I happened to have participated in census work before. For instance, in a country like China, the national census conducted every ten years generally yields accurate overall data, but the data for individual regions is indeed based on estimates. There are several reasons for this:

1. Population Mobility: Generally speaking, in economically developed areas, population figures are often underestimated because a large number of people freely migrate into these regions, and local governments are actually unclear about the exact increase in population. In contrast, in less economically developed areas, population figures are often overestimated because many people leave to work in cities, only returning to their hometowns for brief periods each year.

2. Mortality Data: China’s birth data is already quite accurate. Nowadays, the vast majority of babies are born in hospitals, unlike decades ago when midwives would come to homes to assist with deliveries. Moreover, birth certificates must be issued immediately after a baby is born. However, China’s mortality data is not precise, primarily because burials are still common in many rural areas, and these death records are often delayed.

For example, my city conducted multiple rounds of mass COVID-19 testing in 2021. Each time, more than 4.4 million people were tested, but our small city's 2020 census results only showed a population of 3.7 million.