Science is about objective analysis. The closest a good scientist can come to bias is in writing their hypothesis (which is based on logic, not predisposition) or the more likely cause of who is writing their paychecks. Of course if they fudge some data and what they were working on kills a couple people when they cleared it, Mr. Paycheck-Writer isn't going to be very happy.
I bring this up for a number of reasons, chiefly that objective analysis in this thread was tossed out the window somewhere between posts one, three and nine. When the Conchis wanted us to extrapolate Obama's thoughts, championed animal rights with a large list of facts, then told us to stop discussing what he derailed the thread into and get back to the original topic.
I've read that maraschino cherries cause cancer in rats and while I'm sure there is enough red #4 or whatever in them to fry the rat's tiny little brain, the human body is much more resilient against that diabolical nonvegetable-based dye. A good scientist will recognize this and likely design another experiment based off the results of the previous one. Repeating the process dozens or hundreds of times until the data is satisfactory.
Animals make excellent test subjects for a number of reasons. Unlike humans the noble rat does not enter a study as a test subject with a bias, they do not lie about their symptoms, they do not have higher thresholds for pain that vary wildly from human to human, and they won't try to shun or please anyone. Rat diets, environments, activities, and just about every aspect of their life can be controlled, where as humans doing tests are going to want to go back to their families at night, keep eating their diet laced in saturated fat and sugar even if they've been instructed not to, they are going to get lazy or forgetful, they will get worried and stay up all night worrying instead of getting healthy amounts of sleep. They are going to get different amounts of exercise and fresh air, the number of differences between two human lives are staggering.
Often times you want a large sample size from many variations of the human body, and that is certainly fine. But if you want something very specific you almost have to turn to animals.
In another decade when computers are 100 times more powerful than they are now we'll have the processing capacity to work out a lot of questions without moving away from a keyboard.