Roughly speaking, we ship our logs out of Moodle, applying some filtering, and dump them in a big data-warehouse for the researchers to user. The aim of the filtering is to reduce the total volume of data, while (hopefully) not losing anything important. Also, we have a lot of our own activities (e.g. ouwiki, forumng) rather than the standard ones.
Anyway, I am basically confirming that your guess is right.