4 ms·

Perhaps the paper the graph is taken from would be helpful? https://mason.gmu.edu/~atabarro/BriggsTabarrokFirearmsSuicide.pdf https://mason.gmu.edu/~atabarro/Br

Perhaps the paper the graph is taken from would be helpful? https://mason.gmu.edu/~atabarro/BriggsTabarrokFirearmsSuicide.pdf https://mason.gmu.edu/~atabarro/BriggsTabarrokFirearmsSuicid... They go through several different models and their findings that gun ownership is associated with suicide in the US seem rather robust. There is a lot more to suicide of course: divorce, depression, alcohol use and more.

roenxi 6y ago
> their findings that gun ownership is associated with suicide in the US seem rather robust. Robust is a strong word for that finding. I haven't read the paper extremely closely but page 13 suggests they have a 16 variable linear factor regression for 150 observations. With that sort of model a small but statistically detectable correlation isn't the sort of result that ends arguments. Realistically it should be discarded out of hand as a evidence for any sort of action; the effect is too small and there are too many ways that result could be misleading for both practical and statistical reasons.