re: #3 Charles Johnson

The FBI’s data has the same flaws the Mother Jones data does: it is based primarily on news reports. The FBI does not classify data as “active shootings” or non-active shootings. In fact, the term itself is of relatively modern vintage.

The article is full of other bad analysis. For example:

A run of nine points or more below the orange average line is considered a statistical signal that the underlying process has changed. (A nine-point run below the average is about as likely to occur by chance as flipping a coin nine times and getting heads nine times in a row—the probability is less than 1 percent. The 14-point run we see here is even more unlikely to have occurred by chance.)

That would be true if the data were a normal distribution but it’s not. The majority of their data points are below average. The sample is highly skewed by a few years in which there were very few incidents. In fact, there is another run of nine data points right before the period they claim has the sharp rise. There is a run of eight out of nine around the turn of the century. There are other smaller runs. The sample they use is way too small and sparse for the technique they are using (or really any technique).

I don’t see any testing of the null hypothesis here.