SPATIAL FERTILITY ANALYSIS IN A LIMITED DATA SITUATION : THE CASE OF PAKISTAN

This paper, published by the East-West Population Institute, analyzes fertility data from Pakistan as a test case to argue the benefits of an aereal or spatial analysis of fertility data to explain linkages left unexplained by socioeconomic analyis. The spatial approach is based on the assumption that linkages between fertility and socioeconomic characteristics vary systematically over space. On the basis of this assumption, inferences can be drawn about these linkages when data are incomplete or limited, as is particularly true in Pakistan. Basically, inferences are drawn, though with less accuracy, in much the same way that topographic maps are created. (In mapmaking, direct measurements of elevation are made at only a few points, with the geographer interpolating between the observation points.) In the present study, district level data from 1951 and 1961 censuses were used to construct fertility trend surfaces in three orders. First, or planar-order surfaces indicated that, in general, fertility is highest in the northwest and lowest in the southeast. While second-order surfaces add little beyond this, third-order surfaces, which are the most accurate of all, reveal lower fertility in the south, east, and northeast, thus modifying (without negating) the northwest-southeast generalization revealed in the lower-order surfaces. Positive residuals (exceptions) from third-order surfaces can have a practical value for program planners faced with strategic choices in regard to fund allocation. For example, it can be inferred from 1951 residuals that high fertility districts probably possess abnormal linkages between socioeconomic variables and that resource allocation in these areas would have a higher probability of success. Similarly, the shape and pattern of 1961 residuals indicates that economic development along the Indus River Valley may be responsible for increased fertility rates, implying a strategic response such as including fertility costs in the costs of development projects. The study concludes with suggestions for improving spatial methodology. A 16-item bibliography (1959-1978) is appended.