# Presence/absence data as predictor variable

I am working on a study where I have species presence/absence data for 80 independent sites, as well as the soil carbon values for each of the sites. I want to test whether the presence/absence of each species is related to the site's soil carbon value. In other words, which species coincide with site that have high soil carbon, and which species coincide with site that have low soil carbon.

Since I have a binary predictor variable, and a continuous response variable, what would be the best approach for this analysis?

• I think logistic regression is a good start. Apr 27 at 17:54
• @utobi OP is regarding soil carbon as the response or outcome. I would reach for a generalized linear model and expect that logit or log link would be useful depending on whether soil carbon is a proportion or percent or expressed absolutely. Apr 27 at 18:32
• @NickCox, yes that's what the OP wrote. However, from the problem description, I find it more natural to think of soil carbon as a predictor, i.e. the reason for the absence of species, whatever they are, as also confirmed by the second part of gung's answer (+1 by the way). Apr 27 at 19:07

The "best" analysis depends on what your question is and how you are thinking about the relationships. For example, what do you think is a (possible) function of what? You state that presence/absence is a predictor of carbon in the soil (presumably a continuous variable). As stated, determining if a continuous variable differs between two groups would be a $$t$$-test. If you had a bunch of different species that could be present or not, it might be a factorial ANOVA. This is not my field, but I would guess the thinking is something like 'this species generates carbon that then becomes incorporated into the soil (or leaches it out of the soil)'. On the other hand, if you wanted to see if the presence of carbon predicted whether a given species were present, you would probably want logistic regression. In that case, the thinking would be something like 'this species needs carbon to survive, if there is very little in the soil, this species won't be there'.