Data Limitations
The FALCOT 92 was not meant to be a study of childhood mortality rates for
this population. Therefore, possible shortcomings of the data set should
be evaluated with this in mind. The limitations we are about to present
force us to review the results on childhood mortality as rough estimates
rather than well established unbiased estimates:
Sample Size Limitations
The sample size needed to estimate the probability of dying before exact
age 1, q(1), or the probability of dying before exact age 5, q(5), should
ideally have been many times larger than the one actually obtained. Furthermore,
the current sample size (1223 females) does not permit elaborate analysis
on the regional level, nor does it provide for serious analysis by factors
such as gender or level of education. (See also this book's Appendix A on
sampling strategy).
Questionnaire Limitations
Some important questions relating to data collection on children ever born
and children dead were missed during questionnaire development. In particular,
no questions were asked concerning the status of the last birth, such as
date of birth, sex, and whether the child was still alive by the time of
interview. Moreover, no questions were asked to allow estimation using birth
history or life table methods. The questions missing on the status of last
birth will mostly affect fertility calculations.
Parity limitations
For the purpose of parity calculations and analyses, all ever married women
of child-bearing age (15-49 years) in each household should have been interviewed
for the purpose of inclusion in the childhood mortality and fertility analysis.
The exclusion of ever married women (of unknown quantity) from interviewing
for this section makes it very difficult to estimate average parities in
the manner recommended by demographers.
|
