Q-Q Plot Interpretation
A normal Q-Q plot compares the empirical quantiles of your data against the theoretical quantiles of a normal distribution. If the data are normally distributed, the points lie close to a straight line. The shape of any deviation tells you what kind of non-normality is present.
Straight line ideal
Points falling close to the reference line across the full range indicate the data are approximately normal. Small deviations at the tails are common with small samples and not by themselves evidence against normality; pair the visual check with a formal test such as Shapiro-Wilk for n less than 50 or Anderson-Darling for tail-sensitive cases.
Heavy and light tails
An S-shape, with points curving above the line at the upper end and below the line at the lower end, indicates heavy tails (kurtosis greater than 3). The opposite shape, points below the line at the upper end and above at the lower end, indicates light tails. Heavy tails inflate Type I error in t-tests and ANOVA more than light tails do. A monotonic curve (banana shape) indicates skewness rather than tail thickness; a clear bow above the line at both ends marks right-skew, a bow below indicates left-skew.
Decision after a Q-Q
If the Q-Q plot is approximately straight, proceed with parametric tests. If skewed, try a transformation: log or square-root for right skew, square for left skew, Box-Cox for an optimised power. If the deviation is in the tails only and n is large, parametric tests are usually robust; for small samples switch to a nonparametric alternative (Mann-Whitney, Kruskal-Wallis, Wilcoxon). Document the visual decision; do not rely on the formal test alone.