How Sample Size Affects Variability of x-bar

Comparing Sample Sizes

In “Standard Deviation of the Sample Mean,” we used \(\sigma_{\bar{x}}=\sigma/\sqrt{n}\) to describe the spread of sample means across repeated samples. Now we can use that relationship to compare sample sizes directly. The key pattern is that multiplying the sample size by 4 divides the standard deviation of \(\bar{x}\) by 2.

Consider three sampling distributions from the same population, with sample sizes \(n=9\), \(36\), and \(144\). The population mean \(\mu\) and population standard deviation \(\sigma\) are fixed. As long as the observations meet the independence assumptions for the formula, all three sampling distributions have the same center, \(\mu\). Their spreads differ: the larger the sample, the smaller the standard deviation of the sample mean.

This is a comparison of theoretical sampling distributions, not a claim that every larger sample will produce a sample mean closer to \(\mu\). The standard deviation describes typical spread over repeated samples. A particular sample mean can still be unusually far from the population mean.

How the Change in Sample Size Changes Spread

The formula from the earlier tutorial gives a quick way to compare two sample sizes without starting over each time. For a fixed population standard deviation, divide the standard deviations for the two sample sizes. The population standard deviation cancels, leaving a ratio determined by the sample sizes alone.

If the new sample size is four times the old sample size, then \(n_{\text{new}}=4n_{\text{old}}\). Substituting this into the ratio gives \(\sqrt{1/4}=1/2\). Thus, the new standard deviation is half the old one. This ratio approach is useful even when the sample sizes are not the specific values 9, 36, and 144.

Standard deviation and variance respond differently to sample size. Since the variance of \(\bar{x}\) is \(\sigma^2/n\), quadrupling \(n\) makes the variance one-fourth as large. Taking the square root of the variance gives the standard deviation, so the standard deviation is halved. Keep the measurement units in mind: standard deviation is measured in the original units, while variance is measured in squared units.

Compare by a Multiplier, Not Just by Recalculating

A ratio can make the pattern clear when the sample sizes are different from those in the main comparison. For example, if a sample size increases from 25 to 100, the new size is 4 times the old size. The standard deviation must therefore be multiplied by \(1/2\). If the sample size increases from 25 to 50 instead, it is multiplied by 2, so the standard deviation is multiplied by \(1/\sqrt{2}\), not by \(1/2\).

More generally, if the sample size changes by a factor of \(k\), the standard deviation changes by a factor of \(1/\sqrt{k}\). This lets you reason about the effect before doing a full calculation. It also helps catch errors: if the sample size becomes four times as large, a proposed standard deviation one-fourth as large is too small under this formula.

Sample Size Needed for a Target Spread

The same relationship can be used in reverse. If you want the standard deviation of the sample mean to be a certain fraction of its current value, determine how much the sample size must change. For instance, making the standard deviation one-half as large requires quadrupling the sample size. Making it one-fourth as large requires multiplying the sample size by 16, because \(\sqrt{16}=4\).

This is sometimes surprising: reducing variability by a large amount can require a much larger increase in sample size. The relationship is not linear. Doubling the sample size does not halve the standard deviation; it multiplies it by \(1/\sqrt{2}\), which is about 0.7071.

What Changes and What Stays the Same

When comparing sampling distributions from the same population, separate their centers from their spreads. The mean of the sampling distribution remains \(\mu\) for each sample size under the conditions described in the earlier tutorial “Mean of the Sampling Distribution of x-bar.” Increasing \(n\) changes the standard deviation \(\sigma_{\bar{x}}\), not the population mean or the center of the sampling distribution.

The formula for standard deviation also does not determine the shape of the sampling distribution. Whether a sampling distribution is approximately normal depends on information about the population distribution and the sample size. Do not infer a normal shape from a small standard deviation alone.

For samples drawn without replacement from a finite population, refer to the 10% condition and finite-population adjustment discussed in “Standard Deviation of the Sample Mean.” When using the usual AP approximation \(\sigma/\sqrt{n}\), the simple halving pattern applies when the samples meet the independence assumptions or the 10% condition supports treating the observations as approximately independent. The exact finite-population adjustment can make the ratios slightly different from the simple pattern.

Common Mistakes and AP Exam Tip

• Halving the sample size instead of taking its square root: For \(n=36\), divide \(\sigma\) by \(\sqrt{36}=6\), not by 36.
• Claiming that quadrupling \(n\) quarters the standard deviation: It quarters the variance. The standard deviation is the square root of variance, so it is halved.
• Changing the center when sample size changes: For samples from the same population under the stated sampling conditions, the mean of the sampling distribution remains \(\mu\). It is the spread that changes.
• Assuming a particular larger-sample mean must be closer to \(\mu\): The rule compares variability over repeated samples. It does not guarantee the result for one observed sample.
• Ignoring units or the context: If individual readings are in units, then \(\sigma_{\bar{x}}\) is also in units. Variance, by contrast, is in squared units.
• Claiming the distribution is normal based only on its standard deviation: The formula describes spread, not shape. Use information about shape only when it is supported by the population and sampling conditions.

For a clear AP response, identify what is being compared, show either the standard-deviation calculations or the ratio of sample sizes, and interpret the result in context with units. A complete explanation might say, “Because the sample size increases from 9 to 36, it is multiplied by 4, so the standard deviation of the sample mean is divided by 2.” This explains the connection instead of simply listing two numerical answers.