Percentages are close to meaningless

I’ve mentioned in the past my great frustrations with the use of percentages in news stories. An article will say “X Caused a 50% drop in robberies” but give no indication of the actual amount. Were there four less robberies? 400? Who knows?

This following Vox article is frustrating on many levels.

Australia confiscated 650,000 guns. Murders and suicides plummeted.

The article makes clear that without a doubt deaths went down after Australia initiated a buyback program to confiscate certain types of firearms. But the exact number is distressingly vague. I’ve read through it and I have no idea what the actual number is as everything is discussed in percentages. Then there’s this.

One study concluded that buying back 3,500 guns per 100,000 people correlated with up to a 50 percent drop in firearm homicides. But as Dylan Matthews points out, the results were not statistically significant because Australia has a pretty low number of murders already.

Bottom line: Australia’s gun buyback saved lives, probably by reducing homicides and almost certainly by reducing suicides.

So in the headline we’re told that murders “plummeted.” In the article we’re told that the amount wasn’t statistically significant and that the amount of homicides “probably” reduced. Not quite the same thing.

  1. No Comments