That's not what they did. They used estimates like it says "according to annual Census Bureau estimates that are not related to the official 2020 census counts. The annual estimates are based on births, deaths, construction permits and other records" and "according to new Census Bureau estimates, which do not reflect the 2020 census counts. The agency will release the final 2020 census tally in March." The article also has some incorrect information like claiming Rhode Island and Connecticut continue "longer-term [population] losses" when neither state has lost population in any census going back at least a century (https://www.census.gov/data/tables/time-series/dec/popchange-data-text.html), so who knows what they mean by longer-term (2 years? 5? 10? 25?) or "losses". Anyway there's no point in using an article when you can go straight to the source and look at the data yourself. Furthermore it's kind of dumb to look at year over year change in a population estimate to infer some kind of trend. It would be like climate scientists using a single year of estimated temperature data to show how quickly the climate is warming.