I've thought about this myself. You would probably need a larger sample though. And if the difference is not statistically significant, a full referendum may be

I've thought about this myself. You would probably need a larger sample though. And if the difference is not statistically significant, a full referendum may be called for. The random selection also does away with a lot of lobbying-related problems.