RANKING RELATIVE RISKS

Because the linearized multistage models q₁* upper-bound potency measure is very nonresponsive to the information on the shape of the dose-response relationship in observed data, the conservative q₁^* values make it difficult to rank the relative risks for different chemicals.  Chemicals that cause  relatively few if any responses for a wide range of low dose values may have q₁^* values that are very close to q₁^* values of chemicals that cause a much larger steady increase in the number of responses over the same range of low dose values.

As illustrated in Figure 8, the multistage models maximum likelihood estimates are more responsive to the observed data and hence have greater capability to differentiate between the shapes of the dose-response relationships of different substances than the  regulatory upper-bound potency measures (like q₁*) based on the linearized multistage model.   Thus, the multistage models maximum likelihood estimates of risk are better at ranking relative risks than the linearized multistage models upper-bound potencies.

>> Figure 8