The right way would be to measure stats for each model, for each manufacturing year, for a precise amount of time T.
Then if your 360 age is n*T, and RROD probability during amount of time T is p
(for a fixed number of played hours per week, in average climate)
Then, the global probability of RROD will be something like this formula :
(this is for very low p and n=300, but you get the idea)

So if 33% was for 1 year and 16% was for 6 months it's coherent but they never communicate the amount of time (age) of all measured xbox'es... So it's not enough precise to allow exact stat to be known. It's 2 figures anyway, for 1 year old, I guess : MS, you are bad console hardware designer.
That's all.
This post has been edited by openxdkman: Feb 14 2008, 10:24 AM