I hate to be that guy, but your first statement is technically incorrect: you can have discrete alternate conformations superimposed in the electron density (usually 2-3 is the most that can be resolved), and you can have different conformations of multiple copies of the molecule. (I've personally worked with both, although the differences in the second case were small.) That's not even counting ensemble-based approaches to modeling the crystal structure, although that's arguably just a different way of representing the uncertainty.

All that aside, crystallization certainly biases it towards specific conformations, which single-particle EM does not.