Go to the article they cite and you will see two things. 1. They claim that longer sentences do in fact work (but are subject to diminishing returns). 2. The evidence is weak and full of caveats. Like it's not like we have RCT running for generations. We have observational studies rife with problems. They understandably try to make the best of what they have, but frankly it's just not good enough.
Yes, it's possible incarceration itself is a sham treatment for criminal behaviour.
> One of the major justifications for the rise of mass incarceration in the United States is that placing offenders behind bars reduces recidivism by teaching them that “crime does not pay.” This rationale is based on the view that custodial sanctions are uniquely painful and thus exact a higher cost than noncustodial sanctions. An alternative position, developed mainly by criminologists, is that imprisonment is not simply a “cost” but also a social experience that deepens illegal involvement. Using an evidence-based approach, we conclude that there is little evidence that prisons reduce recidivism and at least some evidence to suggest that they have a criminogenic effect. The policy implications of this finding are significant, for it means that beyond crime saved through incapacitation, the use of custodial sanctions may have the unanticipated consequence of making society less safe.
https://onlinelibrary.wiley.com/doi/epdf/10.1111/j.1745-9133...