Seems like the data shows a pretty clear correlation between gun deaths, per capita, and gun ownership, per household. Seems like those are reasonable metrics, and their data sources (which they cite) are reasonable. From MotheJones, full article here.