What is heteroscedasticity?
Updated May 16, 2026
Short answer
It occurs when residual variance is not constant across predictions.
Deep explanation
Linear regression assumes constant variance (homoscedasticity). Violations lead to inefficient estimates and unreliable confidence intervals.
Real-world example
Income prediction models often show higher variance for higher incomes.
Common mistakes
- Ignoring funnel-shaped residual plots.
Follow-up questions
- How to fix heteroscedasticity?
- Does it bias coefficients?