The Measurement of Income in Fertility Surveys in Developing Countries
MacDonald, Maurice, and Eva Mueller. 1975. "The Measurement of Income in Fertility Surveys in Developing Countries." Studies in Family Planning, 6(1): 22-28.
Against the background of the interest in relating the influence of economic factors to fertility, especially in the developing world, this study sets out to determine a) whether a detailed sequence of questions about income components yields significantly better results than a shortcut approach; and b) to what extent the relation between income and fertility may be distorted when rough approximations of income are relied upon. 3 surveys conducted in Taiwan between 1967 and 1970 shed light on these questions. They are based on a sample of 2200 couples of childbearing age who were interviewed 3 times. In the first interview, the wife was asked to estimate their monthly income, which was then converted to a yearly basis (YW). 20 months later the husband was asked detailed questions about his income (YH). 8 months later the wife was interviewed again. This time she was asked specific questions about expenditures. This rough expenditure measure was converted to an annual basis (EW). None of the estimates agreed. Comparison with outside data revealed that EW came closest, with YH only slightly behind. Even with some discrepancies attributed to misunderstanding and inadequate information, there does seem to be a relationship between discrepancies and socioeconomic characteristics. No appreciable influence of income on current or desired family size could be found. However, the affluent did appear to use conception to a greater degree. The lack of a relationship between income and current and desired fertility does not seem to be due to the poor quality of income measurement. However, when there is a relationship, as between income and contraceptive use, errors in income measures can weaken the findings. When income data obtained by shortcut methods are analyzed with socioeconomic variables as part of a more complex model, there is danger of further errors, as the Taiwan data suggested that there are biases in income reporting associated with a education, type of employment, and the number of living children. The authors conclude that it is worthwhile to devote care and time to income measurement.