How Sample Size Changes Standard Error and Confidence-Interval Width

The uncertainty of a mean depends on both the spread of individual observations and the number of independent observations. A confidence interval adds a coverage requirement to that uncertainty calculation.

Start with a fully labeled example

Suppose an independent random sample of 25 measurements has mean 50 and sample standard deviation 10. Assume the population is reasonably modeled as normal and its standard deviation is unknown.

$$SE=\frac{s}{\sqrt{n}}=\frac{10}{\sqrt{25}}=2.$$

For a two-sided 95% t interval with 24 degrees of freedom, $t_{0.975,24}\approx2.064$. The interval is approximately

$$50\pm2.064(2)=50\pm4.128=(45.87,54.13).$$

The SD is 10, the SE is 2, and the interval half-width is about 4.13. These quantities answer different questions, despite coming from the same sample.

Increasing n reduces uncertainty, not necessarily data spread

Holding the underlying standard deviation fixed, four times as many independent observations halves the standard error. It does not halve the standard deviation of individual measurements. A t-interval also changes its critical multiplier as degrees of freedom change, so its width is not determined by the square-root factor alone.

Interpret confidence as a property of the procedure

In repeated sampling under the model, about 95% of intervals constructed this way cover the fixed population mean. This does not mean that 95% of individual observations lie within the interval.

Additional observations help only when the design supplies additional information. Repeated measurements of the same experimental unit should not automatically be counted as independent replicates.