You like systems and procedures that produce predictable and consistent outcomes and you look for what could go wrong.

The local models may not have evidence of consistent outcomes.

Again, no consistent outcomes were observed across any group and further, no significant improvements were observed within any individual participant.