DATA EVALUATION METHODS USED IN THE PREVIOUS CENSUSES

DATA EVALUATION METHODS USED IN THE PREVIOUS CENSUSES POST ENUMERATION AND DEMOGRAHIC ANALYSIS Gebeyehu Abelti Deputy Director General, Population & Social Statistics Central Statistical Agency Of Ethiopia Presented at United Nations Regional Workshop on the 2010 World Programme on Population and Housing Census: Census Evaluation and Post enumeration Addis Ababa, 14 -18 September, 2009

OUTLINE o o o 1 2 3 4 5 – – – INTRODUCTION CENSUS UNDERTAKING IN ETHIOPIA POST ENUMERATION SURVEY DEMOGRAPHIC ANALYSIS CONCLUSION

1 INTRODUCTION. . q Ethiopia is Federal state sub-divided into Nine Regional States and Two Administrative Urban Areas ; q Regional States are further subdivided into 78 zones and Over 600 weredas (districts), q Overall, in 11 Regional States there about 15, 000 rural and urban kebeles.

1 INTRODUCTION. . ADMINISTRATIVE HIERARCHY IN ETHIOPIA FEDERAL REGION ZONE WEREDA KEBELE

2 CENSUS UNDERTAKING IN ETHIOPIA q So far, the country has conducted only three National Population and Housing censuses : The First ever Population and Housing Census was conducted in May 1984 (39. 9 million); q The Second Population and Housing Census was conducted in October 1994 (53. 5 million); q The third/Latest PHC was conducted in May and November 2007(73. 9 million). q

2 POST ENUMERATION SURVEY PES was conducted only once; PES was undertaken in 1994 two months after the enumeration; o Pre-test before the PES was carried out in 13 EAs; o The general objective of the PES were to measure coverage and content errors of the 1994. o The specific objectives were: a) Evaluation of the accuracy of the PHC results in terms of o o coverage and content (response quality) for selected characteristics; b)Identification of procedural and conceptual improvements needed for undertaking PHC in the future. Source and causes of coverage and content errors from the census enumerators and respondents perspective. c) Evaluation of the quality of the census enumeration areas as sampling units for inter-censual surveys.

2 POST ENUMERATION SURVEY… UNIVERSE AND SAMPLE DESIGN o The universe defined for the 1994 PES was the population living within a conventional household in the country excluding Afar and Somalie Regions. o More over, residents of collective quarters and homeless persons were excluded from the survey. o The survey was designed to provide reliable estimates of the census coverage error for the whole country and for six major geographic domains. o The sample design was a stratified cluster design and EAs were the sampling units.

2 POST ENUMERATION SURVEY… q Formation of the six domains was based on the administrative division of the country. o Out of the six domains, three were composed of one administrative region each; o The remaining three domains were formed by combining two neighboring administrative regions which are assumed to have similar ecological, cultural and social features. o A total of 290 EAs were selected for the PES. o Concerning field activities data collection was carried out at two different periods:

2 POST ENUMERATION SURVEY… o First period was November 1994 three weeks after enumeration; o The second period was during the last two months of December 1994. n Due to limited logistical and manpower because National Agricultural Sample Survey was planned to be carried immediately after PHC. o Some enumerators and supervisors were selected from the PHC and the remaining from CSO’s permanent field staff. o Enumerators-supervisors ratio was three to one where as in main census it was five to one.

2 POST ENUMERATION SURVEY… Matching and Reconciliation Visits q Manual matching of PES data with the corresponding census data was done at various stages: q Matchers; q Supervisors and semi-professionals; q Senior technicians and professionals. q Reconciliation visits were conducted to make the coverage evaluation complete, to resolve issues not clearly determined during matching.

2 POST ENUMERATION SURVEY… o The PES report was prepared two years after the field work; o The of PES has not been officially released for the public. o In the report it has been indicated that at national level older and rural population were less omitted by the census than their counter parts. o Estimates of coverage and content errors were made but not communicated to the public.

3 Census Data Evaluation through Demographic Analysis o Demographic analysis has been commonly used to evaluate the census data in the previous censuses and planned to be also used for the current as well. o The country has no vital registration system, did not conduct a PES for all censuses but 1994 and hence, demographic analysis has been found to be an important tool for the overall assessment of census data quality. o Age-sex pyramid, summary indices such as Whipple’s Index and Meyers’ Index were commonly used to assess the census data. o Comparisons of the recent census results with the previous census and with the projection results were also used to evaluate the census data.

4 CONCLUSION o PES was NOT conducted for the 2007 PHC due to the fact that two census dates were adopted: one for the regions with the sedentary population and another for the pastoral regions; o Since the preparatory activities of the second has begun immediately after the first enumeration was completed and hence, it was not possible to get adequate manpower and time as well to carry out an independent PES; o Lack of experience. Because few staff members who were involved in the previous census have left the office due to different reasons;

4 CONCLUSION… o Using demographic analysis, so far no separate census data evaluation report was prepared. Instead, the analysis of data quality used to have been provided for some variables as part of the analytical reports. o However, there is an increasing concern about information on the extents of the coverage and content errors from the data users. o The importance of conducting PES for the next census has been underlined by the Census Commission in order to be able to furnish the information.

THANK YOU