In statistics class today, my teacher said "everything in statistics is wrong, but it is the best we can do" (we did probabilitie statistics, but it was a blanket statement concerning all guessing equations). He then proved to us just that by showing an example where the answer was 40% (the real answer) you get by counting. Then, with using the appropriate formula, he got 36.5% for an answer. As a marketing major, I have seen a huge number of examples where these statistics can be made in any kind of numbers you want them to. MPI is using the most biased statistics they can find from a 'reliable' source in order to further their product. They used a study (in unknown conditions) that had radiative heat loss first, probably because that is what most people think of space blankets as helping with. They put VB layer 3rd probably because people consider it uncomfortable to be clammy, even though it is, IMHO the space balnkets 2nd best (just due to the nature of all my other gear that is windproof)warmth retaining feature. I wouldn't put much stock in what they say, especially considering convective heat loss varies at any given moment. They didn't even consider the conductive heat loss required in sleeping situation to warm a sleeping pad or the ground. There is no exact equation that can predict real life radiative heat loss in a backcountry situation. there are too many variables. But, it is a presence, small, but still there, and can be helped by reflective layers.
A few question to ponder concerning the position of reflective layers and loft:
Would your clothes radiate heat if they were warmer than the next layer up?
How much of the heat your clothes radiate would be captured vs. being transfered by the reflective bags radiative heat loss (being wamer than the next layer up)?
Would a 20* sleeping bag radiate heat if it was warmer than the air around it, due to your body heat?
A good something to ponder. I have a boring class to go to now, so I will have plenty of time to think :)