Achievable logo
Achievable blue logo on white background

Descriptive vs. inferential statistics: the difference, with AP Statistics examples

Descriptive statistics summarize the data you collected; inferential statistics use a sample to draw conclusions about a population. See both in one example and where each fits in AP Stats.
Julia Tache's profile picture
Julia Tache
14 Nov 2026, 10 min read
Achievable blue logo on white background
Digital illustration of a scoop of marbles drawn from a large jar, beside a small bar chart and an estimate range marker, representing descriptive and inferential statistics
Achievable
Achievable blue logo on white background

Descriptive statistics summarize the data you actually collected: the mean, the spread, the shape of a graph. Inferential statistics use that data to draw a conclusion about a larger population you didn’t measure, and they attach a measure of uncertainty to it, such as a margin of error or a p-value. “The 120 students we surveyed sleep 6.7 hours on average” is descriptive; “we are 95% confident the mean for all 1,500 students at the school is between 6.48 and 6.92 hours” is inferential.

AP Statistics teaches both, in that order. This guide explains the difference with one worked example, shows which of the five units in the current course (effective fall 2026) is which, and lists the mistakes that cost points on the exam.

The difference at a glance

Aspect

Descriptive statistics

Inferential statistics

Question it answers

What does this data show?

What can this data tell us about everyone it came from?

Works with

The sample (or a whole population you measured)

A sample, to reach a population

Its numbers are called

Statistics: x̄, s, p̂, r

Estimates of parameters: μ, σ, p

Typical tools

Graphs, mean, median, standard deviation, IQR, correlation, a regression line

Confidence intervals, significance tests, p-values

Uncertainty

None stated: it describes what was observed

Always stated: a confidence level or a significance level

Main AP Stats units

Units 1 and 5

Units 3 and 4, built on Unit 2

The vocabulary does a lot of the work. A statistic is a number calculated from a sample, such as the sample mean x̄ or the sample proportion p̂. A parameter is the matching number for the whole population, such as the population mean μ or the population proportion p, and it is usually unknown. Descriptive statistics calculate statistics. Inferential statistics use statistics to estimate or test parameters.

One study, both kinds of statistics

Suppose a high school with 1,500 students wants to know how much its students sleep on school nights. Surveying everyone isn’t practical, so it selects a simple random sample of 120 students and asks each one.

The descriptive part. The 120 answers have a mean of x̄ = 6.7 hours and a standard deviation of s = 1.2 hours. The median is 6.75 hours, a histogram of the answers is roughly symmetric with no outliers, and 78 of the 120 students (65%) report less than 7 hours. Every one of those numbers is a fact about the 120 students in the sample. None of them, on its own, says anything about the other 1,380.

The inferential part: a confidence interval. To estimate the mean sleep μ of all 1,500 students, use a one-sample t interval. The conditions hold: the sample is random, 120 is less than 10% of 1,500, and n = 120 is large enough for the central limit theorem to apply. The standard error is 1.2 / √120 ≈ 0.11 hours, and with t* ≈ 1.98 the margin of error is about 0.22 hours. The interval is 6.7 ± 0.22, or about 6.48 to 6.92 hours. Interpretation: we are 95% confident that the interval from 6.48 to 6.92 hours captures the true mean school-night sleep of all students at the school.

The inferential part: a significance test. Is there convincing evidence that the school’s students average less than 7 hours? The null hypothesis is H₀: μ = 7 and the alternative is Hₐ: μ < 7, tested at α = 0.05. The test statistic is t = (6.7 − 7) / 0.11 ≈ −2.74, which with 119 degrees of freedom gives a p-value of about 0.004. Because 0.004 is less than 0.05, reject H₀: there is convincing evidence that the mean school-night sleep of all students at this school is less than 7 hours.

The same 120 numbers fed both halves. The descriptive summaries (x̄, s and n) are the inputs to the inferential formulas, which is why AP Statistics teaches describing data first. Our guide to finding a p-value from a z-score shows how a test statistic becomes a p-value, using the z version of this calculation.

What descriptive statistics include

Descriptive statistics organize and summarize data so its main features are easy to see. In AP Statistics, describing a distribution of one quantitative variable means covering four things, always in the context of the data:

  • Shape: symmetric or skewed (left or right), and how many peaks it has.
  • Center: the mean (the sum of the values divided by how many there are) or the median (the middle value when the data are in order). The median is resistant to outliers; the mean is pulled toward them.
  • Variability: the range (maximum minus minimum), the interquartile range (IQR, the third quartile minus the first) and the standard deviation (roughly the typical distance of the values from the mean). The variance is the standard deviation squared.
  • Unusual features: gaps, clusters and outliers. A common rule flags a value as an outlier when it lies more than 1.5 × IQR above the third quartile or below the first.

Descriptive statistics also include where a single value stands (percentiles, and z-scores, which count how many standard deviations a value is from the mean) and the graphs that show a distribution: dotplots, stemplots, histograms and boxplots for quantitative data, and bar graphs and two-way tables for categorical data.

For two quantitative variables, a scatterplot, the correlation r and a least-squares regression line describe the relationship in the data you have. In the current course, regression (Unit 5) is taught as description: interpreting the slope, the residuals and r² for the observed data. Inference for the slope of a regression line was removed in the fall 2026 revision.

OpenStax’s Introductory Statistics, a free college textbook, covers the same descriptive tools in its second chapter if you want more worked examples.

What inferential statistics include

Inferential statistics answer questions about a population using a sample, and every answer comes with a statement of how much it could be wrong. The two procedures AP Statistics teaches are confidence intervals and significance tests.

  • Confidence intervals estimate a parameter with a range: point estimate ± margin of error. The confidence level, often 95%, describes the method, not one interval: in repeated random sampling, about 95% of intervals built this way capture the true parameter. Once an interval is calculated, it either contains the parameter or it doesn’t, so “there is a 95% probability that μ is in this interval” is the wrong interpretation. Our confidence interval definition has more on the wording.
  • Significance tests weigh evidence against a claim. The null hypothesis H₀ is the claim of no effect or no difference; the alternative Hₐ is what you are looking for evidence of. The p-value is the probability, assuming H₀ is true, of getting a result at least as extreme as the one observed. If the p-value is at or below the significance level α, you reject H₀; if it is above, you fail to reject H₀.

Two points about the significance level trip up many students. First, α is a probability, usually set in advance at 0.05 (sometimes 0.01 or 0.10), not a percentage like 95%: 95% is a confidence level, which belongs to intervals. Second, α is the probability of a Type I error, rejecting a null hypothesis that is actually true. A small p-value says the data would be surprising if H₀ were true; it is not the probability that H₀ is true, and failing to reject H₀ doesn’t prove it.

Inference works only when its conditions hold, which is why every AP free-response inference answer checks them:

  • Random: the data come from a random sample or a randomized experiment.
  • Independence (10% condition): when sampling without replacement, the sample is no more than 10% of the population.
  • Normal or large sample: for proportions, at least 10 successes and 10 failures expected in the sample; for means, a sample size of at least 30, or a population that is approximately normal.

In AP Statistics the procedures are z intervals and tests for one or two proportions, t intervals and tests for one mean, a mean difference or two means, and chi-square tests for homogeneity and independence. Penn State’s review of hypothesis testing and its review of confidence intervals go through the same logic in more depth.

Why a sample can tell you about a population

Different random samples from the same population give different statistics. That sampling variability is what makes inference possible, because it follows a predictable pattern: the sampling distribution of a statistic. The central limit theorem says that for large enough samples, the sampling distribution of a sample mean is approximately normal, centered at μ, with a standard deviation of σ / √n.

That is the bridge between the two kinds of statistics. The normal model, including the empirical rule (about 68%, 95% and 99.7% of values within one, two and three standard deviations of the mean), tells you how far a statistic is likely to land from the parameter. A margin of error is built from that, and so is a p-value. In the AP course this material sits in Unit 2, between the descriptive work of Unit 1 and the inference of Units 3 and 4.

How data are collected decides what you can conclude

No calculation fixes a badly collected sample. The course’s rule, the scope of inference, has two parts:

  • Random selection lets you generalize. When the individuals in a study were randomly selected from a population, you can extend the results to that population. When they weren’t (volunteers, or whoever was easiest to reach), the results apply only to individuals like the ones studied. A diet study that enrolls only adults over 60 can’t be generalized to all adults.
  • Random assignment lets you claim cause and effect. When treatments are randomly assigned in an experiment, confounding variables are balanced out, so a significant difference can be attributed to the treatment. An observational study can show an association but not a cause.

A population isn’t necessarily a country. It is whatever group the question is about: the 1,500 students at one school, every batch from one factory, or the adults in one city. The sample has to represent that group. Our AP Psychology guide to inferential statistics shows how researchers apply the same sampling ideas to studies of behavior.

Where each one shows up in AP Statistics

College Board revised AP Statistics for the 2026-27 school year, combining the old nine units into five. The current course and exam description sets each unit’s share of the multiple-choice section:

Unit

Share of multiple-choice section

Mostly

1. Exploring One-Variable Data and Collecting Data

20–30%

Descriptive, plus how to collect data

2. Probability, Random Variables, and Probability Distributions

15–25%

The bridge: probability, the normal distribution and sampling distributions

3. Inference for Categorical Data: Proportions

15–25%

Inferential: intervals and tests for proportions, chi-square tests

4. Inference for Quantitative Data: Means

10–20%

Inferential: intervals and tests for means

5. Regression Analysis

10–20%

Descriptive: correlation, least-squares lines and residuals

Units 1 and 5, the descriptive and study-design work, make up 30–50% of the multiple-choice section, and probability and inference make up the rest. Our AP Statistics units guide covers what each unit includes and what changed in the revision.

The free-response section mirrors the same split. It has four questions worth 10 points each: the first is about formulating a question and collecting data, the second about analyzing and interpreting data (mostly descriptive), the third is an inference question (a test or an interval), and the fourth mixes several parts of the course. The exam is 3 hours and fully digital in the Bluebook app, it is scheduled for May 11, 2027, and you need a graphing calculator with statistical capabilities. The formula and table sheet is provided for both sections; our AP Stats formula sheet guide explains what’s on it. The AP Statistics exam page on College Board’s student site has the latest format details.

Common mistakes on the exam

  • Treating a sample result as a population fact. “The mean is 6.7 hours” describes the sample. A claim about all students needs an interval or a test.
  • Mixing up statistic and parameter notation. Hypotheses are always about parameters (μ or p), never about x̄ or p̂.
  • Misreading a confidence interval. Say “we are 95% confident the interval captures the true mean”, not “there is a 95% chance the mean is in the interval”, and not “95% of the data fall in the interval”.
  • Accepting the null hypothesis. A large p-value means you fail to reject H₀; it doesn’t show H₀ is true.
  • Skipping or misstating conditions. Name the random, 10% and normal or large-counts checks with the numbers from the problem.
  • Generalizing too far, or claiming cause from an observational study. Match the conclusion to how the data were collected.
  • Leaving out context. Every description and every conclusion should name the variable and the population in the words of the problem.

Key takeaways

  • Descriptive statistics summarize the data you have; inferential statistics use a sample to reach a conclusion about a population, with a stated level of uncertainty.
  • Statistics (x̄, s, p̂) describe samples. Parameters (μ, σ, p) describe populations and are what inference estimates and tests.
  • Confidence intervals estimate a parameter; significance tests weigh evidence against a null hypothesis. α is usually 0.05 and is the probability of a Type I error.
  • Random selection lets you generalize to the population; random assignment lets you conclude cause and effect.
  • In the five-unit AP Statistics course, Units 1 and 5 are mainly descriptive, Unit 2 is the probability bridge, and Units 3 and 4 are inference.
Julia Tache's profile picture
Julia Tache
14 Nov 2026, 10 min read