Consider several attributive series like that: {species1: number1; species2: number2; ... ; speciesN: numberN}. Each series contains an information about the species composition of some territory.
What statistical methods can I use to detect differences (similarity) between several series (more than 2, much more...)?
I used chi^2-test for 2 series, but when number of series is growing, error of comparison is also growing due to multiple pairwise comparisons.
I understand that you have a large number of species, and
you have composition data of territories.
As you did not highlight any species of interest I assume you want to see the differences between composition patterns.
You could try using dimensionalty reduction methods such as principal component analysis, factor analysis or multidimensional scaling to reduce each composition series to a few characteristic numbers, and then analyse these numbers.
Alternatively you could get an overview of the attribute series if you run a cluster analysis on them.
Then how to compare the territories depends on what attributes (other than the species composition) you want to take into account.
Thank you very much. I thought about using these methods. Simply I was doubtful of availability of any reliability criteria. Sorry for my English
Добро пожаловать.
Если вам нравится мой ответ был бы вы +набрали это? ;-)
спасибо
(translate.google.com)
If you mean I would vote up your answer... I did it right now ;) Or do you mean anything else?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics