Agreed, but there are some clues that testing was faulty (removing microphone and speakers), admittedly in a poorly written article. It's possible that the bias to prove the theory corect skewed the testing process.

http://en.m.wikipedia.org/wiki/Confirmation_bias