So I was out in the field yesterday shooting, and I was messing around with both incident and reflective metering. My understanding is that, if I point the meter in reflective mode at an 18% grey card, and then take an incident reading in the same lighting, I should get the same reading.

However, it just doesn't seem to work that way: almost invariably, when I feel like I'm pointing the reflective meter at an "average" scene, putting the dome on in the same lighting drops the reading (i.e., calls for more exposure) by one to two stops.

I can think of three explanations:
1) my estimate of "average" is one to two stops off;
2) something is peculiar about my meter, causing either me to operate it wrong or it to read wrong; or
3) I don't understand how incident metering works after all.

The meter is a Luna-Pro SBC and I'm pretty sure it's in good repair---it's served me admirably as a reflective meter. So I think something is wrong with me, be it in my sense of what makes a "correct" reflective reading (#1), operating my meter (#2), or understanding the whole point (#3). Help?