A technical working group (TWG) convened in Port Sudan in 2005 determined the majority of the content of the census questionnaires. The short form questionnaire (SFQ) is designed to provide data on population by basic demographic charac- teristics at national level and down to the level of the lowest administrative districts. It contains 10 questions on age, sex, nationality, place of birth and state of usual residence. In addition to these 10 questions, the long form questionnaire (LFQ) includes more detailed modules on housing, mi- gration, education, labour force participation, fer- tility, mortality, agriculture and animal husbandry. Initially, the census questionnaire was to contain a

question on ethnicity and another on religion: the office of the Presidency removed these in March 2007 and replaced them with a question on re- gion of origin.

The pre-enumeration sampling plan, agreed by the TWG at Port Sudan in 2005, was to administer the short form questionnaire (SFQ) to households in 90 percent of enumeration areas and the long form questionnaire (LFQ) to households in 10 per- cent of enumeration areas. Private households, in- ternally displaced persons and refugees within this 10% of EAs were all supposed to receive the LFQ. The sampling frame for the 10 percent sample was generated using pre-enumeration mapping. EAs were stratified according to whether they were urban/rural, and then by geographic code (state, county, payam, boma, EA) within each stratum. EAs were selected with equal probability from this ordered frame, with post-enumeration weights to capture differences in population between the dif- ferent EAs.

Actual sampling did not strictly follow the pre-enu- meration plan: only in 65% of the EAs originally designated LFQ EAs were any long form question- naires administered. While this may have intro- duced a bias into which households were admin- istered the LFQ, post-enumeration GIS mapping found no obvious gaps in geographical or demo- graphic coverage. Technical assistance from USDA was required to re-work the population weights for LFQ households based on the realized sample. The cleaned and edited long-form Census dataset contains 91,449 households, and 542,765 indi- viduals. The realized LFQ sample is therefore 6.5 percent of total households enumerated in South- ern Sudan. The sample includes private house- holds, internally displaced persons, nomads and refugees, but excludes cattle camps, institutional households, homeless people and overnight trav- elers. The weighted sample is representative of a total of 1.28 million households and 7.65 million individuals, which is 92.5 percent of the popula- tion of Southern Sudan

Implementation

Implementation of the 5th Sudan Population and Housing Census in Southern Sudan occurred in three main stages: pre-enumeration, enumeration and post-enumeration. Planning the census was a joint effort between North and South, with budget, work plan and questionnaire content decided by bodies containing representatives from the North and South, as well as external experts.