Makes one kind of wonder just how much money the DoE wasted on the study. As the dictionary says, Stochastic - proceeding by guesswork. The linear-no-threshhold (LNT) theory has always been a statistical compromise based on inferring information where no data existed.
There have been many discussion over the decades as to the real effects of radiation from dose-rates less than 10 R (or so) acute dose. Figure 1 in Reg Guide 8.29 shows these differences of opinion rather well. The levels that are set by regulation - 5 R/year - are so low that any possible real data is pretty much lost in the statistical variables that come with normal living.
"Risk" has never been "directly proportional" to dose at these low levels. There has been a presumption of proportionality as a means to establish regulations but the risk has always only been judged to increase with increasing exposure up to the point where actual harm can be actually determined . . . ergo, Stochastic and non-Stochastic exposures.