If the data are truly sampled from a Gaussian distribution, the QQ plot will be linear.
If the X and Y values are comparable (as they are in Figure 1), then the points are
expected to line up on the line of identity (shown in a dotted line in Figure 1).
Systematic deviation from this ideal is evidence that the data are not sampled from a
Gaussian distribution. But it is hard to say how much deviation is more than you’d
expect to see by chance. It takes some experience to interpret a QQ plot.
The data in Figure 1 were sampled (by computer) from a Gaussian distribution with
a mean equal to 7.0 and a SD equal to 2.0. I repeated this simulation 20 times (with
different random values). Most of those simulations created graphs very similar to
Figure 1. Figure 2 below shows the four (of twenty) graphs that seem to deviate the
most from linearity.
Figure 2
0 5 10 15
0
5
10
15
Predicted Data
Actual Data
0 5 10 15
-5
0
5
10
15
Predicted Data
Actual Data
0 5 10 15
0
5
10
15