statistics

90% Confidence Interval Calculator

Q: What does a 90% confidence interval tell me?

It provides a range of values likely to contain the true population parameter based on your sample. The 90% figure means that if you repeated the sampling process infinitely, approximately 90% of the intervals you calculate would include the actual population value. It quantifies the uncertainty inherent in using sample data to estimate population characteristics. However, for any single computed interval, the true parameter either lies within it or doesn't.

Q: Why is the z-score for 90% confidence exactly 1.645?

The z-score is derived from the standard normal distribution. A z-score of 1.645 means that 90% of the area under the normal curve falls between −1.645 and +1.645 standard deviations from the mean. This leaves 10% split equally in both tails: 5% on each side. Statistical tables and software calculate this precisely. Different confidence levels require different z-scores: 95% uses 1.96, and 99% uses 2.576.

Q: When should I use standard error instead of standard deviation?

Use standard error when you're already working with the variability of the sample mean itself, or when it's been calculated from your data collection process. Use standard deviation when you have raw measurements and need to first account for how spread out your individual observations are. The calculator allows toggling between these because they're related: SE = σ / √n. If you have only SE available, you don't need to convert it—enter it directly to skip a calculation step.

Q: Can I adjust the confidence level away from 90%?

Yes, the calculator allows you to input any confidence level you prefer. Changing the confidence level automatically updates the corresponding z-score. Higher confidence levels (95%, 99%) produce wider intervals because you're demanding greater certainty. Lower confidence levels (85%, 80%) produce narrower intervals but with higher risk of missing the true parameter. Your choice should reflect your field's standards and risk tolerance.

Q: What's the difference between margin of error and standard error?

Standard error measures the variability of the sample mean itself—how much sample means would differ if you repeated sampling many times. Margin of error is the range you add and subtract from the sample mean to build the confidence interval, calculated by multiplying standard error by the z-score. A wider interval occurs with larger standard error, larger z-score, or smaller sample size. Margin of error directly determines interval width.

Q: How does sample size affect the confidence interval width?

Larger sample sizes produce narrower confidence intervals. Since standard error decreases as sample size increases (SE = σ / √n), the margin of error shrinks proportionally. For example, increasing sample size from 100 to 400 reduces the margin of error by half. This is why researchers prioritize collecting larger samples when they want precise estimates, though larger samples require more resources and time.

A 90% confidence interval defines the range where the true population parameter is likely to fall with 90% certainty based on your sample data. Researchers, quality control specialists, and data analysts use this interval to quantify uncertainty around point estimates. The calculator automates the process of computing standard error, margin of error, and upper/lower bounds from your sample statistics.

Last updated: May 9, 2026

Creators Anna Szczepanek, PhD Statistics

Reviewers Wojciech Sas and Davide Borchia

818 people find this calculator helpful

Understanding the 90% Confidence Interval

A 90% confidence interval represents a range of values calculated from sample data. If you were to repeat your sampling process many times and compute the interval each time, approximately 90% of those intervals would contain the true population parameter.

Consider a manufacturing scenario: a facility produces steel components with a target weight of 500 grams. After weighing 150 randomly selected components, the mean weight is 501.2 grams with a standard deviation of 8.5 grams. The 90% confidence interval might be 499.1 to 503.3 grams. This means the facility can be 90% confident that the actual average weight of all components falls within this range.

The confidence level you choose reflects your tolerance for uncertainty. A 90% confidence interval is narrower than a 95% interval (meaning more precision) but involves slightly higher risk of missing the true parameter. Industries requiring stricter quality assurance often prefer 95% or 99% levels.

The Mathematics of Confidence Intervals

The confidence interval calculation involves three key steps: determining standard error from your sample spread, computing the margin of error using the z-score, and then building the interval around your sample mean.

SE = σ / √n

ME = SE × Z(0.90)

Lower Bound = x̅ − ME

Upper Bound = x̅ + ME

σ (sigma) — Standard deviation of your sample, measuring how spread out your observations are from the mean.
n — Sample size; the total number of observations or measurements you collected.
SE — Standard error; the standard deviation divided by the square root of sample size. It estimates the variability of the sample mean.
Z(0.90) — Z-score for 90% confidence level, which equals 1.645. This value comes from the standard normal distribution.
ME — Margin of error; the distance from the sample mean to each boundary of the confidence interval.
x̅ (x-bar) — Sample mean; the average of all your observations.

Step-by-Step Calculation Process

Step 1: Gather your sample statistics. You need the sample mean (x̅), standard deviation (σ), and sample size (n). If you only have standard error available, you can skip the first calculation.

Step 2: Calculate standard error. Divide the standard deviation by the square root of your sample size. For example, with σ = 10 and n = 100, SE = 10 / √100 = 10 / 10 = 1.0.

Step 3: Compute margin of error. Multiply the standard error by 1.645 (the z-score for 90% confidence). Using the previous example: ME = 1.0 × 1.645 = 1.645.

Step 4: Find the bounds. Subtract the margin of error from your mean for the lower bound, and add it for the upper bound. If your sample mean is 50, the interval becomes 50 − 1.645 = 48.355 to 50 + 1.645 = 51.645.

Common Pitfalls and Practical Considerations

Avoid these frequent mistakes when working with confidence intervals.

Confusing confidence level with probability of a specific interval — The 90% confidence level describes the long-run behavior of the procedure, not the probability that a particular calculated interval contains the parameter. Once computed, your interval either contains the true value or it doesn't—the probability is either 0 or 1 for that specific interval.
Using sample standard deviation when population standard deviation is known — If your data comes from a population with known standard deviation, use that value instead of calculating sample standard deviation. Using sample standard deviation when you have the true population value introduces unnecessary uncertainty.
Ignoring sample size constraints and normality assumptions — Confidence interval formulas assume your sample is reasonably large or that data is normally distributed. With very small samples (n < 30) and non-normal data, consider using t-distribution methods instead of z-scores.
Misinterpreting overlap between confidence intervals — Two overlapping confidence intervals do not imply that two groups are statistically similar. Proper hypothesis testing is needed to make such comparisons, not visual inspection of interval overlap.

Frequently Asked Questions

What does a 90% confidence interval tell me?

It provides a range of values likely to contain the true population parameter based on your sample. The 90% figure means that if you repeated the sampling process infinitely, approximately 90% of the intervals you calculate would include the actual population value. It quantifies the uncertainty inherent in using sample data to estimate population characteristics. However, for any single computed interval, the true parameter either lies within it or doesn't.

Why is the z-score for 90% confidence exactly 1.645?

The z-score is derived from the standard normal distribution. A z-score of 1.645 means that 90% of the area under the normal curve falls between −1.645 and +1.645 standard deviations from the mean. This leaves 10% split equally in both tails: 5% on each side. Statistical tables and software calculate this precisely. Different confidence levels require different z-scores: 95% uses 1.96, and 99% uses 2.576.

When should I use standard error instead of standard deviation?

Use standard error when you're already working with the variability of the sample mean itself, or when it's been calculated from your data collection process. Use standard deviation when you have raw measurements and need to first account for how spread out your individual observations are. The calculator allows toggling between these because they're related: SE = σ / √n. If you have only SE available, you don't need to convert it—enter it directly to skip a calculation step.

Can I adjust the confidence level away from 90%?

Yes, the calculator allows you to input any confidence level you prefer. Changing the confidence level automatically updates the corresponding z-score. Higher confidence levels (95%, 99%) produce wider intervals because you're demanding greater certainty. Lower confidence levels (85%, 80%) produce narrower intervals but with higher risk of missing the true parameter. Your choice should reflect your field's standards and risk tolerance.

What's the difference between margin of error and standard error?

Standard error measures the variability of the sample mean itself—how much sample means would differ if you repeated sampling many times. Margin of error is the range you add and subtract from the sample mean to build the confidence interval, calculated by multiplying standard error by the z-score. A wider interval occurs with larger standard error, larger z-score, or smaller sample size. Margin of error directly determines interval width.

How does sample size affect the confidence interval width?