Confidence Interval

What makes an interval a confidence interval

A confidence interval (CI) is a range of values, calculated from sample data, intended to estimate an unknown population parameter (a mean, a proportion, a difference between groups, a regression coefficient) together with a stated confidence level — most commonly 95%, sometimes 90% or 99%. It is reported as a point estimate plus a margin of error (e.g. “6.2 hours, 95% CI [5.8, 6.6]”) and is standard practice in effect size reporting: a p-value alone tells a reader whether a result is unlikely to have arisen by chance, while a confidence interval also conveys the precision of the estimate.

What makes a reported range an actual confidence interval, rather than just any range built from data, is that it comes from a defined estimation procedure — a point estimate plus a margin of error derived from the sampling distribution of that estimator — applied at a stated confidence level. A range without a specified confidence level and without reference to a sampling distribution (see the counter-example below) is not a confidence interval, whatever it happens to look like.

How to correctly interpret a confidence interval

This is the single most commonly misreported statistical concept in student and even published research writing. The intuitive-sounding interpretation — “there is a 95% probability that the true population mean lies within this interval” — is not the correct frequentist interpretation, and stating it that way in a manuscript is a factual error, not just an imprecise phrasing.

In the frequentist framework (the framework nearly all introductory statistics courses and most published confidence intervals use), the population parameter is treated as a fixed, unknown constant — not a random variable. It either is or is not inside any specific calculated interval; there is no probability left to describe once the sample is drawn and the interval is calculated. The 95% describes the procedure, not this particular interval: if the same sampling and interval-construction procedure were repeated many times on new samples from the same population, approximately 95% of the resulting intervals would contain the true parameter, and about 5% would not. Any single interval is simply one draw from that long-run process — it is either one of the roughly 95% that succeeded, or one of the roughly 5% that did not, and there’s no way to know which from the interval alone.

The correct way to state it: “we are 95% confident that the interval [5.8, 6.6] contains the true population mean,” where “confident” refers to the long-run reliability of the method used to construct the interval — not a probability statement about this specific interval. (A true probability statement about the parameter given the observed data is what a Bayesian credible interval provides instead, which is why the two are easy to conflate but are not interchangeable — see Frequentist vs. Bayesian Statistics for how the two frameworks diverge on this exact point.)

How to find (calculate) a confidence interval

The general form is:

Confidence Interval = Point Estimate ± (Critical Value × Standard Error)

1. Calculate the point estimate — typically the sample mean (x̄) or sample proportion (p̂).
2. Calculate the standard error of that estimate — for a mean, this is the sample standard deviation divided by the square root of the sample size (s / √n).
3. Choose the critical value, which depends on the confidence level and which distribution applies:
   • Use the z-distribution when the population standard deviation is known, or the sample size is large (conventionally n ≥ 30, by the Central Limit Theorem). For a 95% confidence level, the two-sided critical value is z = 1.96.
   • Use the t-distribution when the population standard deviation is unknown and the sample is small — the standard case in most real research. The critical value depends on the degrees of freedom (n − 1) and is read from a t-table or computed in statistical software; it is always slightly larger than the corresponding z-value, which widens the interval to account for the extra uncertainty of estimating the standard deviation from the sample itself.
4. Compute the margin of error (critical value × standard error) and add/subtract it from the point estimate to get the upper and lower bounds.

Two relationships worth knowing when reading or planning a study: for a fixed sample size, a higher confidence level (99% vs. 95%) produces a wider interval, because greater certainty about capturing the true parameter requires a larger margin of error. For a fixed confidence level, a larger sample size produces a narrower interval, because standard error shrinks as n grows — this is the statistical basis for the common design guidance that larger, well-powered samples yield more precise (not just more “significant”) estimates. See experimental design for how sample-size planning connects to this.

Worked examples

Example 1 (mean). A survey of 500 researchers finds a mean of 6.2 hours per week spent on grant-reporting administration, with a 95% CI of [5.8, 6.6] hours. Read correctly, this means: the estimation procedure used here would, if repeated across many independent samples of the same size from the same population, produce an interval containing the true population mean approximately 95% of the time. It does not mean there is a 95% chance the true mean falls in this specific range.

Example 2 (proportion). A study reports that 62% of surveyed early-career researchers have a formal data management plan, 95% CI [57%, 67%]. Reporting the interval alongside the point estimate lets a reader judge precision as well as the estimate itself — a 62% estimate from a sample of 40 people carries a much wider, less useful interval than the same 62% from a sample of 4,000.

What a confidence interval is not

A range reported as, for example, “participants’ ages ranged from 22 to 68” is not a confidence interval — it is simply the sample range (minimum to maximum observed value), with no stated confidence level, no reference to a sampling distribution or estimation procedure, and no claim about a population parameter. Calling a bare min-max range a “confidence interval” is a common but incorrect usage.

A confidence interval is also distinct from a prediction interval, which estimates the range for a single future observation (accounting for both the uncertainty in the estimated parameter and the natural variability of individual observations) rather than for the population parameter itself — a prediction interval is always wider than the confidence interval for the same data and confidence level. It is likewise distinct from a Bayesian credible interval, which does support a direct probability statement about the parameter, but only by starting from a prior distribution rather than treating the parameter as a fixed constant.

Why this matters for reporting research

Reporting confidence intervals alongside point estimates and, where applicable, effect sizes, is now standard or required practice across most major journal style guides and reporting checklists in the social, behavioral, biomedical, and health sciences — it lets a reader evaluate both the magnitude and the precision of a finding, rather than a bare statistically-significant/not-significant verdict. See descriptive statistics vs. inferential statistics for where confidence intervals sit within the broader distinction between summarizing a sample and drawing inferences about a population.