Confidence intervals for proportions
A confidence interval turns a single estimate into a range of plausible values, with a stated level of confidence in the method.
Estimate plus or minus margin
Every confidence interval has the form estimate \(\pm\) margin of error. The margin is a critical value times the standard error. For a proportion, the interval is \(\hat{p} \pm z^*\sqrt{\frac{\hat{p}(1-\hat{p})}{n}}\), where \(z^* = 1.96\) for 95% confidence.
\hat{p} \pm z^*\sqrt{\frac{\hat{p}(1-\hat{p})}{n}}Worked example
In a random sample of 400 voters, 220 (\(\hat{p} = 0.55\)) support a measure. The standard error is \(\sqrt{\frac{0.55 \cdot 0.45}{400}} \approx 0.0249\). The 95% interval is \(0.55 \pm 1.96(0.0249) = 0.55 \pm 0.049\), or about 50.1% to 59.9%. We are fairly confident a majority supports it, but only just.
What '95% confidence' means
If we took many random samples and built an interval from each, about 95% of those intervals would capture the true proportion. It describes the reliability of the method. It does not mean there is a 95% probability that this particular interval contains \(p\); the parameter is fixed and either inside or not.
Confidence level and width
Higher confidence needs a wider interval: \(z^* = 1.645\) for 90%, \(1.96\) for 95%, and \(2.576\) for 99%. There is a trade-off between being confident and being precise. To narrow an interval without losing confidence, increase the sample size.
Choosing the sample size
To achieve a margin of error \(m\) at 95% confidence, use \(n = \left(\frac{z^*}{m}\right)^2 \hat{p}(1-\hat{p})\). With no prior estimate, use \(\hat{p} = 0.5\), which gives the largest (safest) \(n\). For a 3-point margin, \(n = \left(\frac{1.96}{0.03}\right)^2(0.25) \approx 1068\), which is why many national polls survey about 1,000 people.
n = \left(\frac{z^*}{m}\right)^2 \hat{p}(1-\hat{p})Conditions
The interval assumes a random sample, independent observations (sample under 10% of the population), and at least 10 successes and 10 failures in the sample. With small counts, use adjusted methods such as the plus-four interval, which adds two successes and two failures before computing.
Margin of error covers only sampling error
The margin of error measures random sampling variability only. It does not cover bias from undercoverage, nonresponse, or misleading questions. A poll can have a tiny margin of error and still be badly wrong if its sample is biased.
Common mistakes
Saying 'there is a 95% probability \(p\) is in this interval'. Thinking a larger population needs a larger sample (the population size barely matters if it is large). Believing the margin of error covers bias. Using \(\hat{p}\) from one group to plan a different study without caution.
Key takeaways
A confidence interval is estimate \(\pm\) critical value \(\times\) standard error. 95% confidence describes the long-run success rate of the method. Higher confidence widens the interval; larger samples narrow it. Plan sample size from the desired margin, and remember that margins ignore bias.
Flashcard
Estimate plus or minus margin of error
Check your understanding
Which change makes a confidence interval narrower?
