Blocking in AP Statistics

In AP Statistics, blocking means grouping experimental units by a blocking variable, a source of extraneous variation in the response, and then randomly assigning treatments within each block. This makes treatment comparisons within more alike groups and reduces unexplained variability.

What is Blocking?

Blocking is a design choice you make before you assign treatments. The CED (1.13.B.2) defines a blocking variable as a source of extraneous variation in the response variable. In a randomized block design, you first sort the experimental units into groups called blocks based on similar values of that blocking variable. Units in the same block are homogeneous (alike) with respect to the blocking variable. Then you randomly assign treatments separately within each block.

Here's the intuition. Say you're testing three teaching methods and prior knowledge strongly affects test scores. If you mix everyone together, the differences in prior knowledge add noise that can hide a real treatment effect. If you block by prior knowledge, you compare methods among students who started at a similar level. Blocking is basically "compare apples to apples, then randomize." It doesn't replace random assignment. It just organizes the randomization so that one known source of variation can't muddy the comparison.

Why Blocking matters in AP® Statistics

Blocking lives in Topic 1.13 Experimental Design in Unit 1: Exploring One-Variable Data and Collecting Data. It supports several learning objectives. AP Stats 1.13.B asks you to identify experimental designs, and blocking is what separates a randomized block design from a completely randomized design (1.13.B.1 vs. 1.13.B.2). AP Stats 1.13.C asks you to justify why one design is more appropriate than another based on the study's goals and the units and variables involved. That is exactly the argument for blocking: one variable is known to affect the response, so you block on it. It also ties to AP Stats 1.13.A, because blocking is one way to get "direct control of potential extraneous sources of variation in the response" (1.13.A.1.iv). The bigger theme is that good design comes before good analysis. Less noise in the data means a clearer signal when you run inference later.

How Blocking connects across the course

Matched Pairs Design (Unit 1)

A matched pairs design is blocking taken to the extreme, with blocks of size two. You either pair up two very similar units or give both treatments to the same unit. Recognizing matched pairs as a special case of blocking makes both designs easier to describe on an FRQ.

Randomization and Random Assignment (Unit 1)

Blocking and random assignment work as a team. Blocking handles one variable you know matters. Random assignment within each block handles all the variables you don't know about, which is what still allows cause-and-effect conclusions (1.13.D.1).

Stratified Random Sampling (Unit 1)

Stratifying is the sampling version of blocking. Both group similar individuals first. Stratifying is about selecting a sample from a population, though, while blocking is about assigning treatments in an experiment. Same instinct, different job.

Paired Data and Statistical Significance (Unit 4)

When a matched pairs design produces paired data, Unit 4 analyzes the differences within each pair. Because blocking strips out variation between units, the treatment effect has less noise to hide behind. That makes a real effect more likely to show up as statistically significant.

Is Blocking on the AP® Statistics exam?

Blocking shows up as an identify-and-justify skill. In multiple choice, a typical stem describes a study and names a variable that might affect results, then asks which design technique best handles it. One example is a training program where nutrition might affect athletic performance. Another is 60 students testing three teaching methods where prior knowledge matters. The answer is a randomized block design on that variable. Watch for distractors. Tasting four chocolates in a random order is randomization, not blocking.

On free response, experimental design questions often ask you to describe or justify a design for a specific context. 2024 FRQ Q3 asked about researchers investigating which of four car models (A, B, C, D) gets the best gas mileage. That's the kind of setting where you need to explain how a design controls variation in the response. When you block on an FRQ, do three things. (1) Name the blocking variable. (2) Explain why it would affect the response variable, in context. (3) State that treatments are randomly assigned within each block and compared within blocks. Leaving out the randomization step or the in-context reason is the most common way to lose credit.

Blocking vs Stratified Random Sampling

Both split individuals into similar groups first, which is why they get mixed up. Stratified random sampling is a sampling method. You divide a population into strata and take a random sample from each, so you get a more representative sample. Blocking is an experimental design method. You group the units you already have and then randomly assign treatments within each block, so you reduce variability in treatment comparisons. A quick test helps. If the question is about choosing who's in the study, think strata. If it's about who gets which treatment, think blocks.

Key things to remember about Blocking

• A blocking variable is a source of extraneous variation in the response variable, and you group experimental units by it before assigning treatments.

• In a randomized block design, treatments are randomly assigned separately within each block, so blocking never replaces random assignment.

• You should block on a variable that you expect to affect the response, and on an FRQ you need to explain that connection in context.

• Blocking reduces variability in treatment comparisons, which makes it easier to detect a real difference between treatments.

• A matched pairs design is a special type of randomized block design where each block has two units or one unit receives both treatments.

• Blocking happens in experiments when treatments are assigned, while stratifying happens in sampling when individuals are selected.

Frequently asked questions about Blocking

What is blocking in AP Statistics?

Blocking is grouping experimental units by a variable that affects the response, then randomly assigning treatments within each group (block). The CED covers it in Topic 1.13 Experimental Design as part of the randomized block design.

Does blocking replace random assignment?

No. In a randomized block design, you still randomly assign treatments, just separately within each block. Blocking controls one known variable, and random assignment controls everything else, which is what supports cause-and-effect conclusions.

What's the difference between blocking and stratified random sampling?

Stratified random sampling divides a population into strata to select a sample, while blocking groups experimental units to assign treatments in an experiment. Strata deal with sampling, and blocks deal with experimental design.

Is matched pairs a type of blocking?

Yes. A matched pairs design is a randomized block design where each block has just two similar units, or where a single unit receives both treatments in random order.

How do you explain blocking on an AP Stats FRQ?

Name the blocking variable, explain in context why it would affect the response variable, and state that treatments are randomly assigned within each block and results are compared within blocks. For example, if prior knowledge affects test scores, you'd block students by prior knowledge before randomly assigning teaching methods.