. . . [P]ersonal experiences are a terrible way to learn something. As human beings, we are all subject to preconceived notions, personal biases, and differing expectations; and of course any one person's personal sampling of something is, by definition, an uncontrolled, unblinded test subject to external influences and all manner of unknown variables. This is why we all know people who have reached opposite conclusions based on their experiences . . . Instead, we turn to well-designed, blinded trials that employ controls to eliminate bad data caused by all of these weaknesses.
This is from Brian Dunning's latest edition of Skeptoid episode 306 Listener Fedback XX. The episode responds to a number of listener feedback emails regarding the definition of skepticism, the weight and proper context to give unexplained anomalies, how much authority a resource Skeptoid ought to be given, as well as the flaws in knowing something to be true due to personal experience. While the program as a whole is worth the 12 minutes listen the above quote really hit home. I ought to put the above on a sound loop and play it every time I am challenged with 'but seeing is believing.'