Does it cause cancer?
The science of epidemiology asks exactly the right question. It asks: Out of 10,000 people exposed to a chemical, how many of them got cancer? Actually epidemiology ask similar questions about lots of diseases and conditions. How good is it at answering the question? First look here Epidemiology to get warmed up.
[The terms: Cohort; case controls, cross sectional and Odds ratio; Standardize Mortality, relative risk are common and you should know them. Also "mortality" refers to death, while "morbidity" means the condition of being diseased.]
If there is sufficient epidemiological evidence, then you really don't need to kill any rats to decide if a substance is harmful. But:
1. A dose-response relationship must be established in order for causation to be proved by epidemiology. In most studies it is impossible to measure the dose. Dose is estimated, based on recollections and similar situations, but it is never certain. In the industrial setting, people who are less frequently exposed often "notice" chemicals, while people often exposed become "acclimated" and do not notice them.
2. Many cancers of interest are rare. A very large data base is required for statistically meaningful results.
3. It is impossible to rule out confounding factors. Any factor that increases incidence may skew the results. For example, there is a toothpaste, the users of which have very high cancer rates. Why
For a common situation, like the association between smoking an lung cancer, where there are lots of cancers, and people can describe themselves by the amount of smoking they do, epidemiology is quite accurate. For most cases of toxicological interest, it is not too useful.