re: #424 Hal_10000

The response to this study and my pointing out of its flaws.

The way you point out the flaws is to take the data, as I did, and run the statistics. If you put that data through a one-way anova, there is a P.048 for the test hypothesis that the post-2011 data is dissimilar to the 1989-2012 data. ACtually, whoops, it’s much lower, I screwed it up, I’ll redo it but it will probably be about 2/3 of that. Sorry, I’m kinda crippled by migraines.

MJ’s analysis, which I skewered when it first came out two years ago, is based on a few dozen hand-picked incidents based entirely on media reports.

Establish why media reports are an insufficient way of tracking mass shootings. Are you contending that there are media reports of fictional mass shootings that are being included, or that actual mass shootings are not being reported on? What is your basis for this contention? When you say ‘hand-picked’, what do you mean—what are the data you’re claiming are excluded?

This analysis, still based on far too little data, is barely statistically significant (

Great. You admit it’s statistically significant. So why are you being weird about it? Could this be noise? Yes. Do we know the chances of that? Yes, it’s less than 5%, unless you can show some systemic problem with the data. And remember, the systemic problem has to show a collection problem that would affect the post-2011 data and not the pre.

But still: go ahead and ‘skewer’ the data. Run the Bayesian analysis. Actually do something, they include their methodology and the raws, so go for it. Just spouting off about possibly problematic areas without even explaining the logic of how they’re problematic is worth fuck-all.