Course contentsShow
← Data Science
Lesson 567 of 2,14515. Regression DiagnosticsFree lesson

Common Q-Q Plot Patterns: Heavy Tails and Light Tails

Identify S-shaped curves in Q-Q plots that indicate heavy-tailed or light-tailed residual distributions.

Common Q-Q Plot Patterns: Heavy Tails and Light Tails

You'll learn to recognize S-shaped curves in Q-Q plots that reveal whether your residuals have heavier or lighter tails than a normal distribution.

Understanding Tail Behavior

When you examine a Q-Q plot, deviations from the diagonal reference line tell you how your residuals differ from normality. Two common patterns involve heavy tails and light tails, both creating characteristic S-shaped curves.

Heavy Tails (Heavy-Tailed Distribution)

Pattern: Points curve above the line at the upper-right end and below the line at the lower-left end—like a gentle S-curve.

What it means: Your residuals have more extreme values (outliers) than a normal distribution would predict. The distribution has "fat tails."

Think of it like: A classroom where most students score around 75%, but you have more students with very low (20%) and very high (95%) scores than expected. The extremes are more common.

Light Tails (Light-Tailed Distribution)

Pattern: Points curve below the line at the upper-right end and above the line at the lower-left end—like a backward S-curve.

What it means: Your residuals have fewer extreme values than normal. The distribution is more tightly clustered around the center with "thin tails."

Think of it like: A classroom where all scores cluster between 65-85%, with almost no extreme highs or lows. Surprises are rare.

Why This Matters

Heavy tails suggest your model occasionally makes much larger errors than expected, which might affect prediction intervals and hypothesis tests. Light tails are generally less concerning but indicate your data is more uniform than the normal assumption.

Key Takeaway: An S-shaped Q-Q plot indicates heavy tails (more outliers), while a backward-S indicates light tails (fewer extremes)—both suggest departures from normality that may require attention.