GPWv4 releases data for five years: 2000, 2005, 2010, 2015, 2020. Is it advisable to compare data sets of different years in order to assess changes in population density over time? Are there constraints in using the GPWv4 target year data sets as a time series?

The GPWv4 data for the five target years (2000, 2005, 2010, 2015, and 2020) are created from the same input population data, not from sources from multiple years; therefore the data for the five target years do not reflect a true time series and should not be used as such.

GPWv4 uses annualized growth rates to adjust the input census counts for each country to a series of target years to allow for global comparison. For each country, exponential growth rates are calculated in each administrative unit by matching the total population from the input data to those from a previous census enumeration. This growth rate is then applied to the input population data to extrapolate it to the five target years.

Thus, the population estimates for all five years are derived from the same input population data, not from different sources. In addition, the same growth rate is used to extrapolate the data to all target years. As a result, the target years do not reflect a true time series.

Feedback and Knowledge Base