Are statistics worthless in examining social issues, or is there something to them?

So…“there are lies, damn lies, and statistics.” I’m all for statistics when we’re talking about the hard sciences. However, the more I see them used, and the more I want to use them, to show how some policy does some thing that helps a social issue, or that it does harm…the less I feel like they make sense.

Statistics are often used by people advocating for either side of a complex social issue to show that there is evidence of a causal link between, lets say, gay marriage and increased tax revenue; or gun ownership and a reduction in crime. I personally was always committed to the fact that states with the death penalty were actually those with the higher crime rates than abolitionist states.

Also, we are also given numbers referred to statistics that are not – people using percentages to show causality or something less drastic. Of course, correlation is not causality, and if you don’t know what the sampling procedure was, or how certain variables were controlled for, you’re left with a sense that something’s been proven when it really hasn’t.

So, can we or should we use statistics in our arguments? If so, what kind and how? How can we introduce them in a responsible manner? And should we be outraged at news outlets and reporters who reference numbers as proof in an deceptive way? And do any of them ever really do anything but?