Adequacy of existing census statistics for basic demographic research

Abstract

Present knowledge of population characteristics and trends in many parts of the world is seriously defective, when compared with the minimum needs for even the most elementary demographic analysis. The deficiencies are especially serious in the countries of Africa and Asia, and to a lesser extent in those of Latin America. Very many of the countries in these regions have had either no population census at all in recent decades, or only censuses known to be grossly deficient in coverage. Only a few of them have census data on sex and age of the population which can meet mild tests of accuracy and of sufficiency for analytical studies. Yet in these regions the census is almost universally the only source of information about population trends, for statistics of births and deaths are usually either totally lacking or too incomplete for use in this connexion.

The population censuses which are being taken in 1950 or planned in 1951 for a great many countries will undoubtedly make an important improvement in this situation. But the extent of the improvement, with reference to the statistically backward regions, may be disappointing unless greater care is taken in these censuses than in previous ones to obtain data which are accurate enough, broad enough in coverage and tabulated in enough detail to satisfy the most important needs for analysis. Such methods as careful verification in selected sample areas of the completeness of enumeration and accuracy of returns, especially on such important subjects as sex and age of the people, would ensure the maximum product of usable statistics.

The need for such improvement is shown in this article by an analysis of the census statistics obtained in each country having 1,000,000 or more population between 1919 and 1949. Tables are presented showing the availability of essential data on population size, growth and structure with reference to sex and age groups for each country, together with the results of simple tests of the completeness and accuracy of the data.