They may be pretty rare in the universe, but they're very common in the planets we've discovered so far because they're really easy to spot with the wobble method.
Yep. The 'wobble' method has a selection bias for large planets with short orbital periods, hence all the hot jupiter's being discovered. The transit method suffers from this bias as well, but not nearly as severe with regard to planet size. The third option is direct imaging, which selects for planets with longer orbital periods (thus having larger angular separation from their star). Unfortunately though, direct imaging is really friggin hard to do and I'm only aware of a handful of discoveries being made this way.
The latest data from Kepler (transit method) shows that planets Neptune-sized and smaller make up the majority of detections, and indeed the frequency of planets appears to decrease with increasing planet size, following the function of 1/R
2. But most of these detections have rather short orbital periods, as expected from the method being used.