I have 2 groups of unknown number of individuals and I am visually assessing the membership of these individuals to one specific morphological definition of ecotype out of 5 possible ecotypes (A, B, Bs, C and D). I hypothesised that these groups belong to the ecotype A.
I have selected 2 clearly distinctive individuals per group and I am testing their ecotype membership based on 9 morphological traits defined in the literature. I have then visually assessed the individuals and scored each trait per individual from 0 to 10, 0 being the most different from the hypothesised ecotype trait definition and 10 being the closest to it.
What statistical analysis should I use to find out the membership likelihood of each individual to one of the 5 ecotype definitions?
Thanks a lot.
Dear Matteo,
It sounds like you can try either multinomial logistic regression or ordinal logistic regression, depending upon whether the ecotypes are nominal or scale. Alternatively you could use a cluster analysis technique, e.g. K means and set the number of clusters to 5, then compare your clusters with the ecotypes.
However, with either technique you will need a lot more data.
Thank you for your response Jos Feys, however I am a little confused. I start with two groups to assess, so I am assuming I only have two populations. Out of these 2 populations I can only visually assess 2 members per population. I want to know their membership likelihood to one ecotype out of 5 possible ecotypes. I assume their ecotype is A, so I score 0-10 "trait similarity" to the known definition of that same trait for type A. I do this over 9 traits. My question is whether both groups, represented by only two individuals per group, are effectively Type A.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis