Post-enumeration
THERE has been much controversy around the 2017 census. Indeed, when the 24th Constitutional Amendment Bill, which would let the ECP use the provisional results for delimitation of electoral constituencies, was presented in the Senate, the opposition refused to support it without certain conditions being met.
These included a third-party validation survey of five per cent of randomly selected census blocks, supervised by a census commission comprised of reputed demographers, the final report of which would be submitted within 30 days. Any deviation of the sample survey results with the provisional census figures beyond plus-minus one per cent would be incorporated in the final census results.
A consensus was finally reached among the major political parties, and on Dec 19, 2017, the 24th Amendment was passed. On Jan 5, 2018, a census commission (of which I was a member) was notified by the Ministry of Statistics. The commission had its first meeting two days later, and was attended by other senior demographers and senior representatives of the MoS and the Pakistan Bureau of Statistics (PBS).
The general consensus was that the census did suffer from several flaws, and that it would be appropriate to conduct a third-party validation exercise. For this purpose, advertisements in newspapers were also published and several firms were shortlisted. Although a second meeting was scheduled for Jan 15, it was cancelled, and the commission was not convened again.
The 2017 census results must be validated.
On May 10, a summary was moved by the MoS recommending that the Council of Common Interests (CCI) revisit its earlier decision and accept the final results of the 2017 census without conducting the post-census survey. The justification for this was based on the view that each enumeration block would yield different results due to migration, births and deaths, that the validation exercise would incur a heavy cost, and that the field work could go on for as long as two months.
It was perhaps not realised that, given a detailed census was conducted in four months in about 165,000 (100pc) blocks in the entire country, it would not have taken two months to conduct a sample survey in 8,000 (5pc) blocks. And any demographer could have advised the PBS on the best way to measure changes in the population over a year due to migration, births and deaths. Besides, the cost estimate of Rs4 billion was exorbitant given that the total cost of the census itself was Rs12.5bn.
Apparently, the reason why the PBS readily accepted this view was because of a technical evaluation of the 6th population census prepared by an expert committee. The committee’s findings are revealing. For example, the report notes that the governing council of the PBS had constituted a sub-committee to advise the PBS. But the sub-committee took the view that “the census [was] not being planned the way it was envisaged” and that “none of its recommendations were being followed in true letter and spirit”.
Furthermore, it found that geography/GIS officers had not been involved in the process of updation work of both urban and rural area frame and maps; that there was a lack of coordination among national bodies; and that poor compliance on the recommendations of national and international agencies, coupled with an acute shortage of technical expertise and staff, “had left the PBS with a number of loopholes”.
Besides, a pilot census was not undertaken despite being strongly and repeated recommended by national and international experts. The sub-committee also believed that, given that there is never an error-free census, and owing to a lack of comprehensive and accurate data from other sources such as civil registration and population registers, a post-enumeration survey (as is the practice in many countries) was needed. This was also turned down by the PBS authorities.
The last meeting of the CCI did not endorse the MoS’ recommendations, and the buck has apparently been passed on to the next elected government. It will be unfortunate if a consensus is not developed to accept and properly utilise the census data. Indeed, using flawed census results for planning purposes could lead to flawed planning as well.
Perhaps the best course would be to let the census commission examine whether the data does suffer from coverage and content errors in certain areas. However, without the validation survey, over- or undercounting in specific areas cannot be detected. If the census results are accepted at face value, they will have long term political and economic consequences. Since the 2017 census was conducted on the orders of the Supreme Court, perhaps the ball is again in its court.
The writer is a professor of demography and served as a member of the PBS governing council.
Twitter: @mehtabkarim
Published in Dawn, July 4th, 2018