If you can't accurately quantify all the concurrent things you're hypothesizing about (like: values as practiced from values as professed) then your "experiments" just show correlation and not causation.

Proving no more than the basic counter-Enlightenment reality that the poor are not (on average) natural noblemen, and need for the sake of their own humanity to be forced to work if they want to eat.

So, just have less than perfect people on average, but keep the rest of the Enlightenment values. I don't see any problem with amending those with what we now know about human nature. (I think if one examines noblemen close enough, one finds a few impure motivations here and there.) This is pretty much the ideology of the folks over at lesswrong.com.