This is a very interesting study.
I agree how they decided to test with real people, but I don't agree that it's a simple choice of one navigation method over another.
I am a firm believe that the content and the information architecture will dictate what type of navigation method one will use.
Of course there are some navigation methods that are 'inherently' frustrating, like Tam mention the flyout menus are just not a good choice.
Personally I do like the "Yahoo" / Directory type of navigation but I'm not sure it's well implemented here.
While working on redesinging the City of San Diego
's website we moved from an older static menu system to the "Yahoo" / Directory type of navigation.
It was perfect to bring most of the important content to the surface; however, we soon found out that we wanted to bring more and more to the front. This lead to large unwieldy text.
In addition, some citizens would be confused and didn't realized that there was more topics than what was just listed under each of the headings.
Granted not many sites have to work with 60+ departments and over 6000+ pages.
Probably the Directory style was the best solution, but it was not without it's drawbacks.
The results: we had to implement a secondary navigation at the top that, although broke the rule of duplicating navigation, was a necessity. The main navigation lead to pages that gave full path navigation to each topic / department.
As far as Intranets go:
I've always been trained that they don't need to act like websites that we normally surf on the web.
Really they should be designed almost like an application where one has all the tools at hand to get their job done faster and more efficiently.
The intranets that I've help design tend toward the boxy modular design. Something similar to the latest My Yahoo pages.
Where boxes of content can be added and removed at the users whim.