A confidence interval for a mean is a range of plausible values for an unknown population mean, based on data from a sample. It matters because a sample mean is only an estimate, and different random samples would give different results. The interval adds a margin of error around the sample mean to show how much uncertainty is reasonable.
A 95% confidence interval is commonly used because it balances precision with reliability.
Understanding Statistics: Confidence Interval for a Mean
The calculation is based on what would happen if the same sampling process were repeated many times. Each sample would produce a slightly different average. Those averages tend to cluster around the population mean, with a spread that depends on how varied the data are and how many observations were collected.
This spread is called the standard error. In real problems, the population spread is usually unknown. For that reason, a t distribution is used instead of a normal distribution.
The t distribution allows extra uncertainty, especially when the sample is small. Its shape depends on the degrees of freedom, which grow as more data are available.
To build the interval, start with the sample average as the center. Then estimate the likely sampling error from the variation within the sample. A critical t value scales that error to match the chosen confidence level.
Higher confidence requires a larger critical value, so the interval becomes wider. This is a tradeoff. More certainty from the method means less precision in the reported range.
Sample size has a strong effect too. When the sample size is multiplied by four, the standard error is roughly cut in half, assuming the data have similar variation. Collecting more data can therefore improve precision, but only if the data are collected well.
The meaning of 95 percent confidence needs careful wording. It does not mean there is a 95 percent chance that the already fixed population mean lies inside one particular calculated interval. The population mean has one true value.
The interval either contains it or it does not. Instead, 95 percent describes the long-run performance of the method. If researchers repeatedly took random samples and built intervals in the same way, about 95 out of 100 intervals would contain the true mean.
A confidence interval does not describe where 95 percent of individual data values fall. Individual values are usually much more spread out than sample averages.
A t interval works best when the sample represents the population of interest and each observation is reasonably independent. A survey of students from one class may not represent every student in a school district. Measurements from the same person taken many times may not be independent either.
Very strong skewness or extreme outliers can make a small-sample interval unreliable because they distort the average and standard deviation. Graphing the data first can reveal these issues. For larger samples, the method is often more robust, though biased sampling remains a serious problem regardless of sample size.
Confidence intervals appear in product testing, medical studies, school surveys, and environmental measurements. A factory may estimate the mean lifetime of a battery. A health study may estimate the mean change in blood pressure.
In each case, the interval helps show whether a reported average is precise enough for a decision. Students should separate precision from accuracy.
A narrow interval can still miss the truth if the sampling method is biased. Check the population, sampling method, sample size, data shape, confidence level, and units before interpreting any interval.
Key Facts
- Point estimate for the population mean: x̄
- Standard error for a mean: SE = s / √n
- t-interval for a mean: x̄ ± t* (s / √n)
- Margin of error: ME = t* (s / √n)
- Degrees of freedom for a one-sample t-interval: df = n - 1
- A larger sample size usually makes the confidence interval narrower because SE decreases as n increases.
Vocabulary
- Confidence interval
- A range of values calculated from sample data that is used to estimate an unknown population parameter.
- Sample mean
- The average value x̄ computed from the observations in a sample.
- Margin of error
- The distance from the sample mean to either endpoint of the confidence interval.
- Critical t-value
- A multiplier from the t-distribution that sets how many standard errors are needed for a chosen confidence level.
- Standard error
- The estimated standard deviation of the sample mean, calculated as s / √n for a one-sample mean interval.
Common Mistakes to Avoid
- Saying there is a 95% chance that the population mean is inside this specific interval. The population mean is fixed, and the 95% refers to the long-run success rate of the method.
- Using z* when the population standard deviation is unknown. For most sample mean intervals with unknown σ, use a t* value with df = n - 1.
- Forgetting to divide the sample standard deviation by √n. The margin of error uses the standard error s / √n, not the raw sample standard deviation s.
- Interpreting a wider interval as a mistake. A wider interval can be correct when the data are more variable, the sample size is smaller, or the confidence level is higher.
Practice Questions
- 1 A sample of n = 25 students has mean study time x̄ = 6.8 hours and sample standard deviation s = 2.0 hours. Using t* = 2.064 for a 95% confidence interval, compute the interval for the population mean study time.
- 2 A sample of n = 40 batteries has mean lifetime x̄ = 18.6 hours and sample standard deviation s = 3.2 hours. Using t* = 2.023 for a 95% confidence interval, find the margin of error and the confidence interval.
- 3 Two studies estimate the same population mean. Study A uses n = 20 and Study B uses n = 200, with similar sample standard deviations and the same confidence level. Explain which interval should usually be narrower and why.