Alexa gets their rankings from the Alexa Toolbar, which people install into their browsers and it basically tracks their activity around the Internet, the websites they visit, and how often they visit.
This is where they get most of their data from, so they really do have a minute set of data to work with as only a small fraction of people actually have the Alexa toolbar installed (guestimates say around 15-20 million downloads…many of these no longer active), with a good deal of these being people that are Internet marketers who want to track their Alexa rankings.
Along with that, they claim to get data from “other” unknown sources as well (yeah…).
Another concern is the fact that the toolbar is considered to be “adware” by MacAfee while Symantec another major anti-virus program considers the toolbar “trackware” because of the nature of how it works (it tracks and stores all of your web activities). If you have it installed, you can best bet that all of your activities are being tracked and the data you are using it for really is deemed quite useless, because it is not real traffic data.