A null hypothesis is the opposite of a hypothesis
and should express all the explanations that are not included in the hypothesis.
In our example, the null hypothesis states that if we remove individuals of one species of mussel
from the experimental areas, the remaining number of mussels will be unchanged
or even decrease in number in comparison with the control areas. This can
be statistically tested when we have counted the number of mussels within
the different control and experimental areas. If we find evidence for an
increase in the average number of mussels within the experimental areas
(where the competitors were removed), then we kan reject the null hypothesis.
We have falsified the null hypothesis and in this way we have
excluded all possible explanations except for those that are expressed in
the hypothesis. When we have obtained evidence from the experiment that
the null hypothesis is incorrect, we can maintain that the hypothesis is
correct. We can then accept the model:
What was predicted in the model did happen. See figure 1.
When we removed the competing
mussels, the other species increased in number. It appears that the two
species of mussel compete with one another - survival rate is less within one
species when there are many of the other.
Keep the null hypothesis
Our experiment can in only one way result
in a different outcome. That is when the evidence from the experiment does not enable
us to reject the null hypothesis. If the average number of one species of
mussel is equal to or less than the other in those areas where competitors
have been removed when compared to the control areas, then we have no evidence
that denounces the null hypothesis. In such a case, we must reject the hypothesis
because we have no evidence that the number of mussels increased as predicted.
Therefore, the model is incorrect. We have now falsified the hypothesis and therefore must
find a new model that can explain why the pattern of mussel distribution
is as seen in the beginning.
10 of 13