We take peoples word on a lot of things.
Not in the world of science - we take the evidence into consideration.
When I read a scientific paper, it is a very rare instance when I accept everything in it, especially the details of the conclusions, even though they are based on results within the paper. Also, individual experiments are sometimes carried out in a problematic manner (sometimes accidentally, sometimes because of necessity), and thus the data carries some qualification to the trained scientist.
Sometimes I read a paper and arrive at an entirely different set of conclusions than the authors, based on their evidence.
Scientists do not communicate with conclusions, they do so with data - if you've ever been to a scientific seminar you know what I'm talking about.
And these are the same people fooled by the Disney experiment Knight mentioned.
They might be the same people, but in the Disney experiment they are communicating with testimony, and in science they are communicating via evidence.
Thus their flaws in memory-based testimony do not apply to evidence-based conclusions.