(May 11, 2023 at 10:23 am)Helios Wrote:Quote:Most gun control studies use the following methodology: look at the homicide rates in some city before some gun control law is passed in that city, and look at the homicide rates after that gun control law is passed, and conclude based on which one is larger whether that gun control law is effective or counter-productive. That methodology is seriously flawed: the vast majority of gun control laws only affect the sales of new guns, so they can affect at most around 1% of the total guns in existence. By comparison, homicide rates change by around 6% from one year to the next. So, the signal-to-noise ratio is at most 1:6. If somebody thinks he has found a statistically significant result of some gun control law, he is almost certainly calculating the p-value incorrectly, or not at all.And more desperate prattle to defend a terrible study by a hack.
Sorry, but we need to work with the data we have. A terrible study is still better than a-priori reasoning, and a-priori reasoning is better than intuition.
