Reading Diagnostics and Choosing the Right Fix

You regress apartment rent on square footage, floor number and distance to the nearest metro station using 3,000 listings. The output shows R² = 0.71, all three coefficients significant at p < 0.001, and Durbin-Watson = 1.98. But the diagnostic plots show:

• Residuals vs fitted: a clear fan shape — spread grows with fitted rent.
• Residuals vs square footage: a U shape (negative in the middle, positive at both ends).
• QQ plot: heavy right tail.

1. Map each observation to the assumption it questions and state whether it threatens the coefficient estimates or the standard errors.
2. Propose a specific remedy for each problem and say which you would apply first and why.
3. Your manager wants to keep the original specification for interpretability but asks "are the p-values still trustworthy?" Give a precise answer.