What Is a Confidence Interval? A Plain-English Guide
When you read that "support for the policy is 52%, with a margin of error of plus or minus 3 points," you're looking at a confidence interval in disguise. Polls, medical studies, A/B tests, and quality-control reports all lean on this single idea to turn a sample into a statement about the whole population.
In this guide, you'll learn what a confidence interval actually is, the formula behind it, and the one interpretation almost everyone gets wrong. We'll walk through a worked example for a mean, unpack what "95% confidence" really promises, and show how sample size and variability change the answer. By the end, you'll be able to read—and build—intervals with confidence.
What Is a Confidence Interval?
A confidence interval is a range of plausible values for an unknown population quantity, calculated from sample data. Instead of reporting a single guess, you report a band that's likely to contain the true value.
You use it because you almost never measure an entire population. You survey 1,000 voters, not all 50 million. You test a sample of 200 light bulbs, not every bulb on the line. Your sample gives you a point estimate—a single best guess like an average or a proportion—but that estimate jiggles depending on which sample you happened to draw. A confidence interval wraps that point estimate in a range that accounts for the jiggle.
For example, a poll might report "52% support, 95% confidence interval of 49% to 55%." The 52% is your point estimate; the 49%–55% range is the confidence interval. It tells you the true level of support across everyone is plausibly somewhere in that band.
The Common Misinterpretation
Here's the trap. A 95% confidence interval does not mean "there's a 95% probability the true value is inside this specific range." Once you've collected your data and computed your interval, the true value either is or isn't inside it—there's no probability left to assign.
What the 95% actually describes is the method. If you repeated your study many times, drawing a fresh sample each time and building a fresh interval, about 95% of those intervals would capture the true value. The confidence level is a property of the procedure over the long run, not a probability about your one particular interval. It's a subtle distinction, but getting it right keeps you from overstating what your data proves.
The Formula: Point Estimate ± Margin of Error
Every confidence interval follows the same simple skeleton:
Confidence Interval = Point Estimate ± Margin of Error
The point estimate is your sample's best guess—a sample mean or sample proportion. The margin of error is how far you reach out on either side. For a mean, the margin of error is:
Margin of Error = Critical Value × Standard Error
The standard error measures how much your sample mean is expected to vary, and it equals the standard deviation divided by the square root of the sample size (s ÷ √n). The critical value comes from your chosen confidence level: for 95% confidence with a large sample, it's about 1.96. Put it together and a confidence interval for a mean is:
Sample Mean ± 1.96 × (Standard Deviation ÷ √Sample Size)
A confidence interval calculator handles the arithmetic and the right critical value for you, but it helps to see the pieces.
A Worked Example for a Mean
Suppose you run a coffee shop and want to know the average daily spend per customer. You sample 100 receipts and find a mean of $12.40 with a standard deviation of $4.00. You want a 95% confidence interval.
First, the standard error: $4.00 ÷ √100 = $4.00 ÷ 10 = $0.40.
Next, the margin of error: 1.96 × $0.40 = $0.784, which rounds to about $0.78.
Finally, build the interval: $12.40 ± $0.78, giving a range of $11.62 to $13.18.
You'd report it like this: "We are 95% confident that the true average spend per customer lies between $11.62 and $13.18." Notice you're making a claim about all your customers, based on a sample of just 100. That's the leverage a confidence interval gives you.
How Sample Size and Variability Affect Width
The width of your interval is the practical payoff, and two factors control it.
Sample size. Because the standard error divides by the square root of n, larger samples produce narrower, more precise intervals. But the square root means diminishing returns: to halve your margin of error, you must quadruple your sample size. Going from 100 to 400 receipts in our example would shrink the margin from $0.78 to about $0.39.
Variability. The more spread out your data, the larger the standard deviation, and the wider the interval. If your customers' spending ranged wildly, you'd need a bigger sample to pin down the average with the same precision. Low-variability populations are easier to estimate.
Confidence level plays a role too. Demanding 99% confidence instead of 95% widens the interval, because you're asking the method to succeed more often. There's an inherent trade-off: more confidence means less precision, and tighter precision means less confidence.
The Link to Margin of Error and Sample Size
Margin of error and confidence intervals are two sides of the same coin—the interval is just the point estimate with the margin of error added and subtracted. That's why pollsters report a margin of error: it lets you reconstruct the interval in your head.
This relationship also runs in reverse, which is enormously useful for planning. Before you collect a single data point, you can decide how narrow you want your interval to be and solve for the sample size that delivers it. If you need a margin of error of plus or minus $0.25 on customer spend, a sample size calculator tells you how many receipts to gather. Designing your study around a target precision beats discovering after the fact that your sample was too small to say anything useful.
Key Takeaways
• A confidence interval is a range, not a single guess, built as point estimate ± margin of error to express the uncertainty in estimating a population value from a sample.
• The common misinterpretation is fatal to clarity: 95% confidence describes the long-run success rate of the method, not the probability that your one specific interval contains the truth.
• For a mean, the interval is Sample Mean ± Critical Value × (Standard Deviation ÷ √n), with the critical value (about 1.96) set by your confidence level.
• Bigger samples narrow the interval, but with diminishing returns—quadrupling the sample size only halves the margin of error, while higher variability widens it.
• Margin of error and sample size are planning levers: decide your acceptable precision first, then size your study to hit it before collecting data.
Confidence intervals translate the messiness of sampling into an honest, readable range. Once you can read the formula and resist the "95% probability" trap, you'll interpret polls, studies, and experiments far more critically. When it's time to crunch the numbers, a reliable confidence interval calculator handles the critical values and arithmetic so you can focus on what the range actually means for your decision.