Correlation and residual plot worksheet with answer key
8 problems on correlation and residual plot, ready to print for class or homework. The answer key prints on its own page, so hand out everything before it.
Name: ______________________________ Date: ______________ Period: ______
Problem 1
A district compares its 40 elementary schools and finds a correlation of between the average number of minutes students read at home each day, , and the school's average reading score, . (a) Interpret in context. (b) Find and interpret it in context. (c) A parent writes that 60 percent of the variation in reading scores is explained by home reading time. Explain what is wrong with that sentence. (d) The district is considering recording reading time in hours instead of minutes. What would that do to and to ?
Problem 2
A moving company predicts a job's labour time in hours, , from the number of rooms in the home, , using the least-squares line . (a) A 6-room job actually took 7 hours. Find the residual and say what its sign means. (b) A 4-room job had a residual of hour. Find how long that job actually took. (c) One job had a predicted time of 6.75 hours. How many rooms did it have?
Problem 3
A plant records the age in years of each packing machine, , and the units it produces per hour, , for five machines: , , , , . The least-squares line is . (a) Compute all five residuals. (b) Show that they add to zero and explain why they had to. (c) Which machines did the model overpredict, and by how much? (d) Describe the residual plot and say whether a line is a reasonable model.
Problem 4
A sleep researcher regresses nightly sleep in hours, , on evening screen time in hours, , for 120 teenagers. The fitted line has slope and the output reports . (a) Find and explain how you know its sign. (b) Interpret in context and say what the rest of the variation represents. (c) Interpret the slope in context. (d) The residual plot is a fan: residuals sit tightly around zero near and spread far above and below the zero line at large . What does that say about predicting the sleep of a teenager with 5 hours of evening screen time?
Problem 5
A marine lab tracks an invasive algae patch. The weeks since it was first recorded, , and its area in square metres, , are , , , , . (a) Find the least-squares regression line. (b) Compute the five residuals. (c) Describe the residual plot. (d) The correlation is , so . Reconcile that with part (c) and say whether a line is the right model here.
Problem 6
In a physics lab, five masses are hung from a spring and the spring's length is recorded. With the mass in kilograms and the length in centimetres, the readings are , , , , , which lie exactly on the line with . A sixth reading is taken at kg, but the student misreads the ruler and records a length of 2 cm. (a) Find the least-squares line for all six points. (b) What happened to the slope, and why? (c) Find and for all six points. (d) Is the bad reading an outlier, a high-leverage point, both, or neither?
Problem 7
A nursery gives six potted seedlings different weekly amounts of water and records the season's growth. With the water in litres per week and the growth in centimetres, the results are , , , , , . The sixth seedling was overwatered and its roots rotted. (a) Using only the first five seedlings, find the least-squares line and . (b) Find the least-squares line for all six. (c) Find and for all six. (d) Explain how one point produced that change, and say what you would report.
Problem 8
A reviewer tests 18 laptops, fixing each screen at a set brightness and timing the battery. Across the machines the brightness settings ranged from 20 to 80 percent, and battery life is in hours. The regression output is below, together with R-Sq = 68.89 percent.
| Predictor | Coef |
|---|---|
| Constant | 12.4 |
| Brightness |
(a) Write the least-squares regression line, defining your variables. (b) Interpret the slope in context. (c) Find . (d) One machine set to 50 percent brightness ran for 8.2 hours. Find and interpret its residual. (e) A nineteenth machine is tested at 100 percent brightness and runs for 9.0 hours. Without refitting, describe what adding it would do to the slope and to , and explain why.
From statslearn.org/worksheets/correlation-residuals-worksheet. Free to print and copy for classroom use.
Correlation and residual plot: answer key
(a) A moderately strong positive linear association: schools with more home reading minutes tend to have higher average scores. (b) , so about 36 percent of the variation in average reading score is explained by the linear relationship with home reading minutes. (c) The parent reported instead of ; the correct figure is 36 percent, and it still describes explained variation, not cause. (d) Neither moves: stays and stays , because is built from standardized values and carries no units.
(a) Predicted 6 hours, so the residual is hour and the model underpredicted by 1 hour. (b) Predicted 4.5 hours, so the job took hours and the model overpredicted by 0.5 hour. (c) rooms.
(a) Residuals units per hour. (b) They sum to , which least squares forces because the line passes through . (c) The 4-year-old machine was overpredicted by 3 units per hour and the 10-year-old by 1. (d) The residuals scatter above and below zero with no curve or fan, so a line is a reasonable model, though five points is thin evidence.
(a) , negative because the slope is negative. (b) About 57.8 percent of the variation in nightly sleep is explained by the linear relationship with evening screen time; the other 42.2 percent is not. (c) Each extra hour of evening screen time predicts about 0.45 fewer hours of sleep, roughly 27 minutes. (d) The fan shows residual spread growing with , so a prediction at 5 hours is far less reliable than one near 0, whatever says.
(a) . (b) Residuals square metres. (c) A clear U shape: positive at both ends, negative through the middle. (d) A line is not the right model. measures how much variation the line removes, not whether the form is right, and the U-shaped residual plot shows the growth curves upward.
(a) . (b) The slope stays exactly 3, because a point at sits in the lowest-leverage position in the set and cannot tilt the line; it only dragged the intercept from 2 cm down to 0.5 cm. (c) falls from to about and from to about . (d) A large-residual outlier in the lowest-leverage position: influential for the intercept and , not for the slope.
(a) with . (b) With all six, , so and the line is . (c) and . (d) The sixth seedling sits 10 litres from and supplies 100 of the 130 in , so its low growth flattens the line entirely. Report both fits and the reason for the point rather than deleting it.
(a) , with the brightness in percent and the predicted battery life in hours. (b) Each extra percentage point of brightness predicts 0.06 hour less battery life, so 10 points predict about 0.6 hour less. (c) , negative because the slope is negative. (d) Predicted 9.4 hours, so the residual is hours and the model overpredicted by 1.2 hours. (e) The new point has high leverage and a residual of hours, so it pulls the right end of the line up: the slope flattens toward and moves toward , weakening the fit.
Every step is worked at statslearn.org/practice/correlation-residuals-practice. Free to print and copy for classroom use.
The key gives final answers only. The full worked solution for each problem is on the matching practice set, and every other sheet is on the worksheets page.