Scatterplots and lines of best fit
A scatterplot question is rarely asking you to compute the line. It usually asks what one number in the equation means in the situation, which is a reading skill with a small amount of algebra under it.
What the graph gives you
The dots are observations. The line is a model fitted through them. When a question hands you an equation such as y = 1.8x + 12 alongside the graph, it is handing you a translation job: turn 1.8 and 12 back into the sentence the situation started as.
Slope is a rate, and it carries units
The slope is the change in y for each one-unit increase in x, measured in the units of the context. If x is hours and y is dollars, a slope of 3 means the model predicts 3 more dollars for each additional hour. The phrasing on the test follows the same shape every time: increases by about, for each additional. Match the number to the unit, then read the sentence back.
The intercept is the starting value
In y = mx + b, the b is the predicted y when x is 0. If x is years since a purchase, then b is the value at purchase, not today. That reading only makes sense when 0 sits inside the data's range, which is worth checking before you choose an answer about it.
Predicted versus actual
Putting an x into the equation gives the predicted value. Reading the dot at that x gives the actual value. The difference between them tells you whether the model runs above or below that observation, and a question that asks how much higher or lower is asking for exactly that subtraction.
Worked example
Four tutoring sessions are recorded at 1, 2, 3 and 4 hours, costing 27, 30, 33 and 36 dollars. The line through them is C = 24 + 3h. The 3 is the rate: each additional hour adds about 3 dollars. The 24 is the predicted cost of a session of zero hours, which is where the booking fee would sit. The model predicts 39 dollars at 5 hours, but 5 is outside the recorded range, so that number is an extrapolation rather than a supported prediction.
Stay inside the data
The model describes the range it was fitted on. Extending it past the last plotted x is a guess dressed as arithmetic, and test questions usually use it as the wrong answer with a phrase such as may not be reliable. A prediction near the middle of the data is safe; a prediction far outside it is the answer to eliminate.
What does the slope mean in a line of best fit?
The predicted change in the y variable for each one-unit increase in the x variable, expressed in the units of the situation. For each additional hour, the model predicts 3 more dollars is the standard sentence.
What is the difference between predicted and actual?
The predicted value comes from substituting x into the equation. The actual value is the observed dot. Subtracting one from the other tells you how far above or below the model that observation sits.
Can I use a line of best fit to predict outside the data?
The equation will always produce a number, but the further outside the observed range you go, the less the model supports it. Test questions treat a distant prediction as unreliable.
How do I know which axis is which?
Read the axis labels, not the title. The x variable is the one being changed or chosen, and the y variable is the one being measured as a result.
Related guides
Updated 2026-10-25. Free to use, no account needed. SAT is a trademark of College Board. Not affiliated with or endorsed by College Board.