Inconsistencies In Census Data: The Case of Kenya’s North Eastern Province Introduction Demographic data is very vital for all development planning in a country, because it provides detailed bench-mark data on all population characteristics. In developed countries, accurate data is available from effective and complete civil registration systems. However, in developing countries where the completeness and coverage of registration of vital events is below 40%, census and survey data are used to fill in the data gaps (UNECA, 2010). Census data, collected every 10 years in most countries, provides vital data at the national level and is the ultimate source of information at the sub-national levels since surveys have not been designed to provide information at these levels. It provides answers to: How many are e? in terms of the total number of people liing in the entire nation, ―Who are we? in terms of age, sex, education, occupation, economic activity and other crucial characteristics, as ell as Where do we live? in terms of housing and access to social amenities. As such, census data plays a key role in planning, administration and allocation of resources to meet present and future demands of the population, to improve their living standards, ensure sustainable development and progress of a nation. Despite the importance of census data, data consistency and credibility has been questioned in developing countries (Mba, 2004). For instance, in Nigeria, since the first post-independence census in 1962, there has been little consensus on accuracy of the results, and up to the 2006 Census, the population count was disputed by various groups in the country, sparking several riots (Ezeah et al, 2013; Sandra Yin, 2007). Some inconsistencies were also noted in the 2011 South Africa census data released by the statistical agency (Berkowitz, 2012). Kenya conducted its last population census in 2009, the 7th census to be taken in the country. During the release of the results a year after the census exercise, the statistics agency, Kenya National Bureau of Statistics (KNBS) recommended a recount in nine districts in the North Eastern part of the country that had shown massive inconsistencies in the data. However, politicians from the region came out strongly to refute the re- count, and to date there is a pending court case to determine whether the results can be published or the recounting exercise should go on. As a result, detailed analytical census and population projection have never been released, and the country continues to use inaccurate data for planning. The KNBS noted an over-count of about 1.1 million people, with over 95% of this registered in the North Eastern province. Against this background, this paper seeks to evaluate the nature of the data inconsistency in the 2009 Kenya census data, and to establish the extent of the irregularities with particular focus on the North Eastern province of the country, using Whipples, Myers and United Nations Age-Sex Score. Background The world over, data yielded by censuses have recorded various levels of completeness, due to several reasons some of which are biases arising from under-reporting in certain ages, double-counting of persons, persons absent on the night of the census and time location errors among others (UNFPA, 1983). These give rise to coverage and content errors, which vary both in nature and magnitude from one country to another. Coverage errors result from omission of certain pockets of the population, while content errors pertain to misreporting or misclassification of events. Kenya has carried out seven censuses; two pre and five post-independence censuses. These successful censuses have been held consistently every 10 years until the most recent of 2009, which recorded 38.6 million people. All these censuses have shown some degree of incompleteness, and data quality seems to have declined over time, with the latest census 2009 raising serious concerns. According to Christopher, 2010, incomplete census coverage results partly from highly varied census questions and 1 elicited responses. Consequently, quite often the subsequent tabulations do not allow for the monitoring of long-term trends. Use of smoothing techniques has been recommended to improve data quality. However, the extend of the irregularities in the 2009 census data could not be addressed by light smoothing methods, necessitating the use of strong smoothing techniques (KNBS, 2012). What could possibly explain this finding? Kenya ratified a new constitution in 2010 which recommended a devolved system of governance, resulting to the creation of 47 counties. This meant that national revenues would be shared to county governments to perform the devolved functions that were initially executed by the central government. Moreover, an additional 80 constituencies were to be created after the constitution was promulgated and before the 2013 general election. The criteria for revenue allocation and creation of the constituencies was heavily weighted on the population size of each county. The 2009 census data was thus a key determinant in amount resources received by each county. According to the commission for revenue allocation (CRA), the allocation formula was such that 45% of resources are allocated in accordance with population density, 25% by equal shares to cater for costs of running the county governments, 20% by poverty levels, 8% based on geographical size, and 2% based on fiscal responsibility. The North Eastern province has the smallest population size compared to other provinces in the country. To increase the resources allocated to the counties in the region, it is hypothesised that political leaders from the region muddied the data by telling communities to over- report at the household level. Evidence of inconsistency in the North Eastern Province For the preliminary findings, we highlight the irregularities pointed out by KNBS, the statistical agency in Kenya. In 2009, census count showed that Kenya had 38.6 million people, 10 million more than the 1999 count. This translates to an annual increase of one million people, and an annual population growth rate of 3%. However, as seen in Table 1, population size in the North Eastern province more than doubled (140% change) during this period, with an annual population growth rate of 8.8%, almost three times the national average. For the counties in that province, Mandera county population increased four-fold (317%), with an annual population growth rate almost five-times (14%) the national average. Table 1: Population Distribution and Inter-Censal Growth Rate for Selected Regions in the Country Enumerated Population 1999-2009 Region 1999 2009 % Change Growth Rate (%) Kenya 28,686,607 38,610,097 34.6 3.0 Nairobi Province 2,143,254 3,138,369 46.4 3.8 N. Eastern Province 962,143 2,311,259 140.2 8.8 Garissa County 262,694 623,060 137.2 8.6 Wajir County 309,268 661,941 114.0 7.6 Mandera County 246,063 102,5756 316.9 14.3 Source: KNBS, 2012 The pyramid below (Figure 1) shows the age-sex distribution of the population in the North Eastern province. Notably, there is an unexplained population bulge at ages 5-19, 60-64 and 70-74. This abnormal distribution is also visible in the calculated sex ratio for the province. As Figure 2 shows, there is a huge departure from the expected sex ratio distribution, with irregularities noted at all ages, reaching a maximum of 202 for those aged 60-64. This finding is corroborated by age ratios which 2 show high fluctuations and large departures from the expected ratio of 100, especially for populations above 65 years. Figure 1: Age-Sex population structure of North Eastern province, 2009 Population Pyramid, North Eastern Kenya 80+ 75-79 Male Female 70-74 65-69 Total Fertility Rate = 6.4 60-64 Total Population = 2,310,018 55-59 50-54 45-49 40-44 35-39 30-34 25-29 20-24 15-19 10-14 5-9 0-4 12 10 8 6 4 2 0 2 4 6 8 10 12 Source: Authors Figure 2: Sex Ratios by Age, Kenya and North Eastern Province, 2009 Source: KNBS, 2012 3 These results show abnormal growth rates which are not plausible given the fertility and mortality rates of the region. The North Eastern province has the highest fertility and child mortality rates countrywide. Therefore, the observed distribution would only be possible if there was a large net migration in the province. It is possible that refugees from war-torn Somalia were hosted by families in this region, which borders Somalia. This is however not supported by the data, with the North Eastern province being a net population looser due to out-migration. This therefore clearly points to deliberate over-reporting of members in a household. This is clearly evidenced by the fact that household size (the average number of persons per household) in the North Eastern province increased by 11 percent, while in the other seven provinces in the country, it either declined or remained constant. Conclusion The 2009 Kenya census data from the North Eastern Province reflects inconsistencies across board; population size, growth rate, age and sex structure which have serious implications on population planning. The collection, analysis and dissemination of accurate demographic information enable policy makers to plan for the future development of a country. Biased or defective census data leads to poor planning and targeting on the part of central government and development planners, which is tantamount to chasing the shadows. This is because it affects the credibility and usefulness of census data for sustainable development. With the worldwide data revolution to ensure monitoring of the sustainable development goals (SDGs), its paramount that data quality is ensured and promoted so as to provide a good basis for monitoring progress of achieving the SDGs.