Using the \(n\geq30\) Rule Without Overstating It
In “Checking the Normal/Large Sample Condition,” we used two routes for supporting an approximately Normal model for the sampling distribution of \(\bar{x}\): the population is approximately Normal, or the sample size is at least 30. This tutorial focuses on how to apply the second route carefully. In particular, we will compare samples of 40 and 25 observations and clarify what the sample-size rule does—and does not—tell us.
The rule is useful because the Central Limit Theorem explains why the distribution of sample means tends to become more nearly Normal as sample size increases, under its assumptions. For AP Statistics mean inference, \(n\geq30\) is the practical large-sample criterion. When it applies, you do not need to establish that the individual observations are Normally distributed in order to use this route.
The cutoff is a practical decision rule, not a sharp mathematical cliff. The difference between \(n=29\) and \(n=30\) is only one observation; the rule does not mean that the sampling distribution suddenly changes shape at exactly 30. It gives a consistent AP criterion for deciding when sample size alone supports the large-sample route. The population’s shape still matters to how convincing an approximation is, especially if the data show unusually strong skewness or extreme outliers.
What the Rule Supports—and What It Does Not
When \(n=40\), the large-sample route is met because 40 is at least 30. This supports using the t procedure’s approximate Normal model for \(\bar{x}\), even if the distribution of individual observations is not Normal. It does not transform the measurements into Normal data, remove skewness from the population, or guarantee that every feature of a sample is harmless.
The rule also addresses only the shape condition. It does not show that a sample is random or representative, and it does not establish independence. As covered in “Checking the Random Condition” and “Checking the 10% Condition for Independence,” those checks have separate purposes. A sample can meet the large-sample rule and still have a problem with how it was selected or with dependence between observations.
For a sample of 25, the large-sample route is not met because \(25<30\). That does not automatically rule out mean inference. Instead, you need evidence that the population is approximately Normal. When the population shape is not provided, a graph of the sample data can offer evidence: a roughly symmetric, unimodal pattern without clear outliers is more reassuring than strong skewness or unusual extreme values. A sample graph is evidence, not proof of the population’s shape.
A Practical Decision Process
For a condition check, state the sample size and name the route you are using. If \(n\geq30\), explicitly say that the large-sample rule is met and connect it to the sampling distribution of \(\bar{x}\). If \(n<30\), do not say that the condition automatically fails; explain what the available information says about population shape. If shape information is absent, say that the condition cannot be verified from the information given.
Count the observations used to calculate the sample mean.
If \(n\geq30\), state that the AP large-sample route is met. If \(n<30\), move to the population-shape route.
For a sample smaller than 30, refer to information about the population or describe relevant features of a sample graph.
Say what the evidence supports about the model for \(\bar{x}\). Check randomness and independence separately.
Worked Examples
Worked Example: A Random Sample of 40 Water Filters
A technician randomly selects 40 water filters from a shipment of 600 and records the volume of water each filter processes before replacement. The distribution of individual filter lifetimes is described as right-skewed. Is the Normal/Large Sample condition supported by the AP rule?
State. Let \(\mu\) be the mean volume processed before replacement for filters in this shipment. We are checking whether the Normal/Large Sample condition supports a t procedure for \(\mu\).
Plan. We will first compare the sample size with 30. Because the filters are sampled without replacement, we will also check the 10% condition for independence. The random selection supports the random condition. These checks address different requirements.
Do. The sample size is \(n=40\), and \(40\geq30\), so the large-sample route is met. The right-skewed description concerns individual filter lifetimes; it does not mean the rule has failed. The large-sample route supports an approximately Normal sampling distribution for \(\bar{x}\), not a Normal distribution of individual lifetimes.
For the 10% condition, compare the sample with the shipment size:
Since \(6.67\%\) is less than 10%, the 10% condition is met. The technician’s random selection supports the random condition.
Conclude. The Normal/Large Sample condition is supported by \(n=40\), which is at least 30. The right-skewed individual lifetimes do not negate this AP large-sample route. The random selection and the 10% check also support the other stated conditions for a one-sample t procedure.
Worked Example: A Random Sample of 25 Trail Distances
A parks department takes a simple random sample of 25 marked walking trails from 400 trails in a region and records each trail’s length. A dotplot of the sample is roughly symmetric, has one main cluster, and shows no clear outliers. Does the sample meet the large-sample rule? Is the Normal/Large Sample condition supported?
State. Let \(\mu\) represent the mean length of trails in the region. We are checking the Normal/Large Sample condition for inference about \(\mu\).
Plan. The sample has fewer than 30 observations, so it does not meet the large-sample route. We will instead use the described shape as evidence about whether an approximately Normal population model is reasonable. Since the sample is selected without replacement, we will check the 10% condition too.
Do. Here \(n=25\), and \(25<30\). The sample therefore does not qualify by the large-sample rule. However, the dotplot is roughly symmetric, has one main cluster, and shows no clear outliers. These features are consistent with an approximately Normal population, so they provide evidence for the other route. They do not prove that the entire population is Normal.
The sample fraction is
Because \(6.25\%\) is less than 10%, the 10% condition is met. The simple random sample supports the random condition.
Conclude. The sample does not meet the \(n\geq30\) route, but the roughly symmetric, unimodal dotplot with no clear outliers provides evidence that the population distribution is approximately Normal. Thus the Normal/Large Sample condition is reasonably supported through the population-shape route, with randomness and independence supported separately.
Worked Example: \(n=40\) Does Not Check Every Condition
A school records the commute times of 40 students who volunteered after seeing an announcement on a student message board. The school wants to use a t procedure to estimate the mean commute time for all students. Does the sample size alone justify the inference?
State. Let \(\mu\) be the mean commute time for all students at the school. We need to distinguish the Normal/Large Sample condition from the conditions about how the data were collected.
Plan. Since \(n=40\), check whether the large-sample route is met. Then consider whether the volunteer method supports a random sample from the school population. A large sample size cannot repair a selection method that may systematically leave out some students.
Do. The sample size is \(40\), which is at least 30, so the AP large-sample route supports an approximately Normal model for the sampling distribution of \(\bar{x}\). But students volunteered in response to an announcement; they were not described as randomly selected. Students with especially long or short commutes may have been more likely to respond. Thus the random condition is not established. Because the method is not described as a random sample without replacement, a 10% check does not solve this selection concern.
Conclude. The Normal/Large Sample condition is supported by \(n=40\), but sample size alone does not justify generalizing the estimate to all students. The volunteer method does not establish the random condition, so the proposed population inference is not adequately supported.
Worked Example: A Small Sample With No Shape Information
A researcher measures the amount of time 25 randomly selected customers spend setting up a new device. No population description, graph, or list of sample values is provided. Can the Normal/Large Sample condition be verified?
The sample size is \(n=25\), so it is below 30 and does not meet the large-sample route. The random selection supports the random condition, but it gives no information about population shape. Since no graph or other shape information is provided, there is not enough evidence to decide whether the population is approximately Normal.
Conclusion. The Normal/Large Sample condition cannot be verified from the information given. A careful response should request the sample data or other evidence about the population’s shape. It should not claim the condition holds merely because the sample is random, and it should not say it fails merely because \(n=25\).
Common Mistakes and AP Exam Tips
- Changing “at least 30” into “more than 30.” A sample of exactly 30 meets the rule. The correct comparison is \(n\geq30\), not \(n>30\).
- Treating \(n<30\) as automatic failure. A smaller sample can still support the condition if the population is approximately Normal. Give the shape evidence rather than stopping at the comparison.
- Using \(n\geq30\) to claim that the data are Normal. The rule concerns an approximately Normal sampling distribution of \(\bar{x}\), not the shape of individual measurements.
- Thinking a large sample guarantees a trustworthy inference. The rule does not establish random selection, independence, or lack of bias. Name and check those conditions separately.
- Overstating what a graph shows. A sample graph can provide evidence about population shape, especially when \(n<30\), but it cannot prove the entire population is Normal. Describe the visible pattern and use cautious wording.
- Ignoring striking data features. The AP rule gives a clear large-sample route, but it is not a promise that any extreme pattern is harmless. If the available data show severe skewness or a very unusual outlier, acknowledge it and consider whether the approximation warrants closer scrutiny.
A full-credit statement is specific about both the evidence and its target. For example: “Because \(n=40\geq30\), the large-sample route is met, supporting an approximately Normal model for the sampling distribution of \(\bar{x}\). This does not establish the random condition.” For a sample of 25, explain that the large-sample route is not met and identify the shape evidence—or the missing evidence—relevant to the other route.
Check Your Understanding
For each situation, identify whether the large-sample route is met and explain what else, if anything, can be concluded about the Normal/Large Sample condition.
- A random sample of 30 greenhouse plants is used to measure weekly growth. Does it meet the \(n\geq30\) rule? State what distribution the rule concerns.
- A sample of 25 randomly selected bus routes has a roughly symmetric dotplot with one main cluster and no clear outliers. Which route may support the condition, and what should you avoid claiming?
- A sample of 40 customers volunteered to report their delivery times. What does the sample size support, and what condition does it fail to establish by itself?
- A random sample of 25 measurements is described, but there is no information about population shape and no sample graph. Is the condition supported, not supported, or impossible to verify from the information given?
- Explain why \(n=30\) is a practical cutoff rather than proof that every sample-mean model is perfectly Normal.