All articles / SAT Math / Problem-Solving & Data Analysis
Two-Variable Data: Models & Scatterplots for the SAT — Complete Study Guide + Free Practice Problems
Everything the SAT tests about two-variable data in one place: scatterplots and correlation, lines of best fit, nonlinear models, residuals, and interpreting models in context — with step-by-step examples, worked problems, and 2 free practice quizzes.
Practice these quizzes
Free · No signupWhat is two-variable data?
Two-variable data records two related measurements for each item in a group, like hours studied and test score, or a car's age and its resale value. The SAT tests whether you can visualize the relationship between the two variables, describe it, and use a mathematical model to make predictions.
Every two-variable data question boils down to one central idea: as one variable changes, what happens to the other? Keep that question in mind and most of these problems become much more approachable.
Scatterplots & correlation
A scatterplot plots each pair of values as a point on the coordinate plane. The overall pattern of the points describes the correlation between the two variables.
| Pattern | Correlation | Meaning |
|---|---|---|
| Points trend upward, left to right | Positive | As x increases, y tends to increase |
| Points trend downward, left to right | Negative | As x increases, y tends to decrease |
| No clear upward or downward trend | None / weak | The two variables show no linear relationship |
Correlation describes association, not causation. A scatterplot showing a strong positive correlation between two variables never proves that one variable causes the other to change — the SAT tests this distinction directly.
The line of best fit
When a scatterplot shows a roughly linear pattern, a line of best fit (also called a trend line or regression line) is drawn through the data to model the relationship as closely as possible. It follows the same slope-intercept form as any linear equation.
A line of best fit for hours studied (x) vs. test score (y) is y = 6x + 58. Predict the score for a student who studies 5 hours.
Interpolation & extrapolation
Interpolation
Predicting a value that falls inside the range of the original data. Generally reliable.
Extrapolation
Predicting a value outside the range of the original data. Less reliable — the trend may not continue.
The SAT often asks you to critique a prediction that extrapolates far beyond the original data range. Even a strong model can become inaccurate outside the observed range — recognizing this limitation is frequently the actual skill being tested.
Nonlinear models
Not every scatterplot follows a straight-line pattern. Data that curves upward faster and faster suggests a quadratic or exponential model instead of a linear one.
| Scatterplot pattern | Best model type |
|---|---|
| Roughly straight-line trend | Linear: y = mx + b |
| Curves up then down (or down then up), symmetric | Quadratic: y = ax² + bx + c |
| Curves increasingly steeply in one direction | Exponential: y = a · bₓ |
If the SAT shows a scatterplot alongside several candidate equations, you often don't need to calculate anything — just match the general shape of the curve to the correct function family first, then narrow down further if needed.
Residuals & model fit
A residual is the difference between an actual data value and the value predicted by the model: residual = actual − predicted. Residuals close to zero mean the model fits that point well.
A model predicts a score of 82 for 4 hours studied. A real student who studied 4 hours actually scored 89. What is the residual?
A positive residual means the actual value was above the model's prediction; a negative residual means it fell below. If residuals show a clear pattern (like a curve) instead of scattering randomly, a linear model is probably not the best fit.
Interpreting models in context
Just like linear function questions, two-variable data questions frequently ask what the slope or y-intercept of a model represents in real-world terms.
A line of best fit for a car's age (x, in years) and value (y, in dollars) is y = −1200x + 22000. What does the −1200 represent?
Test day strategy for two-variable data
| Question signal | Fastest approach |
|---|---|
| "Describe the relationship" from a scatterplot | Check the overall trend direction: positive, negative, or none |
| "Correlation implies causation" style question | Reject it — correlation never proves causation on its own |
| Prediction using a line of best fit | Substitute the given x-value directly into the model equation |
| Prediction far outside the given data range | Flag it as extrapolation — less reliable than interpolation |
| Curved scatterplot with answer choices | Match the curve's shape to linear, quadratic, or exponential first |
| "What does the slope/intercept represent" question | Connect each part of the equation to its real-world meaning in context |
Now put it to work
Two quiz sets, each building on the last — start with Quiz 1 and work through in order, or jump straight to the topic you need.
Two-Variable Data 1
Scatterplots, correlation, and lines of best fit.
Start quiz → Quiz 2Two-Variable Data 2
Nonlinear models, residuals, and interpreting models in context.
Start quiz →