Population Proportion: Definition, Formula and Worked Example

A population proportion is the share of a whole population that has a particular characteristic, written p and always between 0 and 1. If 842 of every 1,000 people in a population have the characteristic, the population proportion is 0.842 (84.2%). Because the whole population is rarely measured, p is usually estimated from a sample proportion, written p′ or p̂. This guide gives the definition, the formula with every symbol explained, a fully worked confidence interval with the arithmetic shown, reading guidance, the conditions that must hold, an APA reporting line, and code for Excel, R and Python.

Definition: what is the proportion of a population?

The proportion of a population is the number of members with a characteristic (the “successes”) divided by the total number of members: p = X / N. It is a parameter: a fixed number describing the whole population, usually unknown. The matching number computed from a sample is a statistic, the sample proportion, which serves as the point estimate of the true population proportion (OpenStax, Introductory Statistics 2e, section 8.3). The NIST/SEMATECH e-Handbook uses the same idea for quality work, where p is the true proportion defective and its estimate is the number of defects divided by the sample size.

The question “what is proportion in statistics?” has the same answer in plain words: a part divided by a whole, expressed as a decimal. Multiply by 100 for a percentage; the two are the same quantity written differently (see proportion vs percentage). For the parameter/statistic distinction in general, see parameter in statistics and population vs sample.

Formula and every symbol explained

Symbol	Meaning
p	Population proportion (parameter): successes in the population divided by population size. Usually unknown.
x	Number of successes in the sample.
n	Sample size.
p′ (p-hat)	Sample proportion (statistic): p′ = x / n. The point estimate of p.
q′	Proportion of failures in the sample: q′ = 1 − p′.
zα/2	The standard normal value that leaves α/2 in the upper tail (1.96 for 95% confidence).
EBP	Error bound for a population proportion, the margin of error.

The estimate is p′ = x / n. The confidence interval for p, in the form OpenStax gives it, is:

EBP = z_α/2 × √( p′q′ / n ), and the interval is (p′ − EBP, p′ + EBP).

The quantity √( p′q′ / n ) is the estimated standard error of the sample proportion. For the general idea of an interval, see confidence interval.

Worked example with the arithmetic shown

This is the setting of OpenStax Example 8.10, recomputed here step by step: a sample of n = 500 contains x = 421 successes. Estimate the population proportion with a 95% confidence interval.

1. Sample proportion: p′ = 421 / 500 = 0.842.
2. Failures: q′ = 1 − 0.842 = 0.158.
3. Check the conditions: np′ = 500 × 0.842 = 421 and nq′ = 500 × 0.158 = 79. Both are greater than 5, so the normal approximation is acceptable under the OpenStax rule.
4. Standard error: √(0.842 × 0.158 / 500) = √(0.133036 / 500) = √0.000266072 ≈ 0.01631.
5. Critical value for 95% confidence: z_0.025 = 1.96.
6. Error bound: EBP = 1.96 × 0.01631 ≈ 0.0320, which rounds to 0.032.
7. Interval: 0.842 − 0.032 = 0.810 and 0.842 + 0.032 = 0.874.

The 95% confidence interval for the population proportion is (0.810, 0.874). With this method, about 95% of intervals built from repeated samples would contain the true p; it does not say p has a 95% chance of lying in this particular interval.

The same data with a Wilson interval

The NIST/SEMATECH e-Handbook (section 7.2.4.1) gives an interval built around the sample proportion with z²/(2n) terms and a denominator of 1 + z²/n, and recommends an exact binomial method when the sample is very small or failures are very few. Applying that formula to the data above (z = 1.96) gives centre = (0.842 + 1.96²/1000) / (1 + 1.96²/500) = 0.845842 / 1.007683 ≈ 0.8394, half-width ≈ 0.0320, so the interval is about (0.807, 0.871). With n = 500 the two methods agree to the second decimal; they diverge as n shrinks or p′ nears 0 or 1.

Interpretation: reading the result

What you see	How to read it
p′ = 0	No sample member has the characteristic. The normal-approximation interval collapses to a single point, so use an exact method.
p′ = 0.5	Half the sample has the characteristic; the standard error √(p′q′/n) is at its largest for a given n.
p′ = 1	Every sample member has the characteristic; the same collapse applies, so use an exact method.
p′ = 0.842, interval (0.810, 0.874)	The estimate is 84.2%; plausible values of p under this method run from 81.0% to 87.4%.
np′ or nq′ at or below 5	The OpenStax condition fails; do not use the normal-approximation interval.

A narrower interval means a more precise estimate; because n sits under the square root, quadrupling the sample size halves the error bound. For planning n before collecting data, see sample size for a quantitative study.

Assumptions and common mistakes

• Successes and failures must both be large enough. OpenStax requires that np′ and nq′ are both greater than five.
• Random sampling. The interval describes sampling error only. A convenience sample can be far from the population however narrow the interval is.
• Confusing p and p′. p is the fixed population value; p′ changes from sample to sample. Report the estimate as p′, never as p.
• Mixing up proportion and percentage. 0.842 and 84.2% are the same value; write 0.842 where a formula needs a proportion.
• Misreading the interval. It is a statement about the method over repeated samples, not a probability that p is inside one computed interval.
• Using the formula on tiny samples. For very small n or very few failures, NIST recommends an exact binomial method.

To compare two proportions rather than estimate one, see the two-proportion z-test.

How to report it (APA example)

Report the sample proportion, the sample size and the interval with its confidence level. A line in APA style: “Of 500 respondents, 421 met the criterion (p̂ = .84, 95% CI [.81, .87]).” State the method (for example, normal approximation or Wilson) once in the analysis section.

How to calculate it in Excel, R or Python

Excel

A1: 421            (successes, x) A2: 500            (sample size, n) A3: =A1/A2                              sample proportion, 0.842 A4: =SQRT(A3*(1-A3)/A2)                 standard error, 0.01631 A5: =NORM.S.INV(0.975)*A4               error bound, 0.0320 A6: =A3-A5     A7: =A3+A5               interval, 0.810 and 0.874

R

x = 421; n = 500 phat = x / n se = sqrt(phat * (1 - phat) / n) c(phat - qnorm(0.975) * se, phat + qnorm(0.975) * se)   # 0.810 0.874 prop.test(x, n, correct = FALSE)$conf.int               # Wilson-type, about 0.807 0.871

Python

from math import sqrt x, n = 421, 500 phat = x / n se = sqrt(phat * (1 - phat) / n) z = 1.96 print(phat, phat - z * se, phat + z * se)   # 0.842 0.810 0.874

FAQ

What is a population proportion?

It is the fraction of all members of a population that have a given characteristic, written p, a value from 0 to 1. It is a parameter and is usually estimated from a sample.

What is the formula for a population proportion?

p = successes in the population / population size. When estimated from a sample, p′ = x / n, with a confidence interval of p′ ± z_α/2 √(p′q′/n).

What is the difference between proportion and percentage?

A percentage is the proportion multiplied by 100. A proportion of 0.842 is 84.2%.

When can I use the normal-approximation interval?

OpenStax states the condition that np′ and nq′ are both greater than five. When it fails, use an exact binomial method, as NIST recommends for very small samples.

Sources

• OpenStax. Introductory Statistics 2e, section 8.3, “A Population Proportion”, and Example 8.10. openstax.org/books/introductory-statistics-2e/pages/8-3-a-population-proportion (date checked: 2026-10-29).
• NIST/SEMATECH e-Handbook of Statistical Methods, section 7.2.4.1. itl.nist.gov/div898/handbook/prc/section2/prc241.htm (date checked: 2026-10-29).