Questions: Residuals and Goodness of Fit (R²)

5 questions to test your understanding

Score: 0 / 5
Question 1 Multiple Choice

A student calculates R² = 0.81 for a regression of exam score on study hours and concludes that '81% of the data points lie on or near the regression line.' What is wrong with this interpretation?

ANothing — R² = 0.81 does mean that 81% of the data points cluster near the line
BShe should say 81% of points lie exactly on the line, not near it
CR² = 0.81 means the linear model explains 81% of the variation in exam scores, not that 81% of points are close to the line
DR² measures the slope of the regression line, not the closeness of points to it
Question 2 Multiple Choice

After fitting a linear regression, you plot residuals vs. fitted values and see a clear U-shaped (curved) pattern. What is the correct conclusion?

AThe regression is overfitted and should be simplified by removing predictors
BThere are too many outliers pulling the line off course
CThe variance of the residuals is non-constant (heteroscedasticity)
DThe linear model is misspecified — the true relationship between x and y is nonlinear
Question 3 True / False

A high R² value is sufficient evidence that a linear regression model is appropriate for the data.

TTrue
FFalse
Question 4 True / False

In ordinary least squares regression, the residuals always sum to exactly zero.

TTrue
FFalse
Question 5 Short Answer

Explain why a pattern in a residual plot indicates a problem with the model even when R² is high.

Think about your answer, then reveal below.