When testing for

statistical significance, there are two possibilities--the data shows something (the

variation is statistically significant), or it doesn't (the variation is statistically insignificant). The null hypothesis states that the variation is due solely to

chance error or

random variation, and is not

statistically significant.

Hence, if we are looking at 60 rolls of a die:

Spots | Count
-------+-------
1 | 12
2 | 8
3 | 7
4 | 9
5 | 10
6 | 14

Under the null hypothesis, the variation is due to chance, and the expected values would be 10, and the variation is

insignificant or a

fluke.

The opposite of the null hypothesis is the alternative hypothesis, which states that the variation is statistically significant and outweighs chance.