Griffith Feeney's Demography Website

June 10, 2009

ROSTOCK, Germany. Demographic analysis can get complicated, but for any number of practical reasons—lack of time being perhaps the most common—we want to keep things simple when we can. Here’s an example of how simply we can do a quick assessment of the quality of reporting of children ever born (CEB) and children surviving (CS) data in the 2007 census of Palestine.

The Palestine Central Bureau of Statistics publication Census Final Results: Population Report, West Bank (February 2009) provides one CEB table, Table 15, pages 119-130, and one CS table, Table 16, pages 131-142. The report is available in PDF format online here; select “census” publications and look for code 1530. That’s 24 pages of tables, each full of numbers, almost 5,000 numbers altogether. But we only need to look at two of these numbers to get an initial assessment of data quality. How so?

First, we look only at total population. This brings us down to the CEB table on page 119 and the CS table on page 131. Next, we look only at total children ever born, which brings us down to one column in each table, the one labeled “Total Number of Children Ever Born Alive” and, under this, “Both Sexes” in the CEB table, and similarly in the CS table.

Then we focus on the 30-34 age group. Why? Because for this age group the proportion of deceased children among all children born approximates under five mortality, often denoted q(5), the proportion of children born who die before their 5th birthday, for which we can easily obtain other estimates for comparison.

The tables show 237,458 children ever born and 230,466 children surviving for this age group. By subtraction, the number of deceased children is 6,962, so the proportion deceased is 0.0293, or 29.3 per thousand. If this estimate is about right, the quality of reporting, for this age group, at least, is good. If it’s too low—too high is most unlikely—low, the quality of reporting is not so good.

The simplest way to get an estimate is to go to the UN Population Division website unpopulation.org, look for Data Online, then Panel 2, then select “Under-five mortality” in the “Variables” window and “Occupied Palestinian Territories” in the “Country/Region” window, then click “Display.”

This shows that the Population Division estimates for 1995-2000 and 2000-2005 of 30 and 24 per thousand, respectively. The 29.4 estimate from the children born and surviving from the 30-34 age groups refers to approximately 7 years prior to the census, which was taken in the first half of December 2007. so it refers to the end of 2000. Averaging the UN estimates gives 27 per thousand, slightly below 29.4, so there the comparison is very good. There is no evidence of under reporting of deceased children.

In short, the quality of the children ever born and surviving data in the 2007 census of Palestine appears to be good. We can look forward with high hopes that more detailed investigation will provide valuable evidence on the level and trend of infant and child mortality throughout the country.

Of course we do not need the census for estimates at the national level. These can be gotten from a DHS or other population survey. But even a large sample survey strictly limits our ability to get estimates for geographic areas or other subgroups of the population.

The census, being a complete enumeration, makes it possible to calculate estimates for governorates, cities, towns, camps, and socioeconomic groups. The ability to calculate infant and child mortality for many such population subgroups is the rationale for including the children ever born and children surviving questions in the census.