I appreciate the insight. I was a Math Instructional Coach in 2025 and was skeptical about the huge increase in our ELA scores while our Math scores creeped up 2% (typical growth) in our school. It is frustrating that decisions are made (Receivership, CSI, TSI) etc. based on data that is not factual and could have implications on funding, staffing and curriculum. NYS should address this and provide transparency.
Thank you for the article; great info regarding the test and scoring, as well as the need to dig deeper to understand score swings. And thanks for hosting test data from across the country and making it accessible.
I’m wondering why you don’t include opt-out numbers, given that they have such a large impact on calculated proficiency rates—often making them unrepresentative of the cohort they are meant to reflect. I started plotting NY data, including opt-out data (all free and open-source, data-4-good-not-$). Would love your feedback:
Yes, it looks like “bad testing” in 2025, as you point toward. It’s interesting to think about math in this context. For a lot of math testing, I think you would not have to do this sort of correction (or very minor corrections).
It doesn’t address this issue directly, but a comparison to math results would be helpful in eliminating/limiting some possibilities (eg it was decided to make all the tests/cutoffs etc easier in 2025 and ignore normalization that would eliminate this effect).
Thanks for this thoughtful and informative analysis. It is valuable content is made even more helpful by the observations about having open data and a healthy dose of skepticism.
I appreciate the insight. I was a Math Instructional Coach in 2025 and was skeptical about the huge increase in our ELA scores while our Math scores creeped up 2% (typical growth) in our school. It is frustrating that decisions are made (Receivership, CSI, TSI) etc. based on data that is not factual and could have implications on funding, staffing and curriculum. NYS should address this and provide transparency.
Wow. This is a great substack. Looks like you are 100% right there was some 2025 error. Thanks for detailed discussion.
Thank you for the article; great info regarding the test and scoring, as well as the need to dig deeper to understand score swings. And thanks for hosting test data from across the country and making it accessible.
I’m wondering why you don’t include opt-out numbers, given that they have such a large impact on calculated proficiency rates—often making them unrepresentative of the cohort they are meant to reflect. I started plotting NY data, including opt-out data (all free and open-source, data-4-good-not-$). Would love your feedback:
https://substack.com/@openschooldata/note/c-343080414?r=4fe4ek&utm_medium=ios&utm_source=notes-share-action
Fantastic work, save that it detracts from the most important message: the overall incredibly low proficiency rates in New York City Schools.
Yes, it looks like “bad testing” in 2025, as you point toward. It’s interesting to think about math in this context. For a lot of math testing, I think you would not have to do this sort of correction (or very minor corrections).
It doesn’t address this issue directly, but a comparison to math results would be helpful in eliminating/limiting some possibilities (eg it was decided to make all the tests/cutoffs etc easier in 2025 and ignore normalization that would eliminate this effect).
Thanks for this thoughtful and informative analysis. It is valuable content is made even more helpful by the observations about having open data and a healthy dose of skepticism.