Science cannot peel back the fabric of reality and see if X is staring back at us.
What it can do is tell us whether or not there are observations that support things that might be defined as X, or that are X-like, or that contradict X.
Let's suppose that for this X, there are no observations to contradict X, but observations of X do not exist. In fact, credible observations of X are virtually nonexistent.
Now, not observing X or something that supports or necessitates the existence of X doesn't mean X doesn't exist - in fact, X could very well exist but be outside of our capacity to observe - but it also means that there is a substantial basis to say that they might as well not, or that it is likely that they do not.
The point is not to say, "this is evidence for atheism," it's to say, "it's true that there's not evidence for atheism, but that does not make it unrealistic." This is especially true if one goes by the colloquial meaning of the word, which is usually the soft rejection of religion.
Edited by Vardarac, 15 April 2015 - 10:20 PM.