How to Interpret a Confidence Interval (Without Getting It Wrong)

A confidence interval is not a probability. Learn the correct frequentist interpretation, recognize the most common misconceptions, and understand what overlapping intervals actually mean.

The One Sentence to Memorize

If we repeated this sampling process many times and calculated a 95% confidence interval each time, approximately 95% of those intervals would contain the true population parameter.

Notice: we are not saying there is a 95% probability that the true value lies in this specific interval. That would be a Bayesian credible interval, which is different. The 95% is a property of the method, not of any single interval.

The interpretation almost everyone gets wrong

Students, journalists, and even some researchers misinterpret confidence intervals. Here are the most common errors, why they are wrong, and what to say instead:

Common (Wrong) Phrasing Why It's Wrong Correct Alternative
"There is a 95% probability the true mean lies in this interval." The true parameter is fixed (either in the interval or not). The probability is either 0 or 1. The 95% refers to the method repeating, not to this specific interval. "If we repeated the experiment many times, about 95% of the intervals we construct this way would contain the true mean."
"We can be 95% confident the true value is in the interval." "Confident" sounds like a personal belief about this particular interval, which is incorrect in frequentist logic. The 95% is objective (method-based), not subjective. "The method used to compute this interval captures the true parameter about 95% of the time."
"95% of the data falls within the confidence interval." This describes a prediction interval or a range containing 95% of observations, not a confidence interval for a parameter. "A 95% confidence interval is a range that, by the method's design, contains the parameter estimate with 95% long-run coverage."
"The confidence interval is the range where the true parameter definitely lies." Nothing is definite. The true parameter might be outside this interval (it happens ~5% of the time for a 95% CI). "A confidence interval is a range that is constructed using a method that captures the true parameter about 95% of the time."
"If I compute 100 confidence intervals, 95 of them will contain the true value." Technically true in expectation, but it assumes each interval has the same level and the sampling is random. In practice, once computed, each interval either contains it or doesn't (probability 0 or 1). "If we use this method to compute 100 intervals from 100 independent samples, we expect roughly 95 to capture the true parameter."

Reading a confidence interval's width

The width of a confidence interval tells you how precisely you have estimated the parameter.

Narrower = More Precise

A narrow confidence interval (e.g., 95% CI = [102.3, 105.7]) means your estimate is precise. A wide interval (e.g., 95% CI = [85.1, 122.9]) means there is more uncertainty.

Sample Size Drives Width

The larger your sample, the narrower your confidence interval. This is why the formula for CI width includes 1/√n — doubling your sample size reduces the margin of error by a factor of √2 ≈ 1.41. To make a meaningful improvement in precision, you often need a substantially larger sample. If you want to narrow a CI, use our sample size calculator to plan your next study.

Variability in Your Data Matters Too

Even with a fixed sample size, high variability in your measurements (high standard deviation) produces a wider CI. You control sample size; you cannot control variability—but you can measure and report it honestly.

Overlapping confidence intervals ≠ no significant difference

One of the most misunderstood facts in statistics: two confidence intervals can overlap and the parameters they estimate can still be statistically significantly different. Here's why:

Two overlapping 95% confidence intervals that still represent a statistically significant difference. Group A's interval is roughly 102 to 118; Group B's interval is roughly 108 to 124. The intervals overlap visibly from 108 to 118, yet a proper two-sample t-test of the difference between the groups would reject the null hypothesis of no difference. Overlapping CIs Can Still Show Significant Difference Group A 95% CI: [102, 118] Group B 95% CI: [108, 124] Overlap Even overlapping CIs can yield p < .05 for the difference test.
Fig. 1 — Two 95% confidence intervals that visibly overlap yet can still correspond to a statistically significant difference between the two estimates. The width and position of each interval, and the correlation between estimates (if they come from the same sample), all affect whether the difference is significant.

Why does this happen? Because a confidence interval for a single parameter is not the same as a confidence interval for the difference between two parameters. When you test whether Group A and Group B are significantly different, you are testing the difference (A − B), and the standard error of a difference is not simply the overlap or non-overlap of the individual CIs.

A rough rule of thumb: if the 95% CI for one group lies entirely outside the 95% CI for the other, then the difference is definitely significant (p < .05). But if the intervals overlap, you cannot conclude anything without conducting the actual test of the difference. Overlapping CIs do not guarantee the difference is non-significant.

Always test the hypothesis directly using the appropriate test (t-test, ANOVA, etc.) rather than trying to infer from visual CI overlap.

Worked example: interpreting a real confidence interval

A researcher measures the average height of 50 undergraduate students and obtains:

Mean height: 171.2 cm

95% Confidence Interval: [169.4, 173.0]

Correct interpretation: "If this researcher repeatedly sampled 50 students and calculated a 95% confidence interval each time using the same method, approximately 95% of those intervals would contain the true average height of the population of undergraduates."

What this does NOT mean:

Practical use: You can report that the best estimate of average student height is 171.2 cm, and the plausible range (given sampling variability) is approximately 169.4 to 173.0 cm. This interval reflects the uncertainty in your estimate due to sampling. If you want a narrower interval, you would need a larger sample.

FAQ

Does a 95% CI mean there's a 95% probability the true value is inside it?

No. In frequentist statistics, the true parameter is a fixed (but unknown) constant. Once an interval is computed, the true value is either inside it or not—the probability is 0 or 1. The 95% refers to the long-run frequency: if you repeated the experiment and calculation many times, about 95% of the intervals would contain the true parameter. This is a property of the method, not of any single interval. If you want a probability statement about this specific interval, you would need a Bayesian credible interval, which is a different calculation.

What does it mean if a CI for a difference includes zero?

If a 95% confidence interval for the difference between two groups includes zero, it is consistent with the hypothesis that there is no difference at the α = .05 level. You would fail to reject the null hypothesis (H₀: difference = 0). However, this does not prove there is no difference—it only means the data do not provide sufficient evidence to conclude there is one. A wider interval (caused by small sample size or high variability) is more likely to include zero even if a true difference exists.

Why is my confidence interval so wide?

A wide confidence interval reflects high uncertainty in your estimate. Common causes are: (1) small sample size — larger samples give narrower intervals; (2) high variability in your data — measurements that are spread out produce wider intervals; (3) high confidence level — a 99% CI is wider than a 95% CI from the same data; (4) small effect size — when the estimate itself is close to the null, uncertainty can look proportionally large. To narrow an interval, collect more data or reduce measurement variability (if possible).

Can two overlapping confidence intervals still be significantly different?

Yes. A confidence interval for a single parameter is not the same as a confidence interval for the difference between two parameters. Two overlapping 95% CIs can still correspond to p < .05 for a test of the difference. Conversely, two non-overlapping CIs do guarantee the difference is significant (p < .05). Always conduct the formal statistical test rather than trying to infer significance from visual CI overlap.

Is a 99% or 95% confidence interval more precise?

A 95% confidence interval is more precise (narrower). A 99% CI is wider because it has to capture the true parameter 99% of the time instead of 95%. If you want both higher confidence (99%) and a narrow interval, you must increase your sample size. There is always a trade-off: for a fixed sample size, higher confidence means a wider interval. Choose the confidence level based on the consequences of being wrong (α = .05 is conventional but not universal).

How does sample size affect confidence interval width?

The margin of error (half-width of a confidence interval) is inversely proportional to √n, where n is sample size. This means doubling the sample size reduces the margin of error by a factor of √2 ≈ 1.41, or roughly 30%. To cut the margin of error in half, you need to quadruple the sample size. Use our sample size calculator to determine how many participants you need for a desired precision level.

Cross-references

Explore related tools and concepts:

Sources