ASSESSMENT OF THE QUALITY OF AGE REPORTING AND FERTILITY DATA IN THE NATIONAL AIDS AND REPRODUCTIVE HEALTH SURVEYS OF 2005, 2007 AND 2012

Abstract

Content and coverage errors are common features of demographic and fertility history data especially in developing countries. As a result, most surveys include data quality assessment findings to aid correct interpretation and improve future surveys. The National AIDS and Reproductive Health Survey (NARHS) in Nigeria has been conducted five times without any form of data quality assessment. This study was designed to assess the quality of the demographic and fertility history data in NARHS 2005-2012. Data from NARHS (2005, 2007 and 2012) were retrieved for data quality assessment. A total of 52,837 males (15-64 years) and females (15-49 years) respondents were analysed. Data on 51,365 Children Ever Born (CEB) reported by their mothers were also analysed. Demographic techniques involving the calculation of Myers' index, Whipples' index (WI) and Age Accuracy Index were used in assessing the digit preference in age related data, extent of age heaping, consistency of age and sex distribution. Internal consistency of fertility history data according to age group of mothers was also assessed by calculating the mean of CEB, sex ratio of CEB, sex ratio of children surviving and sex ratio of children dead. Myers' index revealed that preferences for digits 0 and 5 were higher in 2012(22.94% and 19.86%) compared to 2005 (19.92% and 18.10%) and 2007 (9.39% and 17.43%). However, terminal digits 3(6.68%), 6(6.84%) and 1(7.10%) in 2005, 1(6.01%), 6(6.32%) and 3(6.64%) in 2007 and 6(4.52%), 3 (5.48%) and 4 (5.66%) in 2012 were all greatly avoided in all the three surveys. WI revealed high heaping at age ending with 0 (195 in 2005, 185 in 2007 and 224 in 2012) indicates that the quality of age data was poor while heaping at age ending with digit 5(104 in 2005, 100 in 2007 and 111 in 2012) indicates that the quality of age data was approximately accurate. The Age Ratio and Sex Ratio also showed inconsistencies in the reporting of age-sex data, as there were over-representation of respondents at age group 30-34 years in 2005 and 2012 among females. Similarly, the age-specific sex ratios fluctuated across all the age groups and all the values deviated from the expected range of 102-107. This was also confirmed by Age Accuracy Index of 42.12 in 2005 and 49.92 in 2012 which showed that quality of age-sex data in these two survey periods were poor. Furthermore, the 2005 and 2007 surveys show inconsistencies in reporting number of CEB as most of the sex-ratios deviated from the normal range of expected values (102-107) while 2012 survey showed little consistency in reporting number of CEB. Demographic and fertility history data collected in National AIDS and Reproductive Health Surveys 2005-2012 showed problems of digit preference, heaping and inconsistencies in reporting age-sex and children ever born. Better quality of age-sex data was reported in the 2007 survey. Hence, adequate training of enumerators and quality of supervision of field work should be improved in future surveys. Similarly, respondents must be sensitized on the importance attached to demographic and fertility history data in order to minimize age misstatement during future surveys.