Assessment of Quality of Data of the 2014 Kenya Demographic and Health Survey (Kdhs)

Abstract

Good quality of data is important in reaching inferences and conclusions that are accurate, reliable and valid. Before analyzing demographic data, it is essential to be familiar with whether the data is accurate and provides acceptable answers. Age data evaluation allows the determination of the direction and magnitude of errors. The need for reliable, valid and complete sets of Kenyan data justified this study. The objective of the study was to assess the quality of the 2014 KDHS data. This was done with a special focus on assessing the completeness of reported date of birth for female interviewees and completeness of reported date of birth of children ever born. It also focused on assessing the degree of age heaping and digit preference by age and sex. The assessment was conducted using age ratios, sex ratios, Myer‟s blended index, and UN Age-sex Accuracy Index. The study also applied an appropriate adjustment method to smooth reported data. The adjustment was applied to remove short-term random variations, or outliers to reveal the important underlying quality data. From the study, it emerged that the 2014 KDHS age data was incompletely reported by 22.5 percent for date of birth for women and children ever born. Some dates, months and years were missing, but since ages had been provided, the missing elements were imputed. Age data was inaccurately reported with high fluctuations in age ratios for males and females, which could be an indication of persons in various ages being carried across age group boundaries or persons misreporting their own ages for various reasons. This compromises the quality of data. Myer‟s index revealed digit preference of age ending with '0' and '5' where the digit ending with '0' is more preferred than „5' for both male and female respondents. Moreover, females showed higher age misreporting than male respondents. Ages ending with 1 have the highest digit avoidance, followed by those ending with 3, 7 and 9. At the same time, the UN age-sex accuracy index revealed general inaccuracy in the 2014 KDHS age data. Six regions recorded highly inaccurate data and two regions had inaccurate data. The findings are consistent with previous study findings on quality assessment on Kenyan data. Although much has been done by agencies dealing with KDHS data collection to ensure that reported data is of the highest quality including rigorous training of interviewers on the most effective data collection methods, there is still room for improvement. This can be done by creating mass awareness and educating people on how false reporting during surveys and censuses distorts information derived from data and how their lives are affected. Since data anomalies could be arising from interviewers‟ motivation to reduce the length of the interviews leading to age misplacement and transfer, a more thorough supervision should be conducted during field work to ensure ages are recorded appropriately.