Do standards and discrepancies in data sources matter?
Do standards and discrepancies matter?
Yes they do. Even if the data were used only for international comparisons, the lack of standards and the existence of sometimes large discrepancies can be very misleading in terms of human development and could lead to wrong decisions about social and economic policies. But components of the Human Development Index (HDI) - GDP per capita, life expectancy, literacy and school enrolment are also used to assess where problems exist within a country and how they may be developing or responding over time to policy decisions.
Small discrepancies may be insignificant but we also see very large ones. For example, the infant mortality rates for a country from the Commonwealth of Independent States (CIS), as quoted by international agencies varied from 19 to 81 deaths per 1,000 live births. And of course mortality is a component of life expectancy. Among the reasons why a national HDI may be different from the global version is that a country may use its own estimates of life expectancy or population.
If standards and discrepancies matter so much then why do the agencies and countries get the numbers so wrong?
In the case of infant mortality quoted above, there were several reasons for the discrepancies, but the most significant reasons were to do with the underreporting of births and deaths in registration. In this case, as in many others, the better figures came from expensive household surveys that did not depend on incomplete registration data.
Another frequent reason for discrepancies between national and international sources is that countries may not give high priority to supplying data to international agencies. The advent of human development reports and MDG indicators has highlighted the importance for countries to share this information in a timely and consistent manner. However, even within a country, different ministries often have different data. Quite typically, a line ministry (education for example) may supply one set of data to an international agency, while another part of government such as the national statistical office may supply different data to the statistical division of a regional commission or to the UN Statistics Division.
So is it the countries that are at fault?
International agencies may also make errors and they vary in the way they treat the data they receive. Some agencies never adjust the data that are sent to them, whilst others do make adjustments in an attempt to standardize the data and make them comparable between countries.
Also sometimes international agencies create their own definitions in order to meet some specific needs. For example, the World Bank converts GDP estimates per head from local currencies to USD purchasing power parities so that the figures for different countries can be compared. This version of GDP is important for calculating the HDI, but it may not be recognized by the country itself. Similarly, there are now estimates of 'functional' literacy for some (mainly industrialized) countries that are quite different from literacy figures available globally.
One of the main reasons why statistics are often not comparable between countries and are not well sourced and calculated is the scarcity of basic information and the lack of expert resources in many countries. Statistics cannot be much better than any other part of the national infrastructure.
How can I differentiate between good figures and bad?
There are several ways. Firstly read the small print. Any good publication or web site will describe the sources and methods in some detail. Secondly, it can be clear from the description of the source or method that some data are not what they seem. For example, unemployment estimates that are based purely on the numbers of people who receive unemployment benefits, or who choose to register at their local unemployment office, must be of very limited value in terms of indicating the availability of people for work. Education enrolment data are generally believed to be better but they still derive from the administration, and enrolment is not the same as school attendance.
Can we, in UNDP for example, do anything about bad data?
Yes we can, even though we are users rather than producers of data. Firstly, we can refuse to publish misleading figures. To publish figures we are not sure about compounds the problem of bad data. It is our responsibility, through our global reach and our national presence, to facilitate national/global cooperation. We have many examples through the visibility given to statistics in the global Human Development Report.
Where will I find good statistics and good advice on statistics?
Human development is a broad field and requires statistics from very many sources. At the international level there are many good and well-documented sources including the Human Development Report itself and our associated web site. Further excellent general sources are the UN Statistics Division site and the World Bank too; most notably their annual World Development Indicators which has very good notes on all of their wide variety of statistical tables.
At the national level the best source is almost invariably the national statistical office. Not only are they usually by far the largest sources but they can also advise on other sources in their field.
Some general sources:
1. UNDP/HDRO (including HDI calculator)
Note: HD Insights are network members' contributions and do not necessarily represent the views of UNDP.
Back to the list of past editions.