Is it safe to pool the blank control data with the solvent control data?

摘要

Ecotoxicological studies, where the test chemical is dosed with a solvent, are often performed using two controls, a blank control (BC) and a solvent control (SC). In many guidelines, it is recommended to combine the data from the BC and the SC if they do not differ significantly and otherwise only use the data from the SC for the significance test on the effects of the chemical treatment. In this paper the validity of the conditional testing scheme is investigated by simulating data sets with and without difference between BC and SC. These simulations show that this testing scheme does not lead to a valid statistical test.