← Lessons

quiz vs the machine

Gold1400

Machine Learning

The Confidence Intervals

Reporting a range of plausible values instead of a single estimate.

5 min read · core · beat Gold to climb

The Confidence Intervals

A single estimate, like a sample mean, hides its uncertainty. A confidence interval reports a range of plausible values along with how confident the method is.

What it means

A 95 percent confidence interval is built by a procedure that, across many repeated samples, would capture the true parameter about 95 percent of the time. The confidence is a property of the method, not of any one interval. For a specific computed interval the true value is either inside or not.

How width is set

An interval centers on the estimate and extends by a margin of error, which is a critical value times the standard error. Several factors shrink the width:

  • A larger sample lowers the standard error, narrowing the interval.
  • Less variability in the data narrows it.
  • A lower confidence level narrows it but captures the truth less often.

Link to hypothesis tests

A 95 percent interval and a test at alpha 0.05 agree: if a hypothesized value falls outside the interval, the corresponding test rejects it. The interval also shows the effect size, which a bare p value hides.

Key idea

A confidence interval is a range built so the procedure captures the true parameter a set fraction of the time, narrowing with larger samples and showing effect size a p value hides.

Check yourself

Answer to earn rating on the learn ladder.

1. What does a 95 percent confidence level describe?

2. What happens to interval width as sample size grows?

3. How does a confidence interval relate to a hypothesis test?