They did point that out in the abstract. The data sources being school policy changes and school-age cutoffs does probably help a lot to prove the direction though.