ACT Math: Linear Regression and Best-Fit Lines for Scatter Plots

Linear regression fits a straight line through scatter plot data to model the relationship between two variables. The best-fit line minimizes vertical distances between the line and data points. The equation is y=mx+b, where m is the slope and b is the y-intercept. On the ACT, you usually don't calculate the line from scratch; instead, you're given data or a scatter plot and asked to: (1) Identify the approximate best-fit line, (2) Interpret the slope (rate of change), (3) Interpret the y-intercept (starting value), (4) Use the line to make predictions. Example: A scatter plot shows study hours vs test scores. A best-fit line might be y=10x+50, meaning each additional hour of study correlates with 10 points higher score; students studying zero hours score roughly 50. The best-fit line models the trend, but individual points deviate from it; that deviation is called residual or error.

Correlation vs causation: A strong linear relationship doesn't prove cause-and-effect. Study hours and test scores correlate, but other factors (sleep, prior knowledge) also matter. The line models the relationship, not proof of causation.

Mistake 1: Assuming all data points lie exactly on the best-fit line. They don't; the line is an approximation. Mistake 2: Confusing correlation with causation. "Ice cream sales and drowning deaths are correlated" (temperature causes both), but ice cream doesn't cause drowning. Mistake 3: Predicting outside the data range (extrapolation). If data shows hours 0-10 on the x-axis, predicting for hour 100 is unreliable. Mistake 4: Misinterpreting slope. Slope=5 means for every 1 unit increase in x, y increases by 5 units. If slope=-2, y decreases by 2 for every 1 unit increase in x. Always double-check: Is the slope positive or negative? What units am I using? Am I predicting within the data range?

Checklist: (1) Identify the best-fit line equation. (2) Interpret slope in context (does the direction match the scatter plot?). (3) Interpret y-intercept (does it make sense?). (4) Use for prediction only within data range. (5) Remember: the line models trend, not exact values.

Scenario 1: Scatter plot of age vs salary. Data ranges age 20-60, salary $30k-$100k. Best-fit line: y=1500x-30000 (roughly). Interpretation: For each additional year of age, salary increases by $1500. A 40-year-old earns roughly $30,000 (slope×40+intercept). Scenario 2: Scatter plot of temperature vs ice cream sales. Data ranges 70-95°F, sales $100-$500. Best-fit line shows positive slope. Interpretation: As temperature rises, sales rise. This is correlation; temperature doesn't directly cause ice cream sales, but both respond to seasonal factors. Scenario 3: Scatter plot of hours studied vs test score. Data ranges 0-8 hours, scores 50-95. Best-fit line: y=6x+50. Interpretation: Each additional hour correlates with 6 points higher. Prediction: 5 hours study → 80 points. For each scenario, write the slope's interpretation and a sample prediction within the data range.

Daily drill: Find scatter plot data online (any topic). Estimate the best-fit line. Calculate slope and y-intercept. Make predictions within the range. Compare your line to actual data.

About 1-2 statistics questions per ACT Math section involve linear regression or best-fit lines. These combine graphing, equation interpretation, and prediction—multiple skills at once. If you master the concept, these questions are straightforward. Many students find statistics intimidating, so correctly answering regression questions boosts your relative score and demonstrates data literacy that extends beyond the test.

Spend 2-3 days on linear regression. Practice interpreting slope and y-intercept, making predictions, and recognizing the limits of extrapolation. By test day, regression questions will feel manageable.