Acute bioassays with control mortality

When substantial mortality occurs in the control group of an acute bioassay, probit analysis and other methods to estimate the LC50 are usually supplemented with Abbott's correction (Abbott, 1925). This paper discusses the probit model with this adaption. Estimates of the LC50 obtained by probit analysis ignoring natural mortality or by probit analysis applied to data corrected with Abbott's formula beforehand are compared with maximum likelihood estimates based on the model. It is demonstrated that ignoring control mortality can lead to biased estimates of the LC50 even at low percentages of control mortality. For interpretation of the LC50 a distinction is made between control mortality caused by natural processes and control mortality caused by experimental manipulation.