But the demographic survey seems to have been selective about what facts i t would report and connections it would make. Take the health issue. The documen t concludes that the high risk of infant, child and maternal mortality is associ ated with pregnancies where mothers are too young, too old, or have already had several children. But a discussion of poverty is missing from the list of factor s related to health. It would be difficult to deny that poverty, lack of access to safe water, poor housing, poor hygiene and unsanitary conditions all have a s trong bearing on the health of the mother and child. Although the NDS collected data on housing characteristics, it did not include any data on income.ª¥ 

A closer look at the fertility behaviour of the poor is important because of the extensive literature on the ¡°replacement effect¡± of high infant mortali ty . Statistical studies in various countries show high fertility among the poor as a rational desire to have children who will survive into adulthood to help take care of them. This helps to explain why many poor women have babies at such sho rt intervals. The 1993 NDS would have been a good opportunity to verify the vali dity of this behaviour in the Philippines. ª¥ 

The NDS avoided collecting data on socio-economic variables that would have a serious effect on these health issues. But, in one area, it made painstaking efforts to quantify fertility preference to derive figures for planned and unpla nned pregnancies. It concluded that ¡°if all unwanted births were avoided, the t o tal fertility rate would be 2.9 children, which is almost 30% less than the obse rved rate. ¡±This, too, was used to establish an ¡°unmet¡± need requiring a gove rnment programme.ª¥