The search for substitute data to replace information that would have been available from properly kept civil registration records has resulted in many developing countries using censuses and sample survey of adult women to collect two types of data average number of children born alive (or parity « P ») and fertility rates during the preceding twelve months (from which the cumulative period fertility rate « F » is calculated) A man of statistics has thus become available, but up to now these have only been used, to study populations whose fertility rates were constant over time (P/F method) In this article we show how, by inverting the traditional hypotheses on fertility rates (which are now considered as variable) and on retrospective survey data (considered as sufficiently good), these data can reflect the decline m fertility rates among populations in which the second stage of demographic transition has begun A new method used to reconstitute fertility rates by five years age groups and by periods is explained, and applied to census data from Marocco (1982) This simulation leads to results very similar to those obtained by the World Fertility Survey.
Recent changes in marriage patterns in the Muslim Arab world are analyzed. The author notes that the problems posed by the early age at marriage for women, virtually universal marriage, and a surplus of women in the marriage market were traditionally resolved partly by polygyny but primarily by repudiation. Changes over the past 25 years have tended toward a stabilization of marriages and a decline in repudiation of wives.
Age at Death and Mortality Levels. An Attempt to Estimate the Completeness of Death Registration from the Age Distribution of Deaths. An Application to Tunisian Data. The decline in mortality is normally measured by period indices. From the cohort point of view, what is being measured is an increase in the ages at death of successive cohorts. There is, therefore, a relationship between the level of mortality and the age distribution of deaths. This relationship enables us to estimate the number of deaths after the first birthday in countries where death registration is defective from the age distribution of recorded deaths and to calculate mortality tables from first birthday onwards. A relationship between the level of mortality and the age distribution of deaths during the first year of life also exists, but it is different. The causes of death which are most difficult to control are those which occur during the first days of life. By using statistics from 18 developing countries we were able to construct a relationship between the proportion of neo-natal deaths and the infant mortality rate. This makes it possible to estimate the infant mortality rate, when the proportion of deaths in the first month of life is known.