In case anyone cares: I just finished a 25-page paper in which I showed statistically that the banker has an impact on the results of a game of Deal or No Deal, but the contestants do not. I'll post a PDF or XPS of the paper tomorrow, perhaps. I primarily wasn't testing contestants, so it is possible that my random sampling of contestants was biased and there exists contestants for which there is a statistically significant difference and all of my participants just happened to be very equally matched. I'm a little doubtful, though.