One more area: (most) female dating advice. "I want a sweet nice guy" vs "all my ex-es were jerks". Guess who she's attracted to.
"Never" is a strong word, but I think it's a good general rule to stand by.
It's true that the only way to be sure of what people will pay is to see what they pay. But by the time you have an actual product to sell, you will have asked people for advice and feedback about 8 million times. It's very natural at that point to go for 8 million + 1 and say, "What would you pay for this?"
Pricing is one of the few things in business that's traditionally amenable to gathering hard data. Which is why I'm so excited about the Lean Startup approach; there one tries to apply the same data-driven spirit to everything.
You say to yourself: "I have a pretty good idea of how much work and expense it will take me to build this", then you go off and try to sell it as though it was already done or just getting polished for release. Once you have orders (or even better, cash advance!), you (privately) commit to either finish the product or cancel/refund orders if you fail^Whave a great learning experience.
It's dishonest, but it works, and if you structure your offer fairly (don't induce reliance on the vaporware, only induce desire), it's not evil.
Successful established authors do this before they write books. It's a standard technique in publishing.
See also KickStarter.
It's great that you're naturally that level-headed, but I think when you have a crowd of people telling you X, it's easy to forget that they're not customers - let alone that customers will usually tell you that they want to pay less ;)
And everyone's price point is different, of course... Some would never buy, others would spend far more than most people. It's all about finding that sweet spot.