That's the average from a National Science Foundation report, counting only active semesters (wall-clock time is more like 9 years on average): http://www.nsf.gov/statistics/nsf06319/ [Figs 4-14 and 4-15]. It does include the years of classes, not just the research years, but classes are usually only two full-time years or so (less if you got a master's first).
It's quite possible to do it in less, if things go right, though in the areas of CS I'm familiar with doing it in under 5 is a stretch (maybe if you're doing a pure theory thesis). But often things don't go right. Lots of people end up writing their PhD on their 2nd or 3rd project, not their first, either because it turned out not to interest them or turned out to be fatally flawed. Or they end up having to switch advisors for reasons ranging from personality clashes to the advisor leaving their university.
Things can also go faster or slower depending on how funding goes. The ideal is to be funded on a grant or fellowship to do your own research full-time. Less ideal is being funded on a grant to do something other than your research, often code-monkey implementation for some big DARPA project or something, which means you basically have a separate part-time job; or if you're supported by TAing, that's a different kind of part-time job.
Yeah I understand how it works (I have done postgraduate research) - I didn't know you were including undergraduate/postgraduate studies, I thought you were meaning thesis only!
Err, I wasn't including undergraduate, just graduate. At least at the schools I know about, there isn't really much of a distinction between "thesis" time and PhD time generally. There's quals, but they aren't that important at a lot of places. Mostly if people say "I'm a 4th-year PhD student", it means they entered the program 4 years previously. They probably spent most of the first two years taking classes, though that varies too (I spread my classes over three years to have more time to start research earlier). I'm not even really sure how much time to allocate to my thesis, because there wasn't some official point at which I stopped doing classes and started doing thesis.
Edit: Actually, it just occurred to me you might be in Europe, where the norm is a three-year PhD thesis, in a pretty time-defined program. The PhD program norms are pretty different between Europe and the U.S. overall (I'm in the U.S.). Is that an accurate guess? If so, that'd explain the confusion. =] The U.S. program is fairly ad-hoc: you enter grad school and then there's various things that you can do in various orders, for various durations, depending on the school. Often you don't have to get a masters first either, so you sort of get that along the way, and you might be doing PhD research at the same time as that too.
Good guess. Where I'm from you can start on your phd work in your fifth year of study normally (for some earlier), and it takes 2-3 years after that to complete your project + thesis.
It's quite possible to do it in less, if things go right, though in the areas of CS I'm familiar with doing it in under 5 is a stretch (maybe if you're doing a pure theory thesis). But often things don't go right. Lots of people end up writing their PhD on their 2nd or 3rd project, not their first, either because it turned out not to interest them or turned out to be fatally flawed. Or they end up having to switch advisors for reasons ranging from personality clashes to the advisor leaving their university.
Things can also go faster or slower depending on how funding goes. The ideal is to be funded on a grant or fellowship to do your own research full-time. Less ideal is being funded on a grant to do something other than your research, often code-monkey implementation for some big DARPA project or something, which means you basically have a separate part-time job; or if you're supported by TAing, that's a different kind of part-time job.