2
$\begingroup$

I've sampled 2 sites for wasps. At each site, I used 20 traps to catch wasps. I sampled each site once a month for 3 months:

Site 1 = 20 samples + 20 samples + 20 samples

Site 2 = 20 samples + 20 samples + 20 samples

So sampling effort and sample size are identical for each site. The traps at each site were setup in the same spots each time. But because I had 20 available, they were not set up simultaneously (one lot were set up for a week, then moved to the other site for a week).

I'm now wanting to construct MDS plots to look at how the wasp species, sites, plant species, trap placement, climate data etc are all related.

My question: When constructing a (dis)similarity table for MDS analysis, do I have to have 120 samples, or can I pool the site data together to just have 40 samples (20 from each site).

Put another way, do I need to split my data up into 3 chunks relating to sampling period or can I pool the data from each period?

$\endgroup$
  • 1
    $\begingroup$ You can decide to do it various ways. 1) Combine all the data in one distance matrix and do MDS of it (single-source = classic or simple MDS); 2) Do Replicated MDS; this model assumes individual sources (distance matrices) each is a noised copy of the hypothetical prototypical one; 3) or do Weighted aka INDSCAL model MDS which is further "individualistic" - it assumes each individual distance matrix have different accents (weights, importance) of the same dimensions implied by all the sources. $\endgroup$ – ttnphns Oct 11 '16 at 10:02
2
$\begingroup$

Got some stats help from a consultant and thought I'd post here in case it helps someone. Essentially, their advice was that it's perfectly OK to sum together the sampling periods (in my case). They said that it can be useful to do the analysis several different ways and then think about what is being shown and what is most useful to your question. I'll be doing quite a few MDS plots in different configurations, although I think that pooling together into the 40 samples will probably be the most useful.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.