To cut your standard error in half, you need four times as much data. Going from 100 to 400 people cuts the SE in half; going from 100 to 200 only cuts it by about 29% — a change you can verify with the percent change calculator.

How Standard Error Is Used

Standard error is the building block for confidence intervals and hypothesis tests. A confidence interval is just your estimate plus and minus a critical value times the SE, and that second part is exactly what the margin of error calculator reports:

  • Use a z value when the sample is large (n ≥ 30), such as 1.96 for 95% confidence — see the z score calculator and normal distribution calculator.
  • Use a t value when the sample is small (n < 30), with degrees of freedom equal to n − 1. The t test calculator uses this same SE in its test statistic.
  • Proportions use z values.

About 68% of sample means fall within 1 SE of the true mean, and about 95% fall within 2 SE. That is why scientists, pollsters, and doctors report standard error or margin of error next to their results. Once you have a test statistic, the p value calculator turns it into a significance level, and the effect size calculator shows how large the difference actually is.

Quick Tips

  • Small SE = precise estimate. Large SE = shaky estimate.
  • Never compare SE values across studies with different units or scales.
  • For proportions, check that n·p̂ and n(1 − p̂) are both 10 or more. If not, the normal formula can give misleading results — a binomial distribution calculator is safer for small counts.
  • Always use the sample standard deviation (dividing by n − 1) when working from raw data.
  • Screen your data for extreme values first with the outlier calculator, since one stray point can inflate s and your SE.
  • For a broader summary of your dataset in one place, try the statistics calculator, or check relationships between two variables with the correlation coefficient calculator and linear regression calculator.

Formulas used

Standard error of the mean
SE_{\bar{x}} = \frac{s}{\sqrt{n}}
Standard error of a proportion
SE_{\hat{p}} = \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}
Sample proportion from counts
\hat{p} = \frac{x}{n}
Sample mean from raw data
\bar{x} = \frac{\sum_{i=1}^{n} x_i}{n}
Sample standard deviation from raw data
s = \sqrt{\frac{\sum_{i=1}^{n}(x_i - \bar{x})^2}{n-1}}
Confidence interval using the t-distribution (n < 30)
\bar{x} \pm t^{*}_{\alpha/2,\, df = n-1} \cdot SE
Confidence interval using the z-distribution (n \ge 30, or proportions)
\hat{\theta} \pm z^{*}_{\alpha/2} \cdot SE
Normal approximation check for proportions
n\hat{p} \ge 10 \quad \text{and} \quad n(1-\hat{p}) \ge 10

Frequently asked questions

Do I need to enter the sample mean to get a standard error?

No. The standard error of the mean only needs s and n. The mean (x̄) is optional.

Add it if you want the confidence interval, the labeled bell curve, and the population-vs-sampling table. Leave it blank and you still get the SE.

Why does my answer show a plus-minus sign?

Because the standard error works in both directions. Your sample result can land above or below the true value.

So an SE of 2 means your sample mean is likely off by about 2 in either direction.

Can I type 42 instead of 0.42 for the proportion?

No. The p̂ box needs a decimal from 0 to 1, so 42% must be typed as 0.42. Typing 42 shows an error.

To skip the math, pick Count entry and type the successes and trials instead.

Can the standard error be zero?

Yes. If your standard deviation is 0, or if p̂ is exactly 0 or 1, the SE is 0.

That means every value in your sample was the same. The chart is hidden because there is no spread to draw.

Why does the calculator divide by n minus 1 for raw data?

Because your list is a sample, not the whole group. Dividing by n − 1 keeps the variance from coming out too small.

This is the sample standard deviation, and it is the right one to use for standard error.

What is the green sample-size note under my answer?

It shows what would happen if you added about 10% more data. It gives the new SE and the percent drop.

Use it to see if collecting more data is worth the effort before you start.

Why did I get an orange warning on my proportion?

It shows up when the normal shortcut may not fit your numbers. That happens when n is under 30, or when n·p̂ or n(1 − p̂) is below 10.

The SE still shows, but treat it with care and think about an exact or Wilson interval instead.

Why is my bell curve centered on 0?

Because you did not enter a sample mean. With no center to use, the tool centers the curve on zero and labels the axis as the distance from your mean.

Type your mean in the x̄ box to see real values on the axis.

What does the Smart rounding option do?

It picks a sensible number of digits for you. Big numbers get 2 decimals, mid-size numbers get up to 4, and tiny numbers get 4 useful digits.

All rounding is display only. The math behind it always uses full precision.

Can I paste data copied from a spreadsheet?

Yes. Paste straight from a column or row. Commas, spaces, tabs, semicolons, and line breaks all work, and you can mix them.

Any entry that is not a number gets flagged so you can fix it.

Does the raw data mode accept negative numbers and decimals?

Yes. Values like -4.5 and 0.02 are fine.

One warning: do not use a comma as a decimal point. Typing 3,5 is read as two values, 3 and 5.

Is standard error the same as SEM?

Yes, when you are talking about a mean. SEM stands for standard error of the mean, which is the first mode of this tool.

Proportions have their own standard error, so SEM does not apply there.

Can I use a population standard deviation in the s box?

Yes. Type σ in the s field and you get σ ÷ √n.

Just note that the confidence interval still switches to t when n is under 30. With a known σ, a z interval is the textbook choice.

Why is my confidence interval section asking for a mean?

An interval needs a center point. Without x̄, there is nothing to add and subtract the margin of error from.

Enter the mean, then choose 90%, 95%, or 99% from the dropdown.

Why is my t interval wider than a z interval?

Small samples carry more doubt, so the t-distribution has fatter tails and a bigger critical value.

That makes the interval wider. As n grows, t gets closer and closer to z.

Can this tool compare two samples?

No. It handles one sample at a time.

For the standard error of a difference between two means or two proportions, you need a two-sample tool or test.

Does it adjust for a small population?

No. There is no finite population correction here.

If your sample is more than about 5% of the whole group, the true SE is a bit smaller than what this tool shows.

How do I save my results?

Click Download results (PNG). It saves an image with your inputs, key values, the interval, and the steps.

It works only after a valid calculation, so fix any red errors first.

Do I have to press Calculate every time?

No. Results update as you type or change a setting.

The Calculate button is there if you want to force a refresh, and Reset puts everything back to the starting example.

What sample size gives a good standard error?

There is no single magic number. It depends on how spread out your data is and how precise you need to be.

Use the sensitivity note and try a few values of n to see how fast the SE shrinks for your data.