First of all, people seem to be a bit confused with regards to probability theory.
Lets say (for the sake of argument, not as a matter of fact), that a gereral HD has a 3% failure probability in its first year of operation. With one HD in your PC, you thus have a 3% risk of HD failure during the first year.
Or more commonly expressed:
P(HD failure) = P(failure of disk 1) = 0.03 = 3%
Now, assume you want two identical HD's in a RAID-0 config. Your risk of failure is now:
P(HD failure) = P(failure of disk 1 or failure of disk 2 or both) = 1 - P(no failures) = 1 - 0.97 * 0.97 = 0.0591 = 5.91%
Thus the risk of failure is almost double that of one disk.
Now, with RAID-1, the calculation is thus:
P(HD failure) = P(failure of disk 1 and failure of disk 2) = 0.03 * 0.03 = 0.009 = 0.09%
Now, how's that for a reduction in risk?
Now, this is all statistics of course. And it assumes that all HD failures (drive 1 vs drive 2) are completely unrelated, which is not the case (imagine a PC case which turns out to be poorly ventilated, it will slowly cook both drives).
But regard it as a starting point for judging risks with a bit more solid foundation than mere handwaving.
Second of all, the "risks" are probably not that interesting anyway....its 100% certain, that without backups, given enough time, you will loose data (RAID-1, RAID-5, RAID-6 etc does not protect against user error, programming errors, viruses, breakins etc).
So, go with any solution which suits you, but also invest in a backup solution and use use it wisely.