Well, I see that ADSLguide is probably not a good tool for US residents. I hadn't given this much thought until recently, but I clearly see now - and I admit many would say it's obvious - that the results from any speed testing service will be affected by the geograpic proximity between the user's location and the testing site.
Even within the U.S., if I go to dslreports.com or http://www.speakeasy.net/speedtest/ and run tests from different cities, I'll get different results. So, it seems best to consistently test from the same locations to get valid, comparable test results.
Different speed tests use different methods and different filesizes etc so even if two test sites are next door to each other, results can vary dramatically.
In the UK, BTw, who control most of the UK DSL infrastructure, have set up a very useful speed test which eliminates the ISPs systems. Therefore, the user can tell if the problem lies between the exchange and the computer or with the ISPs own infrastructure. The problem is that the test only applies to the IPStream delivery method.
Basically, if you find a "local" test site such as DSLReports that the majority of users have confidence in, stick with it.