Thank you for your constructive message. The original post was written by a junior researcher and does not accurately reflect our work; we apologize for the frustration it caused. We will update it soon. We are happy to correct any inaccuracies in light of new evidence. You are welcome to DM me here

When we use data from the past to describe the present, the figures we record are marked as unreliable in the dataset so researchers can filter according to the level of reliability

The 2001 Census is the only official Ukrainian census of the 21st century. It is less than ideal but still the best quality data we have on ethnic identification. We subtract Crimea in 2014 because the state doesn’t administer education here after this date

No, because our dataset ends in 2020

Because the territory is occupied by Russia and Ukraine does not administer public services here. We consider it part of Ukraine prior to 2014.

My point is that two research projects arrived at the same number independently

We subtracted the Russian population living in Crimea to arrive at this number. You can find a very similar figures in other datasets such as the Ethnic Power Relations (EPR) dataset

Based on respondents’ self-identified ethnic origin in the 2001 Ukrainian census

The 2012 legislation enabled regional languages to function, in practice, similar to state languages at the local level. Our coding is still based on Ukrainian law.

This refers to the 2012 Law on the Principles of State Language Policy in Ukraine, which allowed regional languages—spoken by at least 10% of a locality’s population—to be granted a status that enables their use in administration, education, and public services.

We base this figure on the nationality question in the 2001 Ukrainian Census, which asked respondents to self-identify their ethnic origin. The question about native language was asked separately in the same census.

We will detail all this information in our codebook and introductory article, but you are right to point out that we should provide more clarity at this stage and format

We base our coding on the 1989 Law on Languages of the Ukrainian SSR, which declared Ukrainian the primary language of instruction. The fact that it was compulsory should not be taken to mean that it was taught in every school, either in law or in practice.

You are pointing to the de jure vs. de facto distinction. We chose de jure for global comparability; tracking on-the-ground realities would make such a dataset impossible. It's a trade-off. If you're focused on the lived experience in Ukraine, a quantitative research design is not the right choice

No data—qualitative or quantitative— can claim to capture the totality of lived experience. This is why we limit ourselves to de jure language policies. If you believe we have made a factual error, we are happy to revise our coding in light of new evidence

To clarify, this is a cross-national dataset recording state language policies across 163 countries. It is a public good that researchers can adapt to their research interests. Your critique is akin to asking, "Why doesn't the World Bank data track the sources of global inequality?"