I think there is a typo in the paper, that was carried over to the article. These two sentences appear to contradict one another as written:

> Interestingly, we observe significantly improved student test scores in the second year of the ban (about 2-3 percentiles higher than the year before the ban) when suspensions revert to pre-ban levels.

> Overall, we show that student test scores improved by 0.6 percentiles, with the ban increasing spring test scores 1.1 percentiles in the second year relative to the spring test right before the ban took effect.

Instead, I think the 1.1 percentile gain should be about the first year, and a 2-3 percentile gain by the second year. That is consistent with the graph.

But yes, a fairly small gain. I agree that much of the gain could be recovering from losses during the pandemic. Also the FAST is a new test that started in the 2022-2023 school year, so some of this could also be due to students and teachers adjusting to the new test and improving over time.