[janzer]

Let's make up some numbers, say 98% commit a single murder and 2% commit 20 leaving an average of 1.38. If the new system stops every murderer after the first one, the number of murders have a 27.5% reduction. Actual reduction quoted in the article is 12.8%.

So it could very well be true that the average number of murders committed is close to 1 and at the same time the reduction is wholly accounted for by stopping multiple murders.

I still doubt that either cause speculated on in the article are the actual full reason, or even possibly the primary reasons, but it would at least be possible with the information provided.