3 strikes, redux

We still don't know what effect New Zealand's 3-strikes law has had on recidivism.

Graeme Edgeler had undertaken a bit of ocular least squares last year; I'd commented on it here. It then looked like the number of second strikes had dropped, although the conclusion was still tentative. It turns out that the data Graeme was using wasn't as comparable across the two time periods as he had thought, and so he retracted his post. I've updated my post as consequence.

In theory, the law should be effective and at lower cost than American alternatives, for the reasons I laid out here. The American version gets rid of marginal deterrence by having the same hefty penalty across broad categories of offending, so you may expect a severity shift in offending among third-strike offences. New Zealand sets things on the third strike to the maximum penalty for the offence, without parole, unless it would be manifestly unjust to impose the penalty without parole, so proportionality across offence categories is maintained. 

Judges so far seem to be undermining Parliament's intent in the legislation. The two second-strike cases that have come up involving murder, where judges are to impose the maximum sentence without parole unless the 'without parole' provision were manifestly unjust, have both had the non-parole period waived as being manifestly unjust; the only third strike offence so far has also used the 'manifestly unjust' provision for waiving of the without-parole condition. Radio New Zealand reports on the three cases here. 

It would not take many more uses of that provision before offenders expected that judges had no intention of following Parliament's intent, and so deterrence would be lost. Maybe each of the three cases so far really are the exceptions that Parliament intended have this kind of exception. But say that the provision were intended for 10% of cases. The odds that the only three to come up all happened to be manifestly unjust are one in a thousand. Is it more likely that we are seeing the one-in-a-thousand sequence, or that the judges have little intention of following Parliament's intent?  

And so any future empirical study might expect a rise in strikeable offences from 2017 onwards, as judicial credibility diminished - barring changes in sentencing practice that change expectations.

UPDATE: In comments below, Andrew Geddis notes it's been five second-strike murders, and that the judges deemed applying Parliament's intended sentence in all cases would be manifestly unjust. I doubt that's an accident.