Teachers who pass the tests required for this probably are from the top of their teacher cohort regarding intelligence, consciousness, empathy, motivation etc and they are likely better paid, so less stressed from life stuff like bills or whatever.
The peers all come from non-random families who generally have good life outcomes either way. These kids will model good behavioral norms. If kids can interact, they will influence each other. It's not only disruption or no disruption, but likely more subtle.
The effects were only shown here up to end of kindergarten. It likely disappears later.
My default assumption remains that these fashionable methodologies do very little actually. The correction is probable almost entirely due to who the people involved are. I know that the study participants were randomized to Montessori or not, but the rest of peers and the teachers are not random.
The real test would be for a Montessori kindergarten to drop the methodology while keeping the same people.
