And what Friednman is doing is a data trawl. That is different to setting a research question and collecting data to specifically answer that question. There are so many problems with longitudinal cohort studies but if they are set up with an end in mind and those potential problems identified and mitigated useful data can be developed. I am not convinced that was the case here.

Is entering school good or bad? Depends on the child, their family and the school.