I think some of this stems from an oft-cited statistic showing that households with guns are more likely to be involved in a gun-related crime. The problem there is causality. The statistic might be true, but it doesn't mean that owning a gun is dangerous. For example, violent felons are probably more likely to carry guns than the law-abiding population, and thus any statistic that distinguishes only between gun owners and non-gun owners will be skewed. Furthermore, some people arm themselves because they are in some sort of danger, and this further skews the statistics.