Is it legit to compute the mean of log values?

Question

I have two vectors where each element represents a value log2 transformed.

v1 = c(1.4, 2.1, 1.9)
v2 = c(-1.2, -2.2, -1.9)

I'd like to compute the mean of v1 and v2 to show that v1 has higher average abundance than v2. Can I simply compute the mean of each vector to accomplish this? mean(v1) would give 1.8 and mean(v2) give -1.77

Regards,

Answer 1

Taking the mean of a set of logarithms is a kind of average, so in some sense "legit".

In fact, as @gnasher729 comments,

The mean of logarithms is the logarithm of the geometric mean of the original values. So you are comparing the geometric means of your original values.

Whether that can serve to measure "average abundance" depends on what "abundance" means in context and what conclusions you want to draw from a higher average abundance.

So the answer to the question is "it depends".