The count and the instrument both have histories
A category grows between two censuses. That observation is a starting point, not a complete explanation. People may have moved, been born or died, changed how they described themselves, or encountered a questionnaire that recorded their answers differently. A good trend note checks the instrument alongside the population.
The Census Bureau's account of the 2020 race and Hispanic-origin questions documents changes to examples, write-in areas and processing. It increased captured write-in text from thirty to two hundred characters and assigned up to six detailed codes per write-in area rather than two. That can change the information carried into a tabulation.
Sources: s1
The warning also matters outside the decennial census
An American Community Survey user note cautions about comparisons with earlier race estimates beginning in 2020 because of similar questionnaire and coding changes. This is a version boundary. It does not establish that every observed difference is artificial, and it does not quantify a universal adjustment that can be subtracted from a chart.
Before comparing, keep survey, year, geography, universe and category definition together. Read the relevant technical note and identify whether responses, codes or published groups changed. If no defensible bridge exists, display the break visibly and explain what remains comparable instead of drawing one uninterrupted line.
The category is part of the measurement
A statistical label is part of a collection and reporting system. It is not a complete account of a person's identity. This distinction helps prevent a measurement change from being narrated as a sudden transformation of everyone placed inside the category. The methodological footnote is essential to understanding what the number says.