Different definitions of sex sound like a plausible way of fudging the numbers (I don't see any other ways of fudging the numbers that aren't outright lying), and fudging the numbers seems like a good motivation to have different definitions of sex.
The lie detector in the OP makes outright lying seem a bit more plausible though, if it's true.
From The New York Times:
So how sketchy is the research on human sexual behavior, anyway?