Arguing that testing using animal species different from the target species is USELESS just because it does not yield definite results misses the point. Nobody said that it WAS "defienite" but the this lack does not make an argument for "useless".

There are both positive and negative results from testing.

If a drug is tested on a few mammalian species and in every case the results are that the drug is toxic the process does not continue to find out if "toxic in humans also?" If it doesn't seem to work in any of those test the process does not proceed to confirm that it also does not work in humans. There is of course a chance that the results in humans WOULD be different but the probablity low.

If a drug is tested on a few mammalian species and is safe and effective, then it still has to be tried in humans. There is that small probablity that in humans it will be unsafe or ineffective -- but in this case since it is proposed to use the drug in humans that must be done. Remember, a 1% chance that a drug safe for other animal species is unsafe in humans is not begligible if we are talkign about 500 proposed new drugs.

Arguements against animal testing of drugs for humans are best made on moral grounds -- that even though this is a statisticly useful process it is wrong. Arguing that the reason we should not use aninal testing "because it doesn't work" are silly bwecuase that is contrary to fact >

Where does your figure of 1% come from?

I was making no particular claim as to what the "unreliable" percentage was. Merely indicating that even a small probability of an event is not negligable when the number of trials is large enough.

But yes, unlike in the social sciences where results get published based upon ridiculously low confidence levels, that's about right for experiments in the biological sciences. In other words, the protocol of the experiment would be set up using enough subjects that there would be about a 99% confidence level.

People have some funny ideas about what conducting an experiment and noting the results means. It does NOT mean that the results aren't purely by chance. It means (depending upon how the experiment was set up) that there is a calculatable probablility of the results meaning what they appear to mean (and of course 1 minus that probability that it's chance). The "usual standards" differ in the various disciplines as I already alluded to.

The point I was trying to make, if we are ONLY 99% sure that the results of an experiment were not pure chance, then of 3,000 published reports, about 30 of them would be expected to be nonsense.

1% is very optimistic and not very realistic! It’s like saying you are ALMOST perfect!

Stop arguing!!