Assessment of quality of data: the case of the 2008-09 KDHS
This study focused on the assessment of the quality of the Kenya Demographic Health Survey (KDHS) data, specifically the 2008-09 survey. It set out three objectives: to determine the extent of age heaping or digit preference for males and females in the 2008-09 KDHS; to examine age misreporting and transfers of respondents across age boundaries; and to determine sex ratios by age in the 2008-09 KDHS data. It utilised the Myers' Blended Method for data quality checks for ages given in single years and the age and sex ratios and the United Nations Joint Score methods to assess the quality of age reporting in five-year groups. The study established that the 2008-09 KDHS age data is highly inaccurate, with more women than men having their ages reported as either unknown or not given at all and age heaping rampant among males and females across the ages. Higher heaping was observed in even age groups compared to the odd age groups for both sexes. Females generally misreported their ages more compared to the males in the survey. Respondents had preferences for ages ending in terminal digits 0 and 5 for the males and 0, 5 and 8 for the females, with the exception (for both sexes) of ages 5 and 15 years. Ages 55, 48 and 68 years for the females were other exceptions to this observation. Overall, males and females avoided ages ending in terminal digits 1, 7, and 9. The data is also characterised by systematic errors brought about by age misreporting as is evidenced by age ratio values. For the males, there were preferences for the age groups 30-34 and 70-74 years resulting in unusually more than the expected numbers while age groups 65-69 and 75-79 years were avoided giving way to unusually fewer than the expected nunibers. On the other hand, females reported preference for the 10-14,20-24,50-54 and 70-74 years age groups, and avoided the 15-19, 55-59 and 75-79 years age groups. In terms of numbers, females outweighed males all through except at birth. This in turn suggests that individuals concerned had their ages carried across age group boundaries, either to the next lower or higher age group, a character more pronounced among females compared to males. The errors detected in the 2008-09 KDHS data are therefore likely to have compromised its quality and the accuracy of the various demographic measures derived out of it. The study therefore calls for intensive training of KDH.S. enumerators in order to reduce errors pertaining to respondents' ages in the future. Often, in estimation of ages, they base their figures on physical attributes, marital status among others, but it would be desirable if they sought documentary proof when in doubt. The study recommends too that populations be educated through mass media on the need to report their ages as accurately as. is possible. Cultures or traditions that influence misinformation on age should be discouraged. It is prudent too that other methodology be employed to assess KDHS data to confirm the study findings and correct the errors for better and quality DHS data.