How does only looking at showdown hands bias your sample?

[poincaraux]

(First of all, I think I must just suck at searching today, because this must have been discussed here before. Please feel free to point me at older threads and tell me how to search better.)

I'm a NL newbie, and I could certainly use some help on my hand-reading skills. It seems like one easy way to start is to datamine a bunch of hands, select the ones that go to showdown, break things up by position and action (limp, miniraise, raise, re-raise, call, X-reraise, X-call, etc.) and look at the distribution of hands. I'll start with pre-flop hands/action. Obviously, I'm going to be mostly restricted to hands that go to showdown. How can I expect that to bias my results?