The reason people attempt to overclock is obviously to improve performance, but the reason it's sometimes or mostly possible requires a little explanation, albeit a highly simplified one.
When Intel and AMD make different speed CPUs, they all basically come from the same production process and materials. They then test the CPUs that were produced. For example with Pentium 4, some chips from the same batch will be stable at 1.5 Ghz, some will be stable at 1.6, 1.7, etc. So Intel will then slap a 1.6 label on certain chips and 1.9 on certain chips because those are the speeds they were tested as "stable" at. Intel will tend to be a little conservative with their ratings, so it can be assumed that most chips can actually be pushed just a little further, especially at the low end of a processor family.
For example, AMD's Athlon XP chips are all made from the same production process and materials, even though the speed ratings vary from 1.5 to 1.9. If this is the case, most overclockers assume that some of the 1.5 chips can be pushed, because they came from the same batch as the 1.9 but were labeled 1.5 due to testing results.
With this knowledge in hand, you can save some dough (if you're brave) by pushing a slower chip. If you have the right cooling in place, it can be done.
Again, this is a very simplified explanation.