ABSTRACT
Modern society demands that government provide protection against possible harm from the results of technological improvement. Very often the process of regulating the introduction of a new substance relies on research which requires careful statistical design and analysis. This chapter describes the statistical analysis of the brain tumor data from long-term feeding studies on aspartame. Bias, or lack of accuracy, could easily arise with data of this type. The likelihood of finding a tumor depends not only on the size and location of the tumor, but also on the number of sections of material examined for each site. It is common practice in the analysis of results, such as those from the aspartame experiments, to compare separately each dosage group receiving a compound being tested for carcinogenicity with control group. The procedure used in the analysis reported was to first test for the possibility of a linear dose effect on tumor incidence rates among rats fed aspartame, not including controls.
