Technical and political aspects of the 2006 Nigerian population and housing Census

Nigeria’s last census was in 2006. If the decennial rule is followed, the country is barely three years away from another headcount. In this paper, we examine the technical and political aspects of that census in order to derive lessons for subsequent censuses. The focus is not on the census results but rather on the processes leading up to and including the actual enumeration. We describe the connections between population size, revenue allocation and political representation as a means of understanding the social and political dynamics that could undermine the execution of a technically adequate census. These connections are examined through reference to logistic, recruitment and enumeration procedures of the 2006 Nigerian Census. We argue that, like most post-independence Censuses before it, there were motivation and opportunities for manipulating the Census figures. These parameters have not changed.


Introduction
A census is a technical means of collecting data on basic social, economic and demographic characteristics of a population. It is a complex and costly exercise implemented on behalf of government by specified agencies and a host of collaborating partners. The effort required to organize and implement it is justified because evidence-based planning for development purposes is possible only where there is dependable information on relevant population attributes.
The Nigerian Census is a tense political activity. The perceived constitutional connections among population size revenue allocation and constituency delineation have been the main cause of inter-ethnic friction in the five censuses taken in the country since Independence in 1960 and prior to the 2006 census. Two of these censuses -1962 and 1973 -were cancelled outright. The 1991 census was conducted as part of the political transition from military to civilian government. The 15 years between that Census and the 2006 National Population and Housing Census can be traced to the reluctance of national governments to wade into the troubled waters of the Nigerian Census. The country's history of post-colonial census-taking is complicated by the need to mobilize an estimated 389 ethnic groups (Otite, 2000: vii;30;38ff), 36 states and the Federal Capital Territory (FCT) and 774 local government areas (LGAs) for the exercise.
In this paper, we examine technical and political aspects of the 2006 Nigerian Population and Housing Census -that is, the processes leading up to and including the actual enumeration -with a view to uncovering the perceived connections between population size, revenue allocation and political representation, which undermined the execution of an adequately designed census. We do this from the vantage point of roles in the European Union (EU) External Monitoring Mission as Zonal Team Leader (Obono) and National Team Leader, -2007. The paper draws from our acquaintance with the demarcation, logistic, recruitment, enumeration, and other processes of that Census.
Methodologically, we had full access to various preparatory and planning documents 1 , interviewed major stakeholders, including key staff of core departments of the National Population Commission (NPC), external technical support staff, major subcontractors and development partners involved in the census process like UNFPA, UNDP and DFID. We assessed cartography data and enumeration data from the pre-test and Trial Census. As far as we know, this comprehensive information has not been made available to the scientific community by key participants in that exercise before now. We argue that, like all post-Independence Censuses, there were motivation for over-count and opportunities for manipulating the census figures in 2006 and that this can be averted in subsequent censuses through a redefinition of the NPC board, better financial and technical preparation and stronger institutional control and supervision at local government/Comptroller levels.

Background to the 2006 Census
The first population head count in Nigeria was taken in the Lagos area by an American sailor, Captain W. Adams, in 1789, which put the population at 5,000. Other counts of Lagos took place in 1815, 1855, 1861, 1866, 1868, and 1871. Thereafter, censuses were taken on a decennial basis up to 1931 but the first outside Lagos was in 1911 -three years to the amalgamation of the Northern and Southern Colonial Protectorates in 1914. House-to-house enumeration was limited to the main ports and only rough estimates were made of the rest of the territories. According to Aluko (1965: 373), "[the Northern Nigerian] estimates were made on only a sheet of paper and despatched to the Colonial Office" (Aluko, 1965: 373).
The 1931 Census was the first to be centrally planned and coordinated. It was affected by a locust invasion in the Northern Region. The 1931 round was affected in the East by spill-over tensions from the Tax Riots of 1929. The enumeration was incomplete. There was no Census in 1941 because of the Second World War (1939)(1940)(1941)(1942)(1943)(1944)(1945). The 1951The -1953 Census helped shape the political character of all subsequent Nigerian Censuses. It reported the total population as 30.4 million with Northern Nigeria constituting 54 percent of the country's population. These figures were used by the colonial authorities as the basis for allocating regional seats in the 1954 parliamentary elections and the 1959 general elections. They marked the beginning of the politicization of census figures by various ethnic interests.
The preliminary results of the 1962 Census put the national population at 45 million with northern Nigeria totalling 22 million while southern Nigeria had a population of 23 million. They were rejected by Prime Minister Sir Abubakar Tafawa Balewa  amid widespread accusations of inflation of census figures both in Eastern and Western Nigeria (Diamond, 1988). A Central Census Board was established to replace the Regional Census Boards. It conducted the 1963 Census, which placed the national population at 60.5 million and stated that the northern states had been undercounted in 1962. The figure was "adjusted" to 55.6 million, with the Northern region having 29.8 million and the South 25.8 million. This redistribution left the previous legislative apportionment intact (Ekanem, 1972). The reputation of the Nigerian Census as a flashpoint of political controversy was consolidated in these years.
As part of the transition to civil rule, a Census was conducted in 1973. The provisional figures showed that the Northern states accounted for 64.4 percent of the total population (51.4 million out of a total count of 79.8 million people). This resurrected old fears of regional domination. The results were cancelled in 1975 by the administration of General Murtala Muhammed (1938Muhammed ( -1976, which had come to power through a military coup that year. There was no Census until 1991.
Thus, the history of Nigerian Censuses has been the history of political controversy and mutual ethnic mistrust. It is against this background that an assessment of the technical and political aspects of the country's most recent census is important for the lessons it holds for the next census and overall development planning in Africa's largest democracy.

The development context of the 2006 Census
The Obasanjo years were characterized by major political and development reforms in Nigeria. In 2003, the administration privatized the country's four oil refineries, and instituted the National Economic Empowerment Development Strategy (NEEDS), a domestically designed Poverty Reduction and Growth Strategy for fiscal and monetary management. Years of military dictatorship and pre-vious civilian rule had failed to diversify the economy away from its over-dependence on the capital-intensive oil sector, which provided 20 percent of Gross Domestic Product (GDP), 95 percent of foreign exchange earnings, and about 65 percent of budgetary revenues. The subsistence agricultural sector did not keep pace with accelerated population growth and this gave rise to the food importation culture that is a main expression of Nigeria's current food insecurity. From 2004, the government instituted marketoriented reforms. It modernized the banking sector, curbed inflation, and resolved regional disputes over distribution of earnings from the oil industry. GDP rose strongly in 2005, based largely on increased oil exports and high global crude prices. In November 2005, Abuja won Paris Club approval for a historic debt relief deal that, by April 2006, eliminated 25 billion of Nigeria's total 30 billion euros of external debt. These reforms were instrumental to the return of investor and donor confidence in Nigeria. Multilateral support for the Census was an expression of this renewed optimism but it was also based on perceptions of the Census as playing an important role in promoting democracy. As such, support for Census 2006 strengthened global interest in good governance with particular reference to the eighth millennium development goal (MDG) of developing a global partnership for development.

Technical aspects of Census 2006
The main strengths of the 2006 Nigerian Population and Housing Census included adequate planning, funding and technical support for its implementation as well as the supply of scientific expertise, including comprehensive independent monitoring of the Census processes by an External Monitoring Mission constituted by the European Union. It had the additional advantage of substantial publicity of the event and massive stakeholder support and public enthusiasm.
In order to mobilize financial and technical resources for the census, the NPC, in collaboration with UNFPA, organized a donors' conference. The Federal Government of Nigeria (FGN), the EU, the Department for International Development (DFID), the United Nations Development Programme (UNDP) and other development partners made commitments to assist the Census in various ways. The EU made the largest external contribution in support of the Census to the tune of 116.3 million euros. The Nigerian Government allocated to the Census the equivalent of approximately 118 million euros in capital expenditure. It paid the salaries and allowances of over 6,000 National Population Commission (NPC) regular staff and bore the costs of running all NPC offices -i.e. NPC Headquarters, 37 state offices, and 774 LGA offices. DFID committed assistance in the area of satellite imagery, mapping hardware and software and technical assistance in cartography to the tune of 10 million pounds. UNDP not only implemented the major component of the EU Census support, it also provided financial and technical assistance of 3.1 million euros to the Census project 2 . Indirect support was received from government agencies like the Independent National Electoral Commission (INEC), which provided over 160 vehicles, National Postal Services which provided storage containers, the Police and National and State Security Services. Logistic and financial support to NPC and its functionaries was provided by all state and LG authorities.
The EU, which was aligned with the National Census Plan, covered: payment of local costs for training and fieldwork for Trial Census, Census and Post Enumeration Survey (PES); procurement of machine-readable questionnaires and data capturing and processing technology; training and capacity building for National Population Commission staff to undertake the Public Perception Study; support for transparency and accountability measures Funding of national Civil Society Organizations for monitoring of the Census; funding of External Monitoring Missions; and EU Visibility activities and other technical assistance. Payment of allowances to 780,000 ad hoc personnel spread over the entire country was implemented by UNDP.
With regard to the organizational framework of the Census, preparations began in 1999 with the redeployment of technical staff to the then NPC office in Lagos to prepare for a census proposed for 2001 under the UNFPA Chief Technical Advisor, and the former military board of the NPC. The draft census project strategy document was developed in December 1999 by NPC staff with support from UNFPA. Even after the military board was dissolved and before the new civilian appointed board was constituted in 2002, preparations continued under technical direction of then Director General. Several missions from the UNFPA Country Support Team (CST) from Addis Ababa assisted to develop the draft consistency checks and edit specifications, publicity and advocacy strategy, methodology for census mapping, and data processing strategy.
Nevertheless, there was insufficient political will for a census in 2001. In 2003, the Obasanjo regime decided to fully support the census rescheduled for 2005 3 . Activities picked up with internal and external funding. Consultations intensified with census experts, civil society organisations (CSOs), community-based groups, political leaders, policy makers, researchers and the general public. United Nations Population Fund (UNFPA) and EU missions were invited to scrutinize the Census plans and provide input in the revision of the Strategy Document developed by NPC staff (Federal Republic of Nigeria, National Population Commission, 2005) . UNFPA continued to provide the Census Chief Technical Advisor (CTA) while the EU recruited Advisors on Census publicity, information Technology (IT) and Census cartography. DFID recruited an Advisor on OMR/OCR technology. The EU also instituted a series of monitoring missions to enhance the management and implementation of the Census at specific milestones of the Census, including Baseline (before the agreement to support the project); Enumeration Area Demarcation; Pre-test and Trial Census; final preparations; census exercise; Post Enumeration Survey; data processing; and dissemination and publication of Census results.
The Census Technical Group (headed by the NPC Director General) and the Action Plan Committee (headed by the CTA) were established to guide the process. Due to the socio-political challenges historically attached to the Census process in Nigeria, there was need for an effective nationwide communication strategy to educate citizens about the purposes and value of the Census, as well as provide advocacy to specific groups who might be wary of the exercise. The National Publicity Committee was launched, followed by its State Publicity Committees in mid-2004 in all 36 states and the FCT. Members were drawn from public and private sectors and included members of the National Orientation Agency, National Union of Journalists and Teachers, the National Council of Women's Societies, the News Agency of Nigeria, the National Television Authority, and the Federal Radio Corporation of Nigeria. At the grass-roots level, LGA Publicity Committees were inaugurated by the Federal Commissioners in October-December 2004. The LGA committee had the LGA Chair as its committee chairman and the NPC Comptroller as secretary. Its membership included the Divisional Police Officer (DPO) and a representative of the Traditional Rulers' Council. The committees were equipped with publicity materials and held seminars and rallies in senatorial constituencies. Local town criers were used to carry Census messages while market rallies were conducted to present the 2006 Census to market traders and the public.
The use of the GSM cellular telephone network was an innovation of Census 2006. The three major GSM operators in the country at the time disseminated Census text messages to the general public at regular intervals close to Census date. NPC Headquarters and State offices all had GSM hotlines which were available to the public for information and enabled people to report where enumeration had not been done in order to ensure maximum coverage.
Census messages were designed in English and translated into Hausa, Yoruba, Igbo, Ibibio, Efik, and pidgin. They highlighted the importance of being counted, offences related to the Census exercise, sample questions to be asked during enumeration, appeals against migration, and explanation that Census was not a tool for taxation. These messages were disseminated via television, radio, posters, leaflets, billboards, newspaper adverts, and interpersonal media.
In the weeks approaching 21 March 2006 (when actual count was scheduled to start), mass media coverage of the Census increased. Newspapers carried a corner counting down the number of days to the Census date, and ran polls to assess general awareness about the Census. Once the actual count started in the field, articles and stories about the Census shifted to operational problems of the exercise. Although there was resistance from some sectional interest groups in the Niger Delta and parts of the South East, who used the Census as a platform for highlighting pre-existing ethnic agendas, on the whole the nationwide communication and advocacy campaign by NPC was effective. Ninety percent of Nigerians polled by The Guardian were aware of the Census and 91 percent appreciated that Census taking was vital to national development.

Census organizational structure
Overall responsibility for planning, preparation, implementation, coordination, monitoring and evaluation of the Census rested with the NPC Headquarters. State offices of the NPC were responsible for State-level Census activities. The Federal Commissioner in the State and the State Director were in charge of administrative and technical issues. Local Government Area (LGA) offices were responsible for all Census activities at LGA level. They provided Census materials to Census functionaries in the field and retrieved them after the enumeration. An LGA office was headed by a Comptroller who reported to a State Director and the Federal Commissioner of the State. Census field functionaries included Coordinators who led teams of about four Supervisors during fieldwork. Coordinators reported daily to their LGA Comptroller 4 . Supervisors were responsible for five teams of enumerators (i.e. Census implementation in at least five Enumeration Areas). They ensured the smooth progress of enumeration. They established good working contact with the Police, civil and village heads, and other authorities. Supervisors carried out control visits in each EA, to check questionnaires for completeness and accuracy, to ensure that Enumerators visited all dwellings in their EA and to sort out all emerging problems. They remained in close touch with their Coordinators during the field work period 5 .
The Enumerators were key functionaries on whom depended the reliability of the Census data because they were the ones actually conducting the enumeration. Each team of enumerators comprised two people (usually one female and one male) and were supposed to have one EA assigned to it 6 . Their main field tasks were to check their EA maps to ensure that their boundaries were discrete. They also conducted house numbering and household-listing and updated the previous listing information where necessary. During enumeration, Enumeration Teams (ETs) completed all Census documentation on all households in their EA and within households they covered all household members. The completed documents were then submitted to Supervisors for checking and editing.

Cross posting of key staff and field functionaries
Cross-posting of key Census staff was done in order to reduce the risk of Census manipulation from NPC staff 7 . Senior management of NPC had been aware that if political manipulation of the Census figures occurred, it would be with the collusion of NPC regular staff, or some key fieldwork staff like Supervisors. It took steps to cross-post these functionaries at five levels, viz. Federal Commissioners; State Directors; Heads of Data Processing Centres; Comptrollers (transferred to different LGAs within the same States); and Supervisors. Apart from the Comptrollers, all these functionaries worked not only outside their usual states of work, but also outside their states of origin. The Federal Commissioners remained in their States of origin to ensure the recruitment of Census functionaries and the compilation of the final "frozen" list of functionaries before moving to their cross-posted state to supervise the rest of the census preparations. They were finally cross-posted in December 2005 and most resumed at their cross-posted state only in January 2006, barely two months before the Census. Comptrollers were cross-posted from their usual LGAs office to another LGA in the same state. The move may have weakened the connection between them and the local authorities but, again, it occurred late into the process. Many Comptrollers did not have time to get acquainted with the geography or politics of their new LGA as they arrived only a few weeks before the Census. Funding for transportation of Census functionaries and materials did not arrive from Headquarters at the beginning of the Census when it was most needed. Consequently, Comptrollers relied on Local Government Area Chairmen and other local leaders to fund some aspects of their logistics. In our view, cross-posting should always be accompanied by financial autonomy and institutional guarantees of transparency and accountability to reduce the risk of staff interference with census processes.

The census instruments and census tests
The 2006 Nigerian Population and Housing Census, which took place from 21-27 March, obtained information on the total number of persons in Nigeria, age/sex and occupational distribution of the population, its levels of literacy, employment and unemployment, as well as the current stock and condition of housing, access to water, electricity and other social amenities. Its methodology, instruments and logistics were pre-tested prior to enumeration. A pre-test in April 2005 revealed weaknesses in preparation. Recommendations for improvements were addressed in the weeks that followed but the Trial Census (29 August-2 September 2005) based on a 5 percent sample of Supervisory Areas (SAs) showed serious logistical and management flaws. The Census was rescheduled from the widely advertised November 2005 to March 2006 to allow for improved preparations. Census 2006 was an innovation in the history of Nigerian Census data collection and analysis. In addition to being the first housing census in Nigeria, it adopted the Optical Mark Reader (OMR) and Optical Character Recognition (OCR) method in the questionnaires. The information collected in the 4. National Population Commission Census Coordinators' Manual. 5. National Population Commission Field Supervisors' Manual. 6. Because cartography staff had carved out too many EAs during demarcation, enumerators were actually given more than one EA to canvass during the actual census, making verification and control difficult. 7. Cross posting was also done in some earlier censuses that have been contested; therefore, it is not clear how effective this has been in reducing political influence on the census. , not only to reduce errors and data processing time, but also as a way of reducing fraud by making it more difficult and, if it occurred, easier to detect. The forms and scanning technology were imported from overseas to guard against falsification (within Nigeria) and other manipulations of the instrument even before it went to the field. A German company was selected by open tender to supply the scanners for scanning the Census forms and a UK company was selected to print and deliver the OMR/OCR Census forms.
The questionnaires were state coded, except for the central reserve of 5 percent that could be dispatched to any state in an emergency. Each questionnaire had a unique number and barcode and specific colour density recognizable only by the Census scanners installed at the Data Processing Centres. Consequently, photocopied forms, even very good quality identical colour photocopies, could be easily recognized and rejected by the scanners, not only because of the repetition of the unique number and barcode, but also because of a different density of colour that was invisible to the human eye.
To facilitate scanning, the NPC01 form was designed to enable pre-coded responses that could be shaded in small boxes as in a lotto ticket. This gave a dense appearance to the questionnaire, which was unfriendly to the human eye. This may have inhibited the checking and spotting of errors during the implementation of the questionnaire in the field. Errors in handwritten responses and codes were easier to spot than errors in the shaded boxes mainly because of the small size of the shaded boxes.

Questionnaire length and content
Since the NPC01 form only contained basic demographic, socio-economic and housing characteristics, the length of the questionnaire was modest in comparison with Census questionnaires in other Censuses (for example South Africa and Lesotho). The NPC01 form did not have questions on fertility and mortality. These were transferred to the PES questionnaire. Our observation is that a short questionnaire is less likely to fatigue the interviewer and interviewee. Accordingly, NPC01 could have been less prone to error than lengthier instruments.
Although the wording of the questions in the NPC01 form was in line with standard Census questions, there were a few design flaws. In some cases, the categories of answers provided were all inclusive. For example, in Q11 (Previous Residence), those who had not moved could not be correctly recorded because a "Not Applicable" category was not included in the design of that question. In Q18 (Occupation), it was difficult to distinguish between "Non-response" and "Not Applicable" whenever that column was left blank. This had implications for the accurate processing and analysis of the Census information.

Contradictions of a de facto Census
Most African countries undertake a de jure Censusone which enumerated the usual residents of a household. Owing to Nigeria's Anglophone colonial census legacy, the sensitive nature of Nigerian Censuses (to which we have drawn attention earlier in this paper), and particularly because of the risk of over-count, the enumeration procedure for Census 2006 was the de facto method -which enumerates persons physically found present in an area on Census night. In Census 2006, every household member was enumerated on a face-to-face basis. Adults were allowed to provide information for younger household members provided these persons were physically present during the enumeration. Thumbprints of each enumerated person were taken.

Absentee heads of households and other issues
Errors emerged from listing "Absentee" heads of households (HHHs), who were the focal persons in NPC01. Next to every person's name and sex, an Enumerator was required to record his/her "relationship to the Head of Household". The codes for this "relationship" included Head of household (only one can exist in a household); Absentee head of household; Spouse; Child; Parent; and Other relative. The HHH was usually the first person to be enumerated and received the entry code number 001. In principle, this person was an adult: male or female. All other members of the household were listed after the head as 002, 003, etc. Unlike most other Census questionnaires, the Nigerian Census form made provision for listing and enumerating an unseen "Absentee Head of Household" despite the overall de facto character of the Census design. The code "Absentee Head of Household" was a potential source of double count at the analysis stage. The absentee head could be captured as a "ghost" HHH at another household where he did not spend Census Night. A polygynous man married to three wives in different localities, and who spent Census night with only one of them, would be identified with a name and a number, enumerated and have his fingerprints taken once in that household. However, the same person would be identified as "absentee head of household" twice in the households of his other wives or concubines, with his name and a number. He would then be counted among the total number of persons in these two other households although there would be no fingerprints for him. Now, total persons in a household was recorded on the first page of the household form NPC01 and on form NPC06, which summarised this for the entire EA. Analyzing the summary data presented on the Census forms without deducting these "ghost household heads" could lead to over-count of polygynous Nigerian males. This may not have been the intention of the designers of the questionnaire, but it was a direct consequence of the inclusion of that category. Because the designers did not specify in the question on marital status whether each married person was in a monogamous or polygamous relationship, it was not possible to obtain information on nuptiality dynamics through this quicker way or to deduct multiple polygynous HHH counts. Correspondingly, the "Absentee head of household" phenomenon grossly underestimated the percentage of female headed households and, since "Absentee head of household" is not a standard classification, reduced comparability of Nigerian Census data with data from other countries.
Beyond these potential sources of errors, whenever "absentee head of household" was identified by question N° 3, Enumerators were instructed not to ask the remaining questions N° 4-19 because the person was not present on Census night. But this instruction was not specified on the questionnaire as a skip pattern and, so, was often not implemented by fieldworkers. Review of the Trial Census data by the authors indicated that Enumerators did not heed the instruction not to fill in other information (name, age, sex, nationality, origin, place of usual residence, etc.) for the absentee head of household. Many Enumerators went ahead to fill in these data for someone they did not see. In theory, the edit specifications for processing the Census data could be designed in such a way that as soon as a person was identified as an absentee head, then all of the "extra data" would be deleted so as not to count an unseen person in a de facto Census. 8 Even if the Enumerator had indeed heeded the instruction and did not collect supplementary information on the "absentee head of household", the lack of information in questions N° 4-19 for this vital focal person around which the relationship of the entire household members was statistically and sociologically constructed weakened the whole edifice of relationships, making it impossible to validate data on other household members linked to him, as there was no reference data on "absentee head of household" for comparison. For example, if an edit check were to specify that the age of a child must be at least 15 years less than that of the HHH, this check cannot be undertaken with an absentee head of household because he would have no age data.
The review of the Trial Census data revealed that that Enumerators often entered a code of 1 for head of household and then "automatically" coded 2 for "Spouse" although, in the questionnaire, "2" was the code for "absentee head of household" and "3" the code for "spouse." Consequently, a non-negligible proportion of households in the Trial Census were recorded as having one HHH and one "absentee household head" at one and the same time. In most of such cases, it was the spouse who had been incorrectly coded as "absentee head of household." With respect to age of respondents, enumerators were trained on the use of a Historical Events Calendar but they seldom had these issued to them. They were thus faced with difficulties when estimating the age of adult respondents. These difficulties are not unique to Nigerian censuses. In rural areas, where oral (as opposed to written) tradition is the main source of information, adults had only vague ideas of their age and some would often report a much higher or lower age than their actual age.
Religion and Ethnicity were not captured by Census 2006. After heated debates in the press and in data users' workshops, when the controversial nature of questions on ethnicity and religion became evident, NPC made a submission to the Federal Government on the merits and demerits of including them in the questionnaire. The National Council of States (comprising the President, Vice-President, former heads of state, governors and traditional rulers) decided to remove these questions from the questionnaire.
In the view of NPC Chair Samaila Danko Makama (2006), "The problem with Nigeria is that 8. The authors did not have access to the consistency checks and edit specifications used by NPC.
we have tended to place more premium on using Census data for revenue allocation than on planning for sustainable development." In his reasoning, since religious and ethnic groups would prefer numerical superiority over each other, it might be safer to ignore religion and ethnicity altogether to reduce the temptation for each group to explore ways of having a competitive edge through the census.

Political environment of Census 2006
The political aspects of the 2006 Nigerian Population and Housing Census can be examined from the context of constitutional provisions that contribute to a politicisation of the exercise. These provisions include the bases of political representation and revenue allocation as well as the composition of the management board of the NPC. While it is clear that the intention of the formulators of the 1999 Nigerian Constitution (and the 1979 Constitution before it) was not to politicise the country's Censuses, the provisions invariably created both an awareness of the possibility of politics and conditions for the politicisation of the process.
For many Nigerian social scientists, with the exception of elections, "population Census has remained the most sensitive and controversial issue in the politics and administration of the Nigerian state" (Mbeke- Ekanem, 2006). While participating in the 2006 Census exercise in Kano, Governor Ibrahim Shekarau of Kano State enjoined Nigerians to consider the exercise as one of the prerequisites for ensuring the growth of the country instead of seeing the head count as an opportunity to outdo one another over numerical strength (Guardian, 22 March 2006: 8).
The politicisation of the Nigerian Census is thus rooted in the national discourse of even development and federal character. With regard to political representation, as noted, the 1953 Census established and validated the practice of allocating seats in the Federal Legislature on the basis of regional population size. The 1954 Lyttleton Constitution allocated seats in the Federal Parliament according to its figures. The Northern Region was allocated half of the seats in the House, while the other two regions shared the other half equally. This was the genesis of national Census rivalry.
The preliminary results of the first post-independence Census were rejected by government without allowing the census process to reach the stage of checks and validation to produce the final results. The head of the Census Team, Mr. J. Warren, was fired by the Prime Minister when it recorded a slight numerical majority for the South. Had they not been rejected, the results would have required a reapportionment of seats in the Federal Parliament and entailed a strategic loss of political advantage to the Northern People's Congress.
The Prime Minister ordered a recount in 1963, with himself as head of the National Census Board. That recount put the population of Nigeria at 60.5 million. In the new figures, the North had a population that exceeded the South's by 8.5 million. The national total appeared too high to Census officials and was pruned to 55.65 million, with a reduced numerical majority (4 million) for the North. The figures were not accepted by the Premiers of the Western and Eastern regions and the results of that Census formed part of a mounting crisis in the Western Region that ultimately led to the succession of military coups d'état in Nigeria. Every Census since then has taken place in an atmosphere of deep rancour among the different states and ethnic groups in the country. The allocation of seats in the Federal Parliament was the motivation for a large regional Census figure in the 1963 Census and this was pursuant to regional interest in revenue allocation. The struggle for allocation became a serious issue with the discovery of oil. The independence constitution favoured a 50 percent derivation formula in revenue allocation, while the other 50 percent went to the federal government.
With the discovery of oil in the late 1950s and the Civil War in the 1960s, the Yakubu Gowon military administration in 1969 gained federal control of oil revenue generated in the country and developed a revenue allocation formula that was, in addition to other factors, based on population size. As part of the transition to civil rule, Gowon called for a Census which was organised in 1973. The massive overcount of this Census, which, with a total population of 79.8 million, gave the north a numerical majority of 23 million, was an invitation to fresh political instability only a few years after a bitter civil war. The government immediately cancelled the results.
The belief that federal budgetary allocations are directly linked to population size promotes a strategic interest in inflating figures. As a way out, Lles Leba (2006) proposed that the direction of revenue flow from the federal government to state and local governments be reversed such that fiscal contributions to the federal treasury by the latter two tiers of government are based on the respective populations in each area. This approach would reduce the motivation to increase population figures, as a large population would imply an increased burden of expected monthly remittance by the State government to the Federal treasury. This recommendation unfortunately assumes that a positive relationship exists between fertility-induced population size and productivity and that it is equitable to impose higher taxes on areas of high population.
Arguments like this do not address the inter-ethnic mix of cities like Kano and Lagos, whose large population size is less the result of fertility increase than internal migration. On the cultural front, revenue allocation and political representation are not significant drivers of fertility in comparison to historical conditions of high infertility rates, pregnancy wastage, infant and child mortality and the social expectations of families (Obono, 2001). While Leba's proposal might indeed serve as a disincentive to Census over-count, it could just as easily lead instead to an undercount as shown in the colonial enumerations conducted before independence (Locoh and Omoluabi, 1995).

Composition of the NPC Management Board
The 1999 Nigerian Constitution (Item 8 of the Exclusive Legislative List) empowers the National Assembly to legislate exclusively on Census. Part 1 of the Third Schedule contains Item J paragraphs 23 and 24, which invest the NPC with the responsibility of conducting periodical enumeration of population and prescribes its composition and powers. Section 153 lists the National Population Commission as one of the Federal Executive Bodies and guarantees its independence and autonomy in that regard, while Section 158 insulates it from Executive control. Section 213 of the Constitution describes the procedures for treating the report of the NPC on Census. There are two principal constitutional provisions on the NPC, which contributed to politicising the 2006 Census. The first is the provision that the Population Commission shall comprise a Chairman and one person from each State of the federation and the Federal Capital Territory (FCT), Abuja (Item J paragraph 23 of Part I of the Third Schedule to the Constitution). This provision conforms to the Federal Character requirements of Section 14 (3), but it also injects political factors into the composition of the Commission charged with conducting a scientific Census for the country by causing members of the Commission to view their membership in terms of political representation of their States/Regions. In a bid to advance diversity and inclusiveness, this Constitutional provision transforms the Commission into a regular political board 9 . The Board is appointed by the President in consultation with the Council of State and confirmation by the Senate [S154 (1) (3)]. NPC board members are thus political appointees and not technocrats per se. It is notable in this regard that the 8-member Board of the Population Commission responsible for the 1991 Census had no such political alliances. The seven Federal Commissioners and Chairman of the previous Board were appointed on professional merit. They were in charge of zones which were defined by geographic workload and not regional representation.
By allowing the appointment of one board member per state, the constitutional provisions anticipated political interference in the Census process. In doing so, however, they created conditions for that interference even while trying to avoid it. In the next sections, we will examine the processes of the 2006 Census that were the immediate contexts and entry points of much political influence. These include the enumeration area demarcation (EAD) process, staff recruitment, Census forms distribution, local logistical and financial support, and the enumeration itself.

Enumeration area demarcation
Nigeria has a total area of 923,768 km 2 of which water takes up 13,000 km 2 . For this reason, EADwhich is the first major field activity in a Censuswas an enormous challenge in the build up to Census 2006. Its main objective was to sub-divide the country into clearly defined small geographic units known as Enumeration Areas (EAs) that an Enumeration Team (ET) was expected to cover during the Census period. At the end of demarcation, a list of EAs, called an EA frame, was produced along with the EA maps and EA codes, in such a way that when placed side by side, no land area in Nigeria overlapped or was omitted. A correct EA frame is the foundation for a Census. It facilitates efficient sharing of the enumerators' workload and constitutes the frame for reporting the Census results at the lowest, verifiable level.
By May 2005, NPC's funding was in poor shape. A sum of 1.3 billion naira (7.9 million Euros), which was outstanding from the 2004 national budget, had not been released 10 . No disbursement from the approved 2005 budget was released and, after exhausting all of its financial reserves, the Commission was slowly grinding to a halt. This situation had implications for the EAD in various ways. Contractors had not received any payments and had stopped 9. By contrast, the compositions of the Independent National Electoral Commission (Chairman and 12 others), the Federal Civil Service Commission (Chairman and 15 others), and the Code of Conduct Bureau (Chairman and nine others) do not reflect State/Regional representation in precisely the same way as the NPC board does. 10. Cash backing had not been received for the 1.3 billion naira outstanding from the 2004 budget allocation by close of accounts. However, since the Due Process Certificates had been obtained, a special dispensation authorized the release of these funds from the 2005 budget allocation while exploring other sources for these funds.
supplying materials for EAD and other preparations. The Abuja headquarters as well as all state offices had not received any funds for running costs for several months. Arrears of staff allowances since 2001 had still not been paid and the workers' union was threatening a nationwide strike action for 2 June 2005. Demarcators had not been paid from March to June 2005. They depended on the goodwill of local authorities for food, accommodation and stipend. In Kano State, demarcators were accommodated and provided with two meals per day by the Local Government. In Lagos State, they received a subsistence allowance. Many other State and Local Government Authorities provided direct and indirect assistance to demarcators. Under these circumstances, the NPC could not reprimand its 8,200 fieldworkers for not doing their demarcation job properly. Enumeration area demarcation in Nigeria was undertaken nationwide from July 2004to November 2005 Owing to the late arrival of funds, most of the demarcation work was compressed into just 10 months (from February to November 2005) --seven LGAs in Phase I (July 2004); 37 LGAs in Phase II (September 2004);and 730 LGAs in Phase III (January-November 2005).
The External Monitoring Mission acknowledged the critical importance of State and Local Government support for the continuation of the demarcation exercise but warned that this brought with it a real risk of compromising the allegiance of demarcators during EAD. This support from local authorities for the demarcation exercise came at the price of a 212 percent increase in the number of EAs. The 1991 Census EAD demarcated 212,072 EAs (NPC, 2004: 2-3). The Census Strategy and Implementation Plan (August 2004) estimated that about 350,000 Enumeration Areas (EAs) and 70,000 Supervisory Areas (SAs) would be delineated during the EAD exercise for the 2005 Census (before it was postponed to March 2006). The figure of 350,000 EAs was computed from the projected national population of 134 million by 2005 derived from the 1991 Census, and the plan to demarcate EAs having from 250-500 persons in rural areas and 400-500 persons in urban areas.
While senior NPC management at Headquarters was aware of this strategy, State NPC offices were not provided with an "expected number of EAs" within which to base their work. They simply went ahead, and with the support of local political authorities, demarcated areas without following guidelines for EAD and, more crucially, without sufficient supervision from Headquarters. At that time, the financial and logistical implications of doubling the number of EAs provided for in the Census Strategy Document were not appreciated by the Cartography Department or by NPC management.
A verification exercise undertaken in May 2005 showed that many fieldworkers were over-estimating the population quick counts and producing too many undersized EAs, especially in political "flashpoints" like Kogi State, which had carved out an impressive number of EAs by overestimating population size in the quick counts. Some local communities had confused Census EAD with electoral delineation. Within this setting, they reasoned that by carving out more EAs, they would increase the number of their electoral wards and their chances of political representation. Reinforced by a combination of poor training and lack of supervision of fieldworkers, the resulting explosion of EAs was inevitable.
By September 2005, about 436,367 EAs had been demarcated for 71 percent of LGAs. By the time EAD was completed, 662,180 EAs had been demarcated for the 2006 Census. This was three times the number of EAs carved out in the 1991 Census. While NPC management was expecting a 65 percent increase in EAs from the 1991 exercise, they were quite unprepared for the 212 percent increase in EAs that was produced for the 2006 Census.
The main question as census night approached was: Who will foot the bill for the extra EAs? The original strategy of one EA per Enumeration Team meant that 662,180 EAs would be canvassed by 1,324,360 Enumerators, 132,436 Supervisors and 33,109 Coordinators, making a total of almost 1.5 million fieldworkers, an increase of 89 percent from the original 787,500 fieldworkers already budgeted for in the Census Strategy Document.
The NPC realised that it lacked the financial and logistic resources to manage this number of fieldworkers. "Right-sizing" became the term used to describe how to assign the 662,180 generally undersized or over-estimated EAs into contiguous units for the original 350,000 ETs and to ensure that an Enumeration Team would be allocated sufficient work for the Census enumeration period. This rightsizing was undertaken from November 2005 to February 2006. Although grouping 662,180 EAs together for 350,000 teams at such a late stage in the Census process was a vast and risky operation, the NPC succeeded with right-sizing the Enumerators' workload but at a cost for adequate retrieval of census control forms and maps. The lessons learned are clear. For such a crucial and fundamental exercise as EAD, upon which the entire Census depends, adequate preparation, funding, implementation, supervision, and assurance of autonomy are needed in subsequent Census operations.

Recruitment of census functionaries
Recruitment of functionaries (Coordinators, Supervisors and Enumerators) was a major weakness of the 2006 Census. Although the EU funded the payment of the Census functionaries, recruitment remained the responsibility of NPC, which also had to certify functionaries before they received EU payment for Census work. In most developing country settings, teachers and other civil servants are the desired Census functionaries because they are well respected in their communities, have a reputation at stake and are easy to track down through government records. The Federal Commissioners of NPC decided against the recruitment of teachers and civil servants as functionaries citing the high rate of unemployment among Nigerian youth and the need to use the Census to alleviate this 11 .
Recruitment was removed from the office of the Federal Commissioner and implemented by the 774 LGA Comptrollers as a means of insuring it against political interference. But all Comptrollers were under the strict supervision of Federal Commissioners, and the Commissioners were political appointees. In this context, employment of functionaries was seen by many as a way of fulfilling a major role expectation in their State constituencies. In the event, cross-posting of Federal Commissioners did not affect this vital aspect of the Census as all recruitment was completed in the States before Federal Commissioners were cross-posted at the end of December 2005.
Lead stakeholders reportedly pressurized Comptrollers to accept their candidates even after list compilation and submission deadlines had passed. Enumerators were chased away from Kpambai, Jenuwan Kogi, Bika, Ussa and Takum LGAs in Taraba State because they were not indigenes of the state. These LGAs complained that their people were not given any political appointments by the Government (Daily Independent, 6 April, 2006). . Officials of the NPC in Osun were assaulted in Ifetedo when attackers stormed the NPC office in protest against the non-employment of indigenes of the community as Census functionaries (This Day, 21 March 2006) .
Lists of successful candidates were posted at LGAs in the first week of March, 2006 but there were problems with the lists. Supervisors (who should have been cross-posted, i.e. undertaken Census supervision in a state other than theirs) constituted the full membership on Supervisors' lists. A memo from the authorities reaffirmed an earlier position on cross-posting, but it came only after protests at NPC Offices had begun to spread around the country. Members of the National Youth Service Corps (NYSC), who ought to have comprised 50 percent of the bona fide Supervisors, made allegations of serious nepotism and corruption in the recruitment exercise. They claimed the existence on the list of what they described as "Ghost Corpers" 12 . They had no confidence in the lists. The development conformed with the widespread view that: In Nigeria, connections confer privilege and impunity in a prendalist regime that is characterized by patron-client relations rather than due process in the provision of basic services.
[Connections] pave the way for one's entry into networks of trust, a compromise that may occasionally shield criminal activity and subvert due process (Obono and Obono, 2012: 237). Since the Census was organized less than a year from scheduled national elections in 2007, recruitment was possibly vulnerable to political interest. Politicians used the opportunity to provide temporary employment to their youth wings. The pressure on Comptrollers to recruit names sent to them by high-ranking officials and members of the political elite was irresistible. It was at the level of Comptroller that much of the reported and observed challenges of the recruitment and enumeration exercises of the census occurred.

Funding and support from local authorities
Local government authorities assisted the Census by making vehicles, motorbikes and guides available to functionaries. The Census depended heavily on this support because funding voted for transportation and other critical census activities did not arrive when it was most needed, especially at the beginning of Census. Some LGAs paid functionaries daily or monthly feeding and/or transport allowances -as had been the case during EAD. Some LGAs gave as much as 30,000 Naira to Supervisors and 10,000 Naira to Enumerators per month. The Comptroller coordinated this process. While UNDP disbursed the official EU payment to functionaries, the Comptroller was responsible for paying out the sums granted by the State or Local Government. Because of the lack of transparency in 11. As political appointees of the ruling party, this choice gave Commissioners an opportunity of doing something for their constituencies and the connection with the LGA Chairmen then appeared "seamless". 12. "Ghost Corpers" was the term used by members of the NYSC to described fictitious individuals whose names appeared on the lists of NYSC Supervisors, although such individuals were not NYSC members.
the disbursement from local authorities, functionaries did not often know how much they were meant to receive or when. There were reports that some Comptrollers paid less than the stipulated amount of LGA support, while some fieldworkers said they did not get any payments at all. It was difficult to corroborate these claims as there were no records of what came to be described as "hospitality packages." The dependency of NPC on LGAs for "hospitality" to its functionaries and the freedom with which this was defined by the authorities posed risks to the integrity of Census data collection. Indeed, the observations of the external monitoring mission showed that the Comptroller was the weakest link in the Census 2006 process. It is at this level that subsequent censuses could require the greatest inter-agency and global partnership collaborations. NPC Comptrollers were at the heart of the distribution process at the LGA level. They were vulnerable to political pressure in the distribution of Census forms, especially from 23 March when "shortages" began to be reported nationwide. In this setting, the relationship between an LGA Chairman and a Comptroller was a patron-client relationship. According to one report, The cross-posted National Population Commissioner in charge of Sokoto State, Otunba Okanlawon, reportedly "decried a situation where Comptrollers who were employees of the Commission behaved as if they had abdicated their responsibilities to Council Chairmen." He decried what he called the direct involvement of some of them with local council officials, adding that in some instances, Council Chairmen spoke to the press as if they were the people conducting the exercise and not the NPC (Meya, 2006: 8).
This situation had ramifications for reports of shortages of census materials at the local government levels. These "shortages" occurred despite the distribution of materials according to the most elaborate plans by NPC Headquarters.

Logistic "shortages" of census materials
Sufficient forms were supplied for the Census exercise. Initially, 25 million NPC01 main questionnaires were estimated for the Census in the Strategy document (NPC, 2004). This was increased to 35.382 million forms by the Chief Technical Adviser. It was finally increased by another 5 percent (1.354 million NPC01 forms) to 36.737 million to make allowance for a central reserve to cater for emergency shortages. By 12 March 2006, 95 percent of the questionnaires had been delivered to the State offices of the NPC. Given the significant challenges that arose in the course of procurement, the timely delivery of the questionnaires was commendable. Bags and other Census materials arrived at state offices after Census activities had already started. There was no information on what precisely caused the delays.
By 22 March 2006, the third full day of field operations, some NPC State offices and State Governors started asking for additional NPC01 forms. According to NPC headquarters, this meant that either the State had completed all NPC01 forms provided within the period (an improbable achievement) or the alarm was being raised with a view to collecting more forms than required in the event they needed more. In Kebbi and many other States, artificial shortages were created by political leaders who "pressurized" Comptrollers and State Directors to provide their constituencies with more forms than was planned, leaving less forms for others.
States reacted to the scarcity of forms in various ways. In Nasarawa State, 1,000 NPC01 and 400 NPC07 forms were photocopied and distributed, and the Federal Commissioner was ready to make more copies if no new consignment of the forms was delivered. Massive photocopying of Census forms was also being done in FCT, Kano, Niger and Ogun before the Presidential directive asking this to stop. In some other states, functionaries used training versions of questionnaires while waiting for more live questionnaires to arrive, while others kept running between the field and any NPC office for fresh materials. In the second half of the Census week, four State Governors sent private jets to Abuja to personally collect Census forms from NPC Headquarters to palliate the reported shortage in their States. They returned with boxes of forms, albeit much less than they requested. The time that Census functionaries spent on travelling to State Offices for supplies or making contingency decisions and implementing alternative interventions (such as making photocopies) impacted on the efficiency of the Census operation.
A substantial part of the irregularities observed in the distribution of materials occurred at the level of the LGA offices. The well-articulated distribution plan for Census forms and materials sent to all Federal Commissioners, State Directors and LGA Comptrollers on 4 March 2006, collapsed at the LGA levels in all States 13 , partly due to pressure on the Comptroller from local authorities. 14 In Gombe, Ogun and Plateau States, inappropriate distribution of materials by the LGA Comptrollers was due to inadequate projections of materials needed in the different EAs. Post-fieldwork reports suggest that there was an element of hoarding of the NPC01, causing artificial shortages on the ground 15 . The phenomenon of hoarding census forms at the start of the census could have been borne out of fear of not getting a "fair share" of forms and being shortchanged by the census exercise. Unfortunately, it would seem that this led to a sufficiently high level of shortages in several states of the federation. Three weeks after the census enumeration exercise, every State (including those which reported major shortages) were returning unused NPC01 forms. It was estimated at the time that over 1 million forms were recovered unused.

Storekeeping systems
Inadequate storekeeping systems made it impossible to track the materials distributed and to predict shortages before they occurred. At each administrative level, actual material inflows could have been compared with expected material inflows as a kind of "early-warning system" on material shortages, giving the various officials the opportunity to request for additional materials before the shortage actually occurred. Such a system would also have offered management protection against unjustified demands from the field for additional materials and helped to track any wrongful allocation. Perhaps this would have been an excellent means of holding officials accountable for any inadequate allocation of materials, thereby ensuring effective management of material flows. A Comptroller was arrested by police in Benue State for allegedly selling NPC01 forms for one million Naira. Benue NPC confirmed the theft of 18 cartons of NPC01 forms at its store in Makurdi Council secretariat 16 . The State office was not able to provide the serial numbers of the stolen forms but it did recover them after the Census. In the view of the NPC Commissioner cross-posted to Enugu State, Dr. Suleiman Bello, complaints of shortage of materials were a ploy to inflate Census figures 17 Against this background, it is instructive to note that the Census questionnaire made provision for recording a maximum of eight persons. A household with more than eight members will require a second or third questionnaire. Nigerian survey data indicate that only about 5 percent of households have exactly eight members. However, results of the pre-test conducted in 2005 indicated an over-representation of "maximum-line" 8-member households in at least 22 of the 37 States. Analysis of the Trial Census data showed the same tendency with an average household size of 6.2 members instead of 5.1 from comparative survey data. Samples of the 2006 Census data were examined at the Data Processing Centres and instead of the expected 5 percent, 8-member households constituted up to 67 percent of households in Katsina State and 100 percent of households in some EAs in Ogun State. This phenomenon of 8 person households (or Maximum Line Households, MLHs) has always been known to demographers at NPC and they have developed a methodology for addressing it. Nevertheless, the fact that it appears only in Nigerian census and not in survey data reflects how the Nigerian census produces some systematic data errors.

Conclusion
Much of the discussions and observations contained in this paper were the result of the authors' participation in an EU external monitoring mission of Census 2006. One of us did so for 36 months. We were on ground to observe the safe retrieval of 36 million questionnaires of the 2006 Nigeria Population and Housing Census from the field for data processing. We conclude that, like other Censuses since Independence, motivations for over-count in the 2006 Census -political representation and revenue allocation -were strong and institutional controls were often subverted by exigencies encountered by census functionaries in the field and this might have undermined data quality. There were, in addition to motivation, opportunities for manipulation of figures despite attempts by NPC to control this through cross-posting of key functionaries. These efforts did not always have the intended effects because of the poor timing of the cross-posting and inadequate planning and resource allocation.
The Comptroller was closest to the LGA authorities and, owing to inadequate resource allocation, was easily the weakest link in the Census 2006 process from Enumeration Area Demarcation, through recruitment of functionaries, distribution of field forms to the actual enumeration itself. It is at the comptroller level that subsequent censuses should focus strong inter-agency and global partnership collaborations. NPC Comptrollers were at the heart of the distribution and retrieval process at the LGA level. They were vulnerable to political pressure from local politicians at every stage of census activity.
The 1991 census is the most recent census conducted by a non-political, "technical" board. It employed that period's state of the art methodologies and technologies; it was assisted by renowned demographers, cartographers and IT specialists and had adequate financial resources from the government of the day. The results of the census were adjudged by neutral commentators and demographers, including one of the authors 18 as being fairly accurate. It is clear that a nonprofessional board consisting of political appointees of the ruling party does not provide assurances of political neutrality. The constitutional changes needed to transform the NPC Board into a smaller, more technical, and effective group is an area that will reveal the strength of Nigeria's political will in having scientific censuses. It will also be the context in which public understanding of the nature and uses of the censuses can be discerned. Without institutional and constitutional changes of this kind, the perceived connections between population size, on the one hand, and revenue allocation/political representation, on the other, will leave all future Nigerian censuses prone to political influence regardless of how technically sound their design may be.