Q-Q Plot
A Q-Q plot (quantile-quantile plot) is a visualization technique for testing normality.
Contents
Description
Rankits are expected quantiles of a normal distribution. By plotting a set of observations according to a measurement (on the vertical axis) and the corresponding rankit (on the horizontal axis), deviations from normality are visualized and can be inspected for broader trends.
- A straight, diagonal line is indicative of a normally distributed random variable.
- Curves suggest a skew.
- A convex (opens upward) curve suggests a right skew.
- A concave (opens downward) curve suggests a left skew.
An uniformly distributed variable would appear as an S-curve.
- Outliers appear far from the line.
- If the line is diagonal near the center but flattens out above and below, there may be too few observations at the tails of the distribution.
On the other hand, if the line becomes steeper away from the center, there may be too many observations at the tails.
It is particularly common to visualize a model's residuals using Q-Q plots, but this method can be applied to any supposedly normal variable.