Population Estimates by Census Tract, New York State, by Age and Sex, 1990-2016.

Iterative proportional fitting was used to develop populations that are consistent with official Census Bureau tract-level populations from 1990, 2000, and 2010 and single-year county-level population estimates published by the SEER program of the National Cancer Institute (https://seer.cancer.gov/popdata/).

The Longitudinal Tract Database (LTDB) (https://s4.ad.brown.edu/projects/diversity/researcher/bridging.htm) was used to report populations using 2010 census tract boundaries.

In effect, the approach assumes that population growth or reduction at the tract level mirrors what is happening at the county level. This is an improvement over linear or geometric interpolation between census years, but is still far from perfect. Census tracts can undergo rapid year-to-year population change, such as when new housing is constructed or, less frequently, demolished. An extreme example is census tract 1.04 in Westchester County, New York, which had a population of 0 in all 3 census years, as it was located entirely within an industrial area. Since 2010, multiple large high-rise condominiums have been constructed here, so that the population in 2018 is probably now in the thousands, though any estimation or projection method tied to the 2010 census will still count 0 people here.

It is conceivable that address files from the United States Postal Service or other sources could be used to capture these kinds of changes; I am unaware of any attempts to do this.

The file contains data for 4893 census tracts. It has been restricted to census tracts with nonzero populations in at least one of the census years. There are other census tracts consisting entirely of water, parkland, or non-residential areas as in the example above, which have been omitted.

These data are used for the calculation of small-area cancer rates in New York State.