I am having trouble deciding what kind of statistical test I need to perform on my data to answer the question I want answered. Perhaps you can help me!
I have measurements of lengths, widths, etc. of different parts of a specimens body, as well as data that reflects shapes. For instance, I have measured the length, width, height, number of spines, and shape of the spines (either convex, concave, straight. These are recorded as Convex = 1, Concave = 2, Straight =3) of what I believe to be three different species within a genus. I have fourteen such measurements for over 300 specimens. Twelve of the variables are measured in cm to the nearest mm, and the other two variables are as outlined above for the spines, in terms of 1, 2, 3. I currently have the data split out into the three different species, but I want to know if there is statistically significant evidence for having three different species in my population... But I don't really know what kind of statistical test could be done to test this. I imagine some form of multivariate statistics will be employed, but I don't know which to pick.
Any help or ideas would be appreciated!
March 19th 2013, 08:06 PM
Re: Which test to use?
If species are different what would you expect to see in the measurements? And if the species were the same could the mean of what you are measuring be different? (for example if male Cod have a length of 20 cm and female cod have a length of 16 cm they are the same species but they are expected to have differences within the species). Would each of the twelve measurements be different or can some be the same? If eleven were different and one was the same are they considered different species?
For when you measure things as either 1, 2 or 3:
Can different species have the same values 1, 2 and 3?
Can the same species have different values of 1, 2 and 3?
You can't perform numerical analysis on bone shape comparing means and such because you randomly chose Convex = 1, Concave = 2, Straight =3. More fundamentally, concave is not twice convex. But you can do some other types of analysis on data like this.