this research is trivially shown to be a pile of garbage. It ASSUMES the only cause of recidivism can be the length of prison sentence, and therefore that relation is cause. It totally ignores that harder criminals, when caught tend to end up with longer sentences (because, well, they do worse crimes..) and that these same harder criminals are more likely to not change their ways.
No, it assumes there might be other causes, and controls for those. Here's the relevant part of the article (which I already quoted in my original post):
A 1999 study tested this assumption in a meta-analysis reviewing 50 studies dating back to 1958 involving a total of 336,052 offenders with various offenses and criminal istories. Controlling for risk factors such as criminal history and substance abuse, the authors assessed the relationship between length of time in prison and recidivism, and found that longer prison sentences were associated with a three percent increase in recidivism.