Skip to content
Tutorials Center
My learning

Topics

Lesson 33 of 37 · Sampling and Inference Basics

Confidence Intervals

Build 95% confidence intervals for means and proportions, interpret them correctly, understand margin of error, and see coverage in a simulation.

A sample mean is a single guess for the unknown population mean. A better statement gives a range of plausible values and says how reliable the method is. That range is a confidence interval.

It uses the standard error from sampling distributions and the z-values from z-scores.

The recipe

Text
estimate ± critical value × standard error

For a mean with a large sample (or a known σ), a 95% interval is:

Text
x̄ ± 1.96 × s/√n

The number 1.96 is the z-value that leaves 2.5% in each tail. The part after ± is the margin of error.

Worked example

A sample of n = 36 students has mean study time x̄ = 72.4 minutes and standard deviation s = 9.

Text
standard error = 9 / √36 = 1.5
margin of error = 1.96 × 1.5 = 2.94
95% interval = 72.4 ± 2.94 = (69.46, 75.34)
Python Confidence interval for a mean

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Higher confidence means a wider interval: you buy certainty with precision.

Small samples: the t distribution

When n is small and σ is estimated by s, the extra uncertainty makes intervals a bit wider. Use critical values from the t distribution with n - 1 degrees of freedom instead of 1.96. For example, the 95% critical value is about 2.262 with 9 degrees of freedom, and about 2.131 with 15. As n grows, t approaches 1.96. In Python, SciPy provides scipy.stats.t.ppf.

Python t critical values approach 1.96

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Confidence interval for a proportion

For a sample proportion p̂ from n people:

Text
p̂ ± z* × √( p̂(1 - p̂) / n )

Example: 540 of 900 voters support a proposal. p̂ = 0.6, SE = √(0.6 × 0.4 / 900) ≈ 0.01633, margin ≈ 0.0320. The 95% interval is about (0.568, 0.632).

How to interpret it

A 95% confidence interval does not say "there is a 95% probability that μ is in this interval". The true μ is a fixed number; it either is or is not inside. The 95% describes the method: if you repeated the sampling many times and built an interval each time, about 95% of those intervals would contain the true value.

Python Simulate the meaning of 95% confidence

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The result should be close to 0.95 (a little below, since using s with 1.96 slightly under-covers at n = 40).

What controls the margin of error?

The margin is z* × σ/√n, so:

  • larger n gives a narrower interval,
  • more variable data gives a wider one,
  • higher confidence gives a wider one.

To plan a survey, you can solve for n. To get a margin of 3 percentage points for a proportion at 95% confidence, using the worst case p = 0.5, you need n ≥ (1.96 × 0.5 / 0.03)² ≈ 1068.

Confidence intervals fix random sampling error only. They do nothing about bias, such as a poorly chosen sample.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. n = 100, x̄ = 40, s = 5. What is the 95% interval (use 1.96)?

    SE = 5/10 = 0.5, and 1.96 × 0.5 = 0.98.
  2. 2. If you change from 95% to 99% confidence with the same data, the interval gets...

    A larger critical value is needed to capture more of the sampling distribution.
  3. 3. The correct interpretation of a 95% confidence interval is...

    The confidence level is a property of the procedure over repeated sampling.
  4. 4. To halve the margin of error you need roughly...

    Margin ∝ 1/√n, so quadrupling n halves it.

Summary

  • A confidence interval is estimate ± critical value × standard error.
  • For means use z (large n) or t (small n); for proportions use √(p̂(1 - p̂)/n).
  • 95% describes the long-run success rate of the method, not one interval.
  • Larger n narrows the interval; higher confidence widens it.

Next: Hypothesis Testing Basics