fom the comments
n. • 4 hours ago
Hey, this is interesting. Right from the above article.. John Donohue says, “ All this work is based on statistical models,” “When the models all generate similar estimates, it increases your confidence that you have captured the true effect”
Isn’t statistical modeling what climate science is based on as well? And we’re expected to just accept it at face value.
Maybe the one of the questions we should be asking is when do we accept statistical modeling as fact and unquestionable ,,,or not?
The only time the output of a statistical model is anything but a hypothesis is when the output of the model can be verified by real or experimental data. A lot of the analysis and modeling done by politically oriented individuals like the authors of the cited article is intended to support their viewpoint, and is not designed to be tested with real data.
As a statistician (and yes, I checked this professor's credentials, and mine are at the same prestige level but from schools ranked higher for applied math, statistics, and modeling), I am not impressed with his work. He's a lawyer and an economist from good schools, except that he's not in fields that emphasize a study of modeling. Models tell you only what the assumptions they are based on predict.
Lawyers and economists are particularly known for tweaking the assumptions until they get the desired result instead of searching for the truth. I would own firearms regardless of what the models said about general effects. However, my best analysis of firearms effects strongly suggests that gun owners in any given neighborhood are statistically safer than not gun owners in the same area, from the same demographics, income, and family structure. Similarly, gun owners are more likely to commit gun suicides than non gun owners, but they are much less likely (with the same caveats on income and demographics) to commit suicide overall.