Short answer
A population parameter describes the entire group you want to study; a sample statistic describes the subset you actually examine. A statistic can estimate a parameter, but it does not become the population quantity merely because you use it that way. The boundary is which group the number summarizes. 1 2
On this page
At a glance
Scroll to compare all columns →
| Question | Population parameter | Sample statistic |
|---|---|---|
| What does it describe? | Entire population | Selected sample |
| Mean notation in Penn State’s treatment | μ | x̄ |
| Proportion notation | p | p̂ |
| Role in estimation | Quantity being estimated | Summary used to estimate it |
These paired quantities describe the same kind of characteristic at different scopes. 1
What each thing is
The population is the collection of people, things, or objects under study. Its parameter summarizes a characteristic across that collection. A sample is a subset studied to learn about the population; its statistic summarizes the collected sample data. Neither term refers to an individual observation, such as one student’s GPA. 2
Key differences
The key difference is coverage, not the arithmetic operation. Both quantities can be averages or proportions. In the sources’ estimation framework, the sample summary is calculated from available observations, while the corresponding population value is the target of inference and may be unknown. Calculating the statistic does not by itself establish the parameter’s value. 1
How to tell them apart
First identify the intended population. Then ask whether the number summarizes that entire group or only the selected sample. That is more useful than asking whether the number is a percentage or an average. The rule has a limit: a bare number without its group and study context cannot reliably reveal which label applies. 1 2
Where they overlap
Parameters and statistics are linked by matching characteristics: a sample mean can estimate a population mean, and a sample proportion can estimate a population proportion. They are therefore complementary parts of inference, not competing kinds of measurement. How well the sample represents the population matters to the accuracy of the estimate. 1 2
Edge cases
The label depends on the study’s boundary. In OpenStax’s example, average points in one math class is a statistic when all math classes form the population. Hypothetically, if that one class were instead the entire population of interest, its whole-class average would be a parameter. The same calculation can have a different role under a different population definition. 2
Why the distinction exists
Examining an entire population can take substantial time and money, so sampling provides a practical way to learn about it. Keeping the terms separate preserves the distinction between the observed sample summary and the broader quantity being investigated. Confidence intervals and hypothesis tests are methods for learning about that population quantity. 2 1
Common misconceptions
A parameter is not defined as a number that can never be known; it is defined by its population scope. Likewise, calling a statistic an estimate does not guarantee that it closely matches the parameter. Nor are the individual measurements themselves the statistic: the statistic summarizes those sample data. 1 2
Examples
In Penn State’s smoking example, 43% of 987 surveyed students reported regular smoking. That percentage is the statistic; the proportion among all students in the specified campus population is the parameter. In its voter example, approval among 1,000 randomly sampled likely voters is the statistic used to estimate approval among all likely American voters, the parameter. 1