Quite a bit.
Any economist will tell you that polls and stats are irrelevant.
Like the 'fact' that 5 out of every 10 stats are made up including this one.
Truth be told any data set can be manipulated to deliver support for any conclusion. You have your opinion then you massage a data set to reveal the results you want. It's not dishonest, it's just possible.
This paper puts into question the methods of derving support, among other things, for the death penalty. Of particular interest in this paper is the WSJ 2007 article that correlated data with death penalties and # of lives saved.