search

diazrenata / scadsanalysis

Research compendium comparing species abundance distributions to their feasible sets.
MIT License
0 stars 0 forks source link

staircase wit #70

Open diazrenata opened 3 years ago

diazrenata commented 3 years ago

things i have thought of 36 hours post submitting revision 1:

  1. small n/s gives you a small mean dissimilarity even if you have a small community. this is OBVIOUS for gentry but also explains how many small communities in fia have a low dissimilarity score. in general, the dissimilarity score is more well-behaved (probably cause it's got some nice calculations in it) than the breadth indices (which are coarse efforts to describe potentially quite strange distributions).
  2. the 95th percentile of the dissimilarity score among sims directly tells you how big a difference there would need to be between the observed and the actual for it to look 'extreme'. that is, what % of the total n would have to be allocated differently. this is a direct measure of the constraint on statistical power.............................