∿ExploreSims
ExploreSims / Statistics / Inference / Confidence Intervals & Margin of Error
Premium experiment • statistical precision

STATISTICS • INFERENCE

Confidence Intervals & Margin of Error

Change sample size, variability and confidence level to see how standard error and confidence-interval width respond.

● INTERVAL ESTIMATIONMean ± margin
Sample mean
Standard error
Margin of error
Confidence interval

Reasoning inspector

live model state
Sample size—
Confidence—
Critical z*—
Standard error—
Margin of error—
Interval width—

Precision has a visible width

A confidence interval combines an estimate with its sampling uncertainty. Increase sample size, variability or confidence level and watch exactly which part of the interval changes.

LEARN WITH THE SIMULATION

Understand Confidence Intervals & Margin of Error by experimenting.

START HERE

A confidence interval combines a sample estimate with an explicit margin of sampling uncertainty. In this normal-approximation model, the interval width is controlled by the sample standard deviation, sample size and chosen confidence level, so each control reveals a different tradeoff between precision and coverage.

BUILD THE IDEA

What you should understand first

  • Compute standard error as the sample standard deviation divided by the square root of sample size.
  • See why larger samples narrow confidence intervals with diminishing returns.
  • Explain why choosing a higher confidence level widens the interval when the data are otherwise unchanged.
  • Distinguish the sample mean from the unknown population parameter the interval is intended to estimate.

TRY IT YOURSELF

Which is more effective for narrowing an interval: doubling sample size or cutting variability in half?

  1. Record the default confidence-interval width.
  2. Double the sample size while keeping the mean, standard deviation and confidence level unchanged.
  3. Reset, then cut the standard deviation in half while keeping the original sample size.
  4. Compare the two widths and connect the result to the square-root relationship in the standard error.
Read the visual:

Doubling sample size reduces standard error only by a factor of √2, while halving the standard deviation halves standard error directly. Sample-size gains are real but show diminishing returns because n appears under a square root.

WHY IT WORKS 01

Why sample size has a square root

Averages combine many independent observations, and independent variation accumulates in variance rather than directly in standard deviation. Dividing an average by n while variances add produces the familiar σ/√n scale. That square root explains why quadrupling sample size, not merely doubling it, is needed to halve standard error.

WHY IT WORKS 02

Precision and confidence trade off

Holding the data fixed, a wider interval covers more plausible parameter values and therefore achieves a higher nominal confidence level. Narrower intervals look more precise, but increasing confidence without collecting more information necessarily pushes the endpoints farther from the sample estimate.

DO NOT MIX THESE UP

Common misconception

A 95% confidence interval does not mean there is a 95% probability that a fixed population mean is inside this one computed interval under the usual frequentist interpretation. The confidence level describes the long-run coverage of the interval-building procedure under its assumptions.

PRACTICE CHALLENGE

Can you halve the margin of error using only the sample-size control?

  • Record the current margin of error and sample size.
  • Increase sample size until the margin is approximately half its original value.
  • Compare the new sample size with four times the original and explain why the relationship is not linear.
Reveal the reasoning

Because margin of error is proportional to 1/√n when the standard deviation and confidence level stay fixed, halving the margin requires √n to double. That means n must be multiplied by about four. The simulation makes this diminishing-return relationship directly visible.

MODEL BOUNDARIES

What this simulation assumes

Every simulation simplifies reality or a mathematical system. These are the assumptions and limits to keep in mind while interpreting the results.

Model assumptions

  • The interval uses a normal critical value with sample standard deviation supplied directly by the learner.
  • Observations are treated as independent and the sample mean is modeled with standard error s divided by the square root of n.
  • The confidence choices are limited to 90, 95 and 99 percent two-sided intervals.

Model limitations

  • Small-sample t critical values, finite-population corrections, clustering, weighting and non-normal robustness are omitted.
  • The simulation illustrates interval mechanics and does not guarantee that a real sampling design satisfies the assumptions.

CONNECT IT

Where this matters

  • Survey and experiment reporting
  • Estimating population means
  • Communicating sampling uncertainty
  • Planning sample sizes for desired precision

WORDS THAT MATTER

Key vocabulary

Standard error
The estimated standard deviation of the sampling distribution of the sample mean, s divided by the square root of n here.
Margin of error
The critical value multiplied by the standard error, giving the distance from the estimate to either interval endpoint.
Confidence level
The long-run coverage rate of the interval procedure under the model assumptions.
Critical value
The normal-distribution multiplier z* chosen to achieve the requested confidence level.