The degree of statistical significance for a result is based on the probability of observing it by random variation alone. This probability is called the p-value for the result. For example, a difference is statistically significant if it is much larger than can easily be accounted for by random variation. It means that the underlying parameters are likely to be different.
By default, SAS/LAB software interprets a p-value as follows:
| p > 3/10 : | No | evidence of a real difference |
| 3/10 > p > 1/10 : | Not much | " |
| 1/10 > p > 1/20 : | Weak | " |
| 1/20 > p > 1/100: | Appreciable | " |
| 1/100 > p : | Strong | " |