Global Population Data Quality Ratings

Updated: January 2023


The accuracy of population data varies widely across countries.

The most comprehensive data on the number of people living in a territory and their demographic profile, a vital component for public sector economic and social planning and also for private sector needs, is usually only available from the result of a census.

Over a sample of 155+ countries, the average census age at the end of 2022 was 10 years. 62 countries had census data older than the recommended 10 -year interval between censuses.


Most countries conduct a regular population and household census every ten years and some even use a five year cycle. National statistics offices produce only estimates of total population numbers and the demographic breakdown for the intervening years. The accuracy of these estimates depends on the coverage of the last census and the elapsed time since the census, the data and assumptions about births, deaths and net migration and a host of other factors related to the capacity of the national statistical office and its ability to carry out its functions unimpaired by political interference.

There are a number of problems which limit the accuracy of these between census population estimates. National censuses require a large amount of resources to carry out and often vary in accuracy even for developed countries. The Brookings Institution, for example, has expressed fears that political interference and budget restrictions may seriously impair the accuracy of the 2020 United States population census. In many developing countries there are large gaps in terms of the years between holding a census. This means that population estimates made become less and less accurate as time elapses.

Over a sample of 155+ countries, the average census age at the end of 2022 was 10 years. 62 countries had census data older than the recommended 10 -year interval between censuses. The problem is particularly serious in Africa. For example the census age in Namibia is 11 years and 38 years in the Democratic Republic of Congo. The problem is not confined to Africa. Lebanon, for example, has not conducted a census since 1943.

Estimating the level and distribution of population between censuses requires accurate data for live births, deaths and net migration. In developed countries the registration of births and deaths is usually comprehensive. In many developing countries, in contrast, lower registration rates, particularly of births and especially in rural areas means that the data available to national statistics offices is often widely inaccurate. According to the World Bank the registration of children under 5 years old is less than 50% in rural sub-Saharan Africa.

This is a major handicap for national statistics offices in most emerging markets many of which have limited resources available to undertake surveys and it is compounded by long gaps between censuses. The same problem exists in many emerging markets for estimating net migration statistics, particularly for countries that have seen uncontrolled economic and war-induced migration. The accuracy of net migration data is also subject to wide errors in some developed countries. A committee of Members of Parliament in the UK once described official net migration data as ‘little better than a guess.”
 

Population Data Quality Rating (PDQR)

In order to take account of these problems and to assess the relative accuracy of population estimates World Economics has produced a Population Data Quality Rating (PDQR) using methodology similar to the Data Quality Rating used to monitor the quality of official Gross Domestic Product (GDP) data across countries. The Ratings currently cover three factors which it is believed determine population data quality. Each factor is evaluated to provide country scores which are combined into the PQR score using a weighted aggregate to reflect the importance of each of the individual factors. The three factors used to judge data quality are:

Last Census Date (years out of date) – weighting: 60%

Statistical Capacity of National Statistics Office – weighting: 20%

Percentage registration of births – weighting: 20%

 

Last Census Date:

This factor is based upon data on the last census year and the years that have elapsed between that year and 2022 plus one to avoid a zero score for countries that are fully up to date. This data is then ranked to give a score between 0 for Lebanon (80 years out of date) and 100 for Brazil which is up to date which are then converted into an index value for each country to make the data comparable with the other two factors. Data on the number of years out of date are provided in the Population Data Quality Ratings Tables along with the source data.

 

Statistical Capacity

The ability of a country to produce accurate data is a function of its statistical capacity which depends on the resources – human, financial and technical – of its national statistics office. The PQR uses the World Bank’s Statistical Capacity Index (SCI) which varies between 0 and 100 to measure this factor and which is based on an a amalgam of three factors: methodology, periodicity and the source data employed.

 

Registration of Live Births

The third factor employed is estimates of the percentage of births registered by country using UNICEF data available from a World Bank dataset. This variable is available in percentage terms and can be converted into an index varying between 0 and 100 and is used as a proxy for the accuracy of population estimates produced between censuses. Data on live births registered by country are provided in the online tables.

The three factors are weighted in terms of relative importance and are combined into a Population Data Quality Ranking (PDQR) score which is ranked based on an index between 0 and 100 for the quality of population data in each country. The PDQR scores are shown in the Population Data Quality Ratings Tables.

The PDQR data is also divided into quartiles which allows countries to be assigned ratings A, B, C and D based on the quality of population data from best to very poor. For example, a country like the Democratic Republic of Congo whose last census occurred in 1984 produces a low Out of Date index of 51.9 which when combined with a SCI rating of 40.0 and an estimate of only 40.1% of live births registered ends up with a low PDQR score of 47.2 well below the average score of 85.4 and is ranked D. Ireland, in contrast, gains a top score of 100.0 and gains an A rating.