After all, it's not like skewing your sample would run any risk of skewing the results, now would it?
"An initial investigation... centred on interviews with 42 gamers... all of whom claimed to have either dealt with real-world problems by reaching for a joystick, or to have seen game graphics in the real world."
"An initial investigation... centred on interviews with 42 randomly selected gamers"
In other news:
"An initial investigation... centred on interviews with 42 confirmed voters... indicated that voting by members of the general public was approaching 100%."
And that's quite apart from the never-ending world championship 'Let's take the #$%^ out of the surveryors', or any question as to whether there was any gain to those interviewed who confirmed they demonstrated what the research was looking for, as opposed to those interviewed who did not.
Heck, even an afternoon off class for further interviews might work.
Please note: I do not intend any direct negative nasturtiums in respect of the researchers.
Never mind. Ignore me. I'm probably having a bad day.