In South Africa, like in many developing countries, the differences between
enrolment totals, estimated by the education authorities, and the numbers of
children in the country, estimated by demographers in the national statistical
agency, defy easy explanations and suggest that one or both sets of statistics are
inaccurate. In South Africa the gap between the two sets of estimates is
substantially larger than one would expect. The typical reasons that have been
found to underlie developing country data problems of this kind are discussed and
their applicability to the South African data is investigated, using a variety of data
sources. It is found that not clarifying the reasons behind the data discrepancies
and not making necessary adjustments lead to distortions in commonly cited
international development indicators that are not insignificant. It is demonstrated
that analysing the various possible reasons for unexplained gaps between
enrolment and population aggregates can reveal patterns that are in general
useful for education planning. For instance, comparing the educational attainment
of adults to enrolment patterns for children in the household data can help to
gauge the extent to which the child enrolment responses are subject to typical
upward biases. The analysis as a whole highlights the importance of collaboration
between the education authorities and national statistical agencies to improve
data collection and imputation techniques on both sides.