Re: 12 ppl
You can't. But this was probably the kind of study where participants tried all setups (normal book, kindle, probably no reading for control. I haven't read the research so I'm guessing here... But that research setup is called Repeated Measures) which increases the power quite a lot, and 12 subjects can give a surprising power if the study is correctly designed.
Of course you are right in that 12 people probably don't represent the general population, and even worse, were probably all college students. So, generalisability may be limited, but the small, quick study may be a really good start! (it would be crazy to start with hundreds of people. Expensive and unwieldy).
Biting the hand that feeds IT
