SAT Problem-Solving and Data Analysis Study Guide
Prepare for SAT Math. Study ratios, percentages, data, probability and statistical claims. Choose a lesson below, or try mixed practice to check your skills.
Alongside the core guide, the topic lessons include additional instruction, 2 worked examples and 6 targeted practice questions. Find these sets under Practice options.
All lessons
7 lessons
- Ratios, rates, proportions, and unitsPractice
- PercentagesPractice
- One-variable data: distributions, center, and spreadPractice
- Two-variable data: models and scatterplotsPractice
- Probability and conditional probabilityPractice
- Sample-based inference and margin of errorPractice
- Evaluating statistical claimsPractice
About this guide
Seven complete lessons, 105 worked examples, 35 topic-practice exercises and 42 mixed-practice questions.
How to use this guide
This volume is designed for active problem solving, not passive reading. The hardest work often happens before the arithmetic: choosing the correct base, denominator, units, comparison, or population. The explanations make those decisions visible.
A study cycle that produces evidence of mastery
Learn the lesson. Read the teaching content at the beginning of a section. Reproduce the key relationships on scratch paper and explain their conditions aloud. A formula without its conditions is only partly learned.
Attempt before looking. Keep the worked solution closed and solve the question. If stuck, open it to read “Recognize the structure,” then close it and try again. Do not confuse following an explanation with being able to create one.
Compare decisions, not just answers. After checking the answer, read the check and the trap. Ask which step would change if the problem asked for a different subgroup, a reverse percentage, a different unit, or a stronger conclusion.
Test independently. Complete the five exit questions at the end of each section before turning to their solutions. Later, use the three mixed sets, which contain no topic labels or worked-example hints in the question prompts.
Return after a delay. Re-solve missed questions from a blank page, then try a different question testing the same skill. The skills map and solution links make targeted review straightforward.
What the labels mean
Foundation isolates an essential idea. Standard combines that idea with typical representations or contextual interpretation. Challenge requires an additional modeling decision, constraint, or subtle distinction. These labels are editorial learning levels, not official SAT difficulty classifications.
How the material is organized
Each of the seven sections contains concept lessons, 15 worked examples, five exit questions, and complete exit solutions. The mixed-practice chapter adds 42 questions in three sets of 14. All 77 independent questions have explanations and links to a related worked example.
Some learning questions ask for explanations or related quantities rather than replicating the test interface exactly. The mixed sets combine multiple-choice and numeric-response questions; they are not adaptive modules and should not be converted into predicted SAT scores.
A note on data and authorship
All example data, studies, and questions are original and hypothetical, not findings about real populations. References support test alignment, calculator instructions, and statistical principles, not the invented data.
SAT and Bluebook are trademarks of College Board. Desmos is a trademark of its respective owner. This is an independent study guide, not a College Board publication or endorsement.
Scope and readiness
How this volume fits the SAT
College Board organizes Problem-Solving and Data Analysis into the seven skill areas used here.[1] The domain accounts for approximately 15% of SAT Math questions.[2] The standard Math section contains 44 questions across two 35-minute modules, with the second module adapted to first-module performance.[4]
The guide covers the listed representations as well as the numerical skills: frequency tables, dot plots, histograms, box plots, scatterplots, and probability tables or area models. It also includes quadratic fits alongside linear and exponential fits. Supporting tools are clearly distinguished from new formula requirements.
What this guide does not turn into an extra syllabus
Long combinatorics calculations, memorized normal-distribution tables, z- and t-test computations, and advanced regression diagnostics are not prerequisites for this volume. Population standard deviation is shown to explain spread, not to require lengthy hand calculation. Residuals, simple interest, the supplied E
Prerequisites to refresh as needed
You should be comfortable with fraction arithmetic, decimal and percentage forms, order of operations, one-variable linear equations, and substitution into a formula. Know that a quotient such as 3/8 is exact, whereas a rounded decimal may not be. You should also be able to read the two coordinates of a point and distinguish a square unit from a length unit.
Five questions that travel through every section
Target: What exactly is being requested: a count, proportion, percent, value, interval, or conclusion?
Base: What is the relevant whole, denominator, or comparison group?
Units: What does each number measure, and will the final units match?
Representation: Which information is actually present in the table, graph, or equation?
Evidence: Is this an exact calculation, a model prediction, a sample estimate, or a claim about cause?
Precision and calculator discipline
Keep exact fractions or extra decimal precision during intermediate calculations. Round only when the question requests it or when formatting a final numeric response requires it. Decide what to calculate before entering numbers. For a statistical claim, additional decimal places cannot repair the wrong sampling method or establish causation.
A useful mastery standard
Move on when you can solve a fresh problem without a hint, explain why its denominator or model is correct, and identify a plausible wrong method. A correct guess is a review item. An incorrect calculation with a sound model needs a different repair from a perfectly calculated answer to the wrong question.