Overview

Averages of samples vary less than individual values. This finds the chance that a sample's mean falls below or above a target, using the central limit theorem.

Enter the population mean and standard deviation, the sample size and a target mean. Use mode 1 for 'at most the target' or 2 for 'at least the target'.

Worked examples

InputsResult
Population Mean (μ)50Population Std Dev (σ)12Sample Size (n)64Target Sample Mean53Probability ofAt or below targetP(sample mean <= target): 97.72%SE: 1.5z: 295% sampling interval for mean: [47.06, 52.94]
Show 2 more examples
Population Mean (μ)100Population Std Dev (σ)15Sample Size (n)25Target Sample Mean95Probability ofAt or below targetP(sample mean <= target): 4.78%SE: 3z: -1.6795% sampling interval for mean: [94.12, 105.88]
Population Mean (μ)50Population Std Dev (σ)12Sample Size (n)36Target Sample Mean52Probability ofAt or above targetP(sample mean >= target): 15.87%SE: 2z: 195% sampling interval for mean: [46.08, 53.92]

P(sample mean <= target) by population mean

With the calculator's starting values, the p(sample mean <= target) is 97.72%. The table shows how it changes with the population mean.

Population Mean (μ)P(sample mean <= target)
30100%
40100%
50starting value97.72%
600%
700%

Other inputs: Population Std Dev: 12 σ · Sample Size: 64 n · Target Sample Mean: 53. Change them in the calculator above for your own numbers.

How it works

The sample mean has its own spread, the standard error, which is σ ÷ √n. The target is turned into a z-score with that standard error and looked up on the normal curve. For skewed populations, the central limit theorem usually needs a sample of 30 or more to be reliable.

The formula

z = (x̄ − μ) ÷ (σ ÷ √n)
σ ÷ √nStandard error of the mean

Practical tips

  • With μ = 50, σ = 12 and n = 64, the standard error is 1.5, so a sample mean of 53 or less has a 97.72% chance.
  • 95% of sample means fall within about 1.96 standard errors of μ, which is 47.06 to 52.94 here.
  • Quadrupling the sample size halves the standard error.

Frequently asked questions

Why is the spread smaller for sample means?

Highs and lows tend to cancel out within a sample, so averages cluster closer to the true mean.

Does the population need to be normal?

Not for large samples. For small samples from skewed populations, the result is less reliable.

What's the difference from a z-score calculator?

A z-score calculator works on single values. This one works on the mean of a sample.