I don't know about you, but I'm rather sick of the way that the word "fact" is twisted. I've already forgotten quite a bit of the statistics class that I took last semester, but one thing will stick with me for the rest of my life: Correlation does not equal causation.
For instance, you could randomly sample thousands of peanut butter sandwiches made in thousands of kitchens every day, and find a high correlation between peanut butter and jelly (together) as sandwich ingredients. Does this mean that peanut butter causes jelly? Of course not. I don't think that you will find anyone even remotely reasonable who would think that. But swap two other correlated things in for the PB and the J, and loads of people will immediately seize on the "evidence" of those two things being found in tandem and immediately connect the dots between cause and effect. Particularly if the cause-and-effect relationship involves losing weight, making money, or one-upping their ideological foes. Then, suddenly, peanut butter causing jelly is a "fact."
Me, I think that the world would be a lot saner if we were more stingy in awarding the term "fact." Personally, I can keep the process decision-making much less cluttered by categorizing nearly everything as either "data" or "opinion."
Contrary to the popular catchphrase, the data do not (typically) speak for themselves. Data must be collected in a manner as unbiased as possible; it must be compiled in a manner as unbiased as possible; it must be contextualized fairly. (And it's all too easy/tempting to screw any one of those steps up.) And, in the end, someone has to put some (or all) of their credibility on the line by drawing a conclusion from both compilation and context--all the while trusting that the data is not already obsolete. And at that point, it becomes an opinion--an opinion, mind you, that can run the gamut from empirically-informed to out-of-the-ear ignorant, but an opinion nevertheless.
Yes, I realize that facts are safer, at least for one's peace of mind. If you cherry pick your facts carefully enough, you can weather all manner and magnitude of contrary data within the fortress of your own correctness. Personally, I pity anyone like that. Because facts aren't eternally true; more than a few never were.
I'm not saying that you have to review every last data point yourself before making a decision. (Unless, of course, such superciliousness is in your job description.) Neither is projecting your biases on someone else's conclusion a valid response. But I think that it's a good mental exercise to minimize the "facts" that you operate with. And be very, very suspicious of those that just happen to support the way that you want the world to be.
