Writing hypotheses

[Seiryuu]

So a few weeks ago my group and I submitted two hypotheses for an experiment with regards to Artemia sp. and the effects of varying intensities of light. From the examples I read in my manual, my group submitted this:

 

H₀: Light intensity will not have an effect on Artemia sp.

H_a: Light intensity will have an effect on Artemia sp.

 

Now I'm sure that that's right, but my question is, exactly how broad/specific should hypotheses like these be? It just seems to broad for me to be considered a hypothesis.

[CharonY]

You can be more specific by framing the null in terms of the parameters being measured. E.g. instead of "effect" you may want to posit that whatever being measured will not be influenced by light intensity (e.g. phototaxis).

[Seiryuu]

You can be more specific by framing the null in terms of the parameters being measured. E.g. instead of "effect" you may want to posit that whatever being measured will not be influenced by light intensity (e.g. phototaxis).

 

Thanks! I'll keep that in mind next time! But should it just be the null? Wouldn't I have to do it for both?

[CharonY]

Thanks! I'll keep that in mind next time! But should it just be the null? Wouldn't I have to do it for both?

 

Yes, you do. Though it depends on the type of test you are interested and can be rather tricky. For instance, the alternative could be that the phototactic is stronger, weaker (i.e. one-tailed), or either stronger or weaker. However, in case of Fisher's approach an alternative is not even used to formulate the test. Here, the experimental data is tested against the probability that it arose from chance alone, assuming H0 is valid.

 

However, in the ideal case the H0 and H1 are formulated in a way that makes it possible to accept H1 by rejecting H0. Though depending on the experiment there is also the chance that you reject H0 but cannot decide on H1.

 

Also note that you do not accept the H0. You either reject it, or fail to reject it. In the latter case it just means that H0 may be a possibility (but so are others).