Multi-Focus on Practices 3 and 4

Analyze a regression model: AP Stats FRQ practice

3 versions of one AP Statistics free-response question, each with a 10-point rubric and a model answer for every point. Write your response on paper before you open the rubric.

These are StatsLearn originals written in the AP style, not released exam questions. Topics: 5.2 Correlation, 5.3 Linear Regression Models, 5.4 Residuals.

Version 1

A community center models weekly attendance y from the number x of classes offered. The least-squares line is y-hat = 14 + 2.1x, with r = 0.72. The observed x-values run from 6 to 24. One week with x = 8 has attendance 34.3.

  1. (a) Identify the variables and interpret the slope.
  2. (b) Compute and interpret the prediction and residual for the stated week.
  3. (c) Compute and interpret r-squared.
  4. (d) Evaluate a prediction at x = 36 and address causation.
Show the rubric and model answers

Part (a): 2 points

  • Identifies the explanatory and response variables. Classes offered is the explanatory variable x. Weekly attendance is the response variable y.
  • Interprets the slope in context. For each additional class offered, predicted weekly attendance increases by 2.1 people, on average.

Part (b): 3 points

  • Computes the predicted attendance. Predicted attendance is 14 + 2.1(8) = 30.8.
  • Computes the residual with the correct sign. Residual = observed - predicted = 34.3 - 30.8 = 3.5.
  • Interprets the residual's direction. The observed attendance is above the model's prediction.

Part (c): 2 points

  • Computes r-squared. r-squared = (0.72)^2 = 0.5184.
  • Interprets r-squared in context. About 51.8% of the variation in weekly attendance is explained by its linear relationship with classes offered.

Part (d): 3 points

  • Identifies extrapolation. Predicting at 36 classes is extrapolation because the observed maximum is 24.
  • Explains why extrapolation is risky. The linear pattern observed inside the data range may not continue that far beyond it.
  • Avoids a causal claim from regression alone. The regression shows an association. Without random assignment, it does not show that adding classes causes attendance to rise.

Version 2

A community center models weekly attendance y from the number x of classes offered. The least-squares line is y-hat = 18 + 2.5x, with r = 0.80. The observed x-values run from 6 to 28. One week with x = 12 has attendance 51.5.

  1. (a) Identify the variables and interpret the slope.
  2. (b) Compute and interpret the prediction and residual for the stated week.
  3. (c) Compute and interpret r-squared.
  4. (d) Evaluate a prediction at x = 40 and address causation.
Show the rubric and model answers

Part (a): 2 points

  • Identifies the explanatory and response variables. Classes offered is the explanatory variable x. Weekly attendance is the response variable y.
  • Interprets the slope in context. For each additional class offered, predicted weekly attendance increases by 2.5 people, on average.

Part (b): 3 points

  • Computes the predicted attendance. Predicted attendance is 18 + 2.5(12) = 48.0.
  • Computes the residual with the correct sign. Residual = observed - predicted = 51.5 - 48.0 = 3.5.
  • Interprets the residual's direction. The observed attendance is above the model's prediction.

Part (c): 2 points

  • Computes r-squared. r-squared = (0.80)^2 = 0.6400.
  • Interprets r-squared in context. About 64.0% of the variation in weekly attendance is explained by its linear relationship with classes offered.

Part (d): 3 points

  • Identifies extrapolation. Predicting at 40 classes is extrapolation because the observed maximum is 28.
  • Explains why extrapolation is risky. The linear pattern observed inside the data range may not continue that far beyond it.
  • Avoids a causal claim from regression alone. The regression shows an association. Without random assignment, it does not show that adding classes causes attendance to rise.

Version 3

A community center models weekly attendance y from the number x of classes offered. The least-squares line is y-hat = 22 + 2.9x, with r = 0.88. The observed x-values run from 6 to 32. One week with x = 16 has attendance 71.9.

  1. (a) Identify the variables and interpret the slope.
  2. (b) Compute and interpret the prediction and residual for the stated week.
  3. (c) Compute and interpret r-squared.
  4. (d) Evaluate a prediction at x = 44 and address causation.
Show the rubric and model answers

Part (a): 2 points

  • Identifies the explanatory and response variables. Classes offered is the explanatory variable x. Weekly attendance is the response variable y.
  • Interprets the slope in context. For each additional class offered, predicted weekly attendance increases by 2.9 people, on average.

Part (b): 3 points

  • Computes the predicted attendance. Predicted attendance is 22 + 2.9(16) = 68.4.
  • Computes the residual with the correct sign. Residual = observed - predicted = 71.9 - 68.4 = 3.5.
  • Interprets the residual's direction. The observed attendance is above the model's prediction.

Part (c): 2 points

  • Computes r-squared. r-squared = (0.88)^2 = 0.7744.
  • Interprets r-squared in context. About 77.4% of the variation in weekly attendance is explained by its linear relationship with classes offered.

Part (d): 3 points

  • Identifies extrapolation. Predicting at 44 classes is extrapolation because the observed maximum is 32.
  • Explains why extrapolation is risky. The linear pattern observed inside the data range may not continue that far beyond it.
  • Avoids a causal claim from regression alone. The regression shows an association. Without random assignment, it does not show that adding classes causes attendance to rise.

Keep going