What kind of variable is your dependent variable? Is it a count of animals, and you don't know how many were in the area to begin with?

Hello Nur,

If your sample is only comprised of rescued animals, then you won't be able to pinpoint features/attributes that distinguish them from non-rescued animals (especially if the vital traits differ by species). At best, you could look for traits that appear in a high fraction of the rescued animals, and propose that these might be salient features for survival via rescue.

Good luck with your work.

Sal Mangiafico, Yes. I do have the count of animals rescued. I have the diversity of the species at the area studied before the flooding, only diversity, not population studies.

David Morse, is there any analysis that could pinpoint the traits that appear in a high fraction among the rescued animals?

Hello again Nur,

The simplest approach would be descriptive. You could enumerate (perhaps in descending order), the most prevalent (as a percent of the rescued sample) features. For example, if the most common feature across species was, "can swim," with 72% of cases (hypothetical data), then you'd likely conclude that this was a potentially salient feature for rescue. If, on the other hand, a given feature appeared at about the same rate as what you'd estimate for all the known species in the biome, then it likely isn't an important feature.

You could run tests on the proportions, assuming that you knew the correct proportions for the full, pre-crisis, population (and could adjust for the relative density of each species within the biome).

Good luck with your work.

I was going to suggest something similar to the suggestion by David Morse . A simple approach is to address each factor separately, as he says, to report e.g. the proportion of animals rescued that can swim was 0.72 and can't swim 0.28. One thing you might consider is to construct confidence intervals for these proportions. Or do a binomial or chi-square goodness-of-fit test to compare the proportions to a 0.50-0.50 split.

The main problem you have in interpreting the results is that you don't know the population in the original condition. So you can imagine that if you had 1 million mice and one elephant in the original area, and you rescue 1000 mice and one elephant, you might concluded that being small and furry leads to an animal being rescued, whereas being giant and wrinkly has the opposite effect. But in reality, you rescued 100% of the elephants, and only 0.1% of the mice. So it's difficult to know what to conclude. I think all you can do is present this difficulty with your results.

I think you are getting ahead of yourself. Your first question should be what design, what methods and measurements, should you use? This depends on what resources you have available, but as David Morse says for many designs there are no statistical methods for the question that you have. Stats ain't magic.