The **log-normal distribution** often fits right-skewed positive data better than the normal - **Power-law distributions** describe phenomena where extreme values are far more common than the normal model predicts - Assuming normality when the data isn't normal leads to **systematic errors** in pred