post-hoc test after principal coordinate analysis

[Marco777]

Hello,

I'm going to explore my dataset, composed by 100 subjects and 14 binary (1,0) variables.
I was wondering if my protocol makes sense:

1) I obtained a distance matrix from the dataset, by using Hamming distance
2) I performed a principal coordinate analysis on the obtained distance matrix
3) to test for significant differences between groups of subjects, I performed a discriminant analysis, using all the principal coordinates from point 2)

Is that strategy correct? Could I use, instead of principal coordinate analysis, a principal component analysis?

Thanks in advance for any feedback
marco

 

[gianmarco]

Hi Marco,
I am not going to fully answer your question, since I have only a limited knowledge of PCO.
I recall to have read a similar question by a user in the mailing list of the PAST program (you can found it just Googling), which is tailored for biologist, enviromental scientist, and the like.

While you can wait for a reply by a member of this forum (I have two in mind), you could also consider to post a question in that mailing list.

I hope that moderators and other smart guys here will not be afraid if I divert you to another discussion list.
Indeed, I have waited some hours before giving this suggestion, and I gave it because no reply showed up.


Best Luck,
Gm

 

[Marco777]

Hi Gianmarco,
thanks for the tip
best
marco