What is AfriPop:demography?

 

Large area spatial demographic datasets that provide per-grid square estimates of numbers of people are widely available through a range of projects, including AfriPop, but similar information on the demographic composition of these populations at a subnational scale are not freely available. AfriPop:demography aims to fill this gap for Africa through the provision of detailed gridded spatial population datasets showing age composition by 5-year groupings and gender. Contemporary data from national censuses, census microdata and household surveys are combined to quantify the subnational age and gender structures of African populations across the continent, and used to adjust existing AfriPop population distribution datasets. Moreover, the collation of subnational inter-censal growth rates enables production of 2000, 2005, 2010 and 2015 datasets to support Millennium Development Goal related assessments, amongst others.

 

 

How are the spatial datasets constructed?

 

National population censuses collect information on the age and gender of those surveyed, and summarise results to a regional level, generally by 5-year age groupings. Where these data are available, they provide a valuable resource, highlighting the substantial variations that often exist across countries and between urban and rural areas in terms of the demographic structures of populations. The example below (figure 1) shows the proportions of the population that are under 5 years old in Tanzania, mapped out to enumeration area level, as well as the difference observed from the UN Population Prospects 2010 reported national-level under 5 percentage of 17.4%. However, for many countries across Africa, such census data are not reported, or recent census data do not exist, presenting a barrier to obtaining contemporary estimates of subnational age and gender compositions. Here, national household surveys play an important role in providing contemporary demographic information that can be summarized at sub-national resolution. While such surveys cover much smaller sample sizes than national censuses, they are conducted regularly in resource-poor countries where census data is sparse. Within AfriPop:demography we have assembled a unique database of census and household survey datasets to give the most complete and up to date picture of subnational variations in age and gender structures ever produced. Figure 2 below shows the data types used to construct this picture and the most recent year of data used. The proportions derived from these datasets shown in the examples in figure 3 are used to adjust existing AfriPop population distribution datasets to produce detailed demographic composition spatial datasets.

 

(a)                                                                                    (b)

Figure 1: Enumeration area level Tanzania maps showing (a) proportion of the population under five years old; (b) difference from UN Population Prospects 2010 reported national-level under 5 percentage of 17.4%.

 

 

(a)                                                                                            (b)

Figure 2: (a) Type of dataset used to construct AfriPop:demography products (AIS = AIDS Indicator Survey, DHS = Demographic and Health Survey, MICS = Multiple Indicator Cluster Survey, MIS = Malaria Indicator Survey); (b) Year of most recent dataset used to construct AfriPop: demography products.

 

 

(a)                                                                                            (b)

Figure 3: The proportion of (a) children under five years old, and (b) women of child bearing age (using standard definition of ages 15-49).

 

 

Figure 4: Inter-censal growth rates used in the construction of AfriPop:demography products

 

 

 

How are the multitemporal datasets constructed?

 

National populations rarely change uniformly across a country—often the populations of certain regions will grow significantly faster or slower due to higher birth rates, migration or conflicts, for example—but this heterogeneity is generally not captured in the development and use of existing spatial demographic datasets. Often, simple national-level growth rates are applied to existing spatial population datasets to estimate populations at risk for target years in influential mapping studies (e.g.1,2,3), ignoring the spatial and temporal variations in population changes. For AfriPop:demography, we aim to better capture this variation through the construction of a database of subnational inter-censal growth rates. Figure 4 shows the most recent subnational growth rates used. These will be updated continually as more data becomes available. These rates are applied to the datasets constructed above to produce estimates of population distribution and composition across Africa for 2000, 2005, 2010 and 2015. Finally, all population totals are adjusted nationally to ensure that they match those published by the United Nations Population Division.

 

 

What datasets are available?

 

The construction, testing and validation of national datasets at the same 100m spatial resolution as the existing AfriPop alpha population distribution datasets is currently underway. In the meantime, alpha version Africa-wide 1km datasets representing children under five years old and women of child bearing age (using the conventional definition of 15-49yrs) for 2000-2015 are freely available for download here. An example of one of the datasets is shown below in Figure 5. Soon, full resolution country datasets for each five year age group by sex will be made available here.

 

Figure 5: Example output dataset of 2010 distribution of women of childbearing age, showing estimated number per 1x1km grid square. The inset box shows a closeup for Ghana.

 

 

Feedback

 

We welcome all feedback on these datasets—if you see errors, or know of data that can be used to improve outputs, or just have questions/comments, please feel free to contact us.

 

 

What is coming next?

 

Dataset updates and the construction of full documentation and metadata are in progress. Through the AfriPop:demography project we have also started work on expanding our demographic databases beyond age and gender, with the aim of exploiting the rich and ever-expanding datasets produced from national household surveys ongoing throughout Africa. Work has begun on producing similar datasets for Asia, through the AsiaPop project, and these will likely be available in late 2012.

 

 

>Go to data download