My understanding was that many of the claims of threshold effects at least partially stemmed from the imprecision of many tests, even at the 1.5-2 sigma levels.

Would you be able to send me some more info? Email's in my profile.