Questions: Inference in Linear Regression

5 questions to test your understanding

Score: 0 / 5
Question 1 Multiple Choice

A regression output shows a slope estimate with p < 0.001, suggesting a highly significant predictor. However, the residual plot shows a strong funnel pattern — residuals spread much wider at high fitted values than at low ones. What should you conclude?

AThe regression is reliable; a significant p-value overrides any concerns about the residual plot
BThe p-value may be misleading because heteroskedasticity distorts standard errors, making inference invalid
CThe funnel pattern is normal and only affects predictions at the extremes, not inference on the slope
DThe solution is to remove the high-leverage points and refit the model
Question 2 Multiple Choice

In a simple linear regression with one predictor, the t-test for the slope yields a p-value of 0.04. What does the F-test for overall model significance return?

A0.0016 (= 0.04²), because F = t²
B0.04 — the same p-value, because F = t² and the F and t tests are equivalent here
CA different p-value that depends on the residual degrees of freedom
DCannot be determined without knowing the number of observations
Question 3 True / False

The standard error of the slope SE(β̂₁) decreases when the predictor values are more spread out — i.e., when Σ(xᵢ − x̄)² is larger.

TTrue
FFalse
Question 4 True / False

In simple linear regression, the F-test for overall model significance and the t-test for the slope test different null hypotheses, which is why they can give different p-values.

TTrue
FFalse
Question 5 Short Answer

Why does heteroskedasticity (non-constant residual variance) threaten the validity of t-tests and confidence intervals for regression coefficients, even when the slope estimate β̂₁ itself remains unbiased?

Think about your answer, then reveal below.