What is a line of best fit on the SAT?

It is the straight line drawn through a scatterplot to summarize the overall trend. It does not touch every point; it sits so the points are as close to it as possible, balanced above and below. Once the line exists it behaves like any linear equation, y = mx + b, where the slope m is the rate and the y-intercept b is the starting value. The SAT usually gives you the line or its equation and asks what the pieces mean or what the line predicts.

How do you tell if a model is linear, quadratic, or exponential?

Match the shape to the model. A linear trend climbs or falls at a steady rate in equal steps. An exponential trend grows or shrinks by a percentage each step, so the curve gets steeper and steeper. A quadratic trend rises then falls, or falls then rises, making a U or an upside-down U with a single turning point. You identify the shape rather than deriving the equation.

SAT Two-Variable Data

Last updated: July 14, 2026

Two-variable data questions are part of the Problem-Solving and Data Analysis domain on the digital SAT, which is about 15% of the math section — roughly 5–7 of the 44 math questions. They give you two quantities that move together, usually as a scatterplot with a line or curve of best fit, and ask you to read a value off the model, interpret the slope or intercept in context, or use the line to predict. The arithmetic is light; the points come from knowing what each part of the model means.

What does this skill actually test?

Whether you can look at two variables plotted together and describe how they relate. One variable on the x-axis, one on the y-axis, a cloud of points, and a line or curve drawn through the middle to summarize the trend. That line is the model. Almost every question is really asking you to read that model or explain one piece of it.

The data shows up as a scatterplot most of the time. Your job is to connect the picture to the words: what the slope means, what the y-intercept means, and what the line predicts for a value you can't see directly.

What is a line of best fit?

It's the straight line drawn through a scatterplot to summarize the overall trend. It doesn't touch every point — it sits so the points are as close to it as possible, balanced above and below.

Once that line exists, it behaves like any linear equation you already know: y = mx + b. The slope m is the rate, the y-intercept b is the starting value. The SAT rarely makes you find the line's equation from scratch. It gives you the line or its equation and asks what the pieces mean or what the line predicts.

How do you interpret slope and intercept in context?

The slope is how much the y-variable changes for each one-unit increase in the x-variable. If a scatterplot shows plant height versus days, a slope of 0.4 means the plant grows about 0.4 cm per day. Not "0.4 total" — 0.4 for every single day. Units are the whole game here.

The y-intercept is the predicted y-value when x is zero — the starting point before anything changes. Same plant, an intercept of 2 means the model estimates 2 cm on day zero. Read the axis labels, attach the units, and the interpretation writes itself.

Linear, exponential, or quadratic — which model fits?

Not every trend is a straight line. The SAT also uses curves of best fit, and it expects you to match the shape to the model.

Linear — points climb or fall at a steady rate. Equal steps.
Exponential — the y-values grow (or shrink) by a percentage each step, so the curve gets steeper and steeper. Think doubling.
Quadratic — the points rise then fall, or fall then rise, making a U or an upside-down U with a single turning point.

You're identifying the shape, not deriving the equation. Steady line, accelerating curve, or a single peak — that's the read.

How the SAT asks these

A few common shapes:

Interpret a coefficient. "What does the slope of the line of best fit represent?" Attach the units from the axes and say it in a sentence.
Predict a value. Plug an x-value into the model, or read straight up from the x-axis to the line and across to the y-axis.
Match the model. Given a scatterplot, decide whether a linear, quadratic, or exponential model fits best.
Compare a point to the line. Ask whether the actual data value at some x is above or below what the model predicts, and by how much.

The calculator is available on the entire math section, and the built-in Desmos calculator will plot points and fit a line for you once the data is entered. Reading the axes and knowing which piece of the model they want is still on you.

Practice routine

Work a set of 12–15 scatterplot questions mixing all four shapes above. After each one:

Say the slope out loud as a sentence with units before you look at the answers, because a slope without units is where most interpretation questions get missed
When you predict a value, check whether the question wants a point read from the line or an actual data point near it — they are not the same number
On model-matching questions, name the shape (line, curve up, single peak) before you read the options, so the picture drives the choice instead of the wording

This skill sits in the Problem-Solving and Data Analysis group, where the math is light but the context reading is where points leak. If you want to know whether yours is clean or still costing you, join the HIROSCORE beta and see exactly where.

HIROSCORE tracks your accuracy on two-variable data separately from the rest of the math section, so you know if it's solid instead of guessing. The GPS for your SAT score.

References