In this case, I think you're right, it's a small sample. More generally, the statistical power of a test [1] relates the sample size with the magnitude of the effect to test.
In other words, subtle effects need a huge population to distinguish from noise, while huge effects will show even with small populations. All this is very sensitive to bias due to selected population as always.
All in all, I agree, more research needed. Anyone knows more research about this, with bigger samples?