Oh, it's all faith X, but that is just my opinion of course. Evidence, and whether it is solid or not, is a relative thing in my mind. I've never come across perfect data, and data is my science. Well, perfect data exists in theory, but not really in practice, which is why statistical inference is about attempting to quantify probabilities and "error".

For example, whenever a scientist claims a result is statistically significant, there is a small but usually non-negligible probability that the reality is in fact the opposite. Commonly that probability is set at 1 in 20, but it is arbitrary. And that is assuming all the assumptions required to use such a statistical test are satisfied. And that the data were collected correctly, and that the data is measuring what is intended to be measured etc.

James.