statistics

95% Confidence Interval Calculator

Q: Why is 1.96 the Z-score for 95% confidence?

In a standard normal distribution, 1.96 standard deviations from the mean capture approximately 95% of the data. This arises from the cumulative normal function: the area between −1.96 and +1.96 equals 0.95. For different confidence levels, the Z-score changes: 1.645 for 90%, 2.576 for 99%. Statistical software and tables provide precise values for any desired confidence level.

Q: Should I use Z-score or t-distribution?

Use the Z-distribution when you know the population standard deviation or have a large sample (n ≥ 30). Use the t-distribution for smaller samples where you estimate standard deviation from the sample. The t-distribution is more conservative, yielding slightly wider intervals—this accounts for extra uncertainty from small-sample estimation. As sample size grows, t approaches Z.

Q: How does sample size affect the confidence interval?

The standard error contains √n in the denominator, so it decreases as sample size increases. Doubling the sample size reduces the standard error by a factor of √2 ≈ 1.41, thus narrowing the interval proportionally. To halve the margin of error, you need to quadruple the sample size. This non-linear relationship means very large samples are required for very tight intervals, making cost–benefit analysis essential in study design.

Q: Can I use a confidence interval to test hypotheses?

Yes. If a hypothesized parameter value falls outside your confidence interval, you would reject it at the corresponding significance level (e.g., reject at α = 0.05 if it's outside the 95% CI). Conversely, if it falls inside, you lack evidence to reject. This equivalence between confidence intervals and hypothesis tests makes CIs more informative: they show not just whether an effect exists, but the range of plausible values.

Q: What if my data is not normally distributed?

Small deviations from normality are often acceptable, especially with larger samples, due to the Central Limit Theorem. For visibly non-normal data, consider a non-parametric confidence interval or transformation. Bootstrap methods, which resample your data repeatedly, provide robust alternatives when normality assumptions are violated. Many statistical packages now offer bootstrap CIs without additional assumptions.

Q: How do I report a confidence interval in research?

Standard format: state the estimate, interval bounds, and confidence level. Example: "The mean weight loss was 4.5 kg (95% CI: 3.2–5.8 kg)." Always include the confidence level; readers cannot interpret the interval without it. When publishing, specify whether you used Z or t, and mention any transformations or adjustments. Wider intervals are not a weakness—they honestly reflect greater uncertainty.

A 95% confidence interval establishes the range within which you can reasonably expect a population parameter to fall, based on sample data. Researchers, quality assurance managers, and data analysts use this calculation to quantify uncertainty around estimates. Given a sample mean, standard deviation, and sample size, the calculator determines the lower and upper bounds—providing both margin of error and the z-score relationship that defines your confidence level.

Last updated: May 7, 2026

Creators Davide Borchia, PhD

Reviewers Anna Szczepanek and Wojciech Sas

8,891 people find this calculator helpful

Confidence Interval Formula

The 95% confidence interval is constructed by combining three key calculations: standard error, margin of error, and the bounds themselves. Each component builds on the previous one.

SE = σ ÷ √n

ME = SE × Z

Lower Bound = μ − ME

Upper Bound = μ + ME

σ (sigma) — Standard deviation of the sample, measuring spread around the mean
n — Sample size; larger samples yield narrower intervals
SE — Standard error; the standard deviation of the sampling distribution
Z — Z-score corresponding to your chosen confidence level (1.96 for 95%)
ME — Margin of error; the distance from the mean to each bound
μ (mu) — Sample mean, the central estimate of the population parameter

Understanding Confidence Intervals

A 95% confidence interval does not mean there is a 95% probability that the true value lies within your computed bounds. Rather, it reflects the long-run frequency: if you repeated your sampling and calculation 100 times, approximately 95 of those intervals would contain the true parameter.

The width of your interval depends on three factors:

Sample size: Larger samples produce narrower intervals because standard error decreases with √n
Variability: High standard deviation broadens the interval; low variability tightens it
Confidence level: Choosing 99% instead of 95% widens the interval (higher Z-score)

For a 95% confidence level, the critical Z-score is 1.96, representing approximately two standard deviations from the mean in a normal distribution.

Practical Interpretation

Consider a market researcher surveying 100 consumers about satisfaction ratings. The sample mean is 7.5 out of 10, with a standard deviation of 1.2. The standard error becomes 1.2 ÷ √100 = 0.12. With Z = 1.96, the margin of error is 0.12 × 1.96 ≈ 0.24.

The 95% confidence interval spans from 7.26 to 7.74. The researcher can state: "Based on this sample, the true population mean satisfaction rating lies between 7.26 and 7.74 with 95% confidence." This interpretation applies to any field—medical trials, engineering tolerances, survey research, or quality control.

Common Pitfalls and Considerations

Avoid these frequent mistakes when working with confidence intervals.

Confusing confidence level with probability — A 95% CI does not indicate a 95% chance the parameter is inside the interval. The true value is fixed; your interval varies across samples. Think of it as a long-run property of the method, not a single-trial probability.
Assuming normality without verification — The standard Z-based approach assumes your data follow a normal distribution. If your sample is small (<30) and the distribution is skewed, use a t-distribution instead. Many tools allow this adjustment in advanced mode.
Ignoring sample size constraints — Very small samples (n < 10) produce wide, less useful intervals. Before collecting data, perform a power analysis to determine the sample size needed for your desired precision.
Overlooking the impact of variability — High standard deviation widens the interval dramatically. Before computing, assess measurement consistency and reduce noise through improved measurement methods or experimental design.

Alpha and P-Value Relationship

The confidence level and significance level are complementary. For a 95% confidence interval, alpha (α) equals 0.05. This represents the two-tailed rejection region: 0.025 in each tail of the distribution. The right-tailed p-value for a 95% CI is 0.025, corresponding to the upper critical value where Z = 1.96.

If you shift to 99% confidence, alpha drops to 0.01, each tail contains 0.005, and the Z-score increases to 2.576, yielding a wider interval. This trade-off between confidence and precision is central to statistical decision-making.

Frequently Asked Questions

Why is 1.96 the Z-score for 95% confidence?

In a standard normal distribution, 1.96 standard deviations from the mean capture approximately 95% of the data. This arises from the cumulative normal function: the area between −1.96 and +1.96 equals 0.95. For different confidence levels, the Z-score changes: 1.645 for 90%, 2.576 for 99%. Statistical software and tables provide precise values for any desired confidence level.

Should I use Z-score or t-distribution?

Use the Z-distribution when you know the population standard deviation or have a large sample (n ≥ 30). Use the t-distribution for smaller samples where you estimate standard deviation from the sample. The t-distribution is more conservative, yielding slightly wider intervals—this accounts for extra uncertainty from small-sample estimation. As sample size grows, t approaches Z.

How does sample size affect the confidence interval?

The standard error contains √n in the denominator, so it decreases as sample size increases. Doubling the sample size reduces the standard error by a factor of √2 ≈ 1.41, thus narrowing the interval proportionally. To halve the margin of error, you need to quadruple the sample size. This non-linear relationship means very large samples are required for very tight intervals, making cost–benefit analysis essential in study design.

Can I use a confidence interval to test hypotheses?

Yes. If a hypothesized parameter value falls outside your confidence interval, you would reject it at the corresponding significance level (e.g., reject at α = 0.05 if it's outside the 95% CI). Conversely, if it falls inside, you lack evidence to reject. This equivalence between confidence intervals and hypothesis tests makes CIs more informative: they show not just whether an effect exists, but the range of plausible values.

What if my data is not normally distributed?

Small deviations from normality are often acceptable, especially with larger samples, due to the Central Limit Theorem. For visibly non-normal data, consider a non-parametric confidence interval or transformation. Bootstrap methods, which resample your data repeatedly, provide robust alternatives when normality assumptions are violated. Many statistical packages now offer bootstrap CIs without additional assumptions.

How do I report a confidence interval in research?

Standard format: state the estimate, interval bounds, and confidence level. Example: "The mean weight loss was 4.5 kg (95% CI: 3.2–5.8 kg)." Always include the confidence level; readers cannot interpret the interval without it. When publishing, specify whether you used Z or t, and mention any transformations or adjustments. Wider intervals are not a weakness—they honestly reflect greater uncertainty.

More statistics calculators (see all)

Dice Probability Calculator Venn Diagram Calculator Frequency Distribution Calculator Combinations with Repetition Calculator ANOVA Calculator P-hat Calculator 90% Confidence Interval Calculator Sample Size Calculator

What data do you have?

Sample mean (x̅)

Standard deviation (s)

Sample size (n)

Standard error

I know the desired value of:

Confidence level

Z-score (Z)

Right tailed p value

Confidence interval

Lower bound

—

Upper bound

—

Margin of error (E)

—

Alpha

—