Data quality in demographic surveys

This study discusses reasons why household-based surveys might under-represent young unmarried and childless women and therefore not capture correctly the life experiences of young women. We examine the differences in marital status data between household-based surveys and censuses in several Asian countries that have experienced increased labor force participation among young women (Bangladesh, Cambodia, Lao PDR and Mongolia). To determine the extent of over-estimation of fertility levels, we adjust age-specific fertility rates on the basis of marital status distribution in censuses. The implications for interpretations of the fertility trends are far-reaching. In Bangladesh, while differences between reported and adjusted values were small for the 1992-93 DHS, the 2007 DHS reported an adolescent fertility (ABR) of 126 and was adjusted to 77 births per 1000 women aged 15-19. When analyzing fertility trends only from DHS, the country achieved in two decades a relatively small deduction adolescent childbearing. Yet the story may, in fact, be different: the postponement of childbearing —concurrent with the postponement in marriage as seen in the census data—resulted in a decline of ABR to half the level from the early 1990s and played a key role in the rapid fertility decline in Bangladesh.

Unwanted fertility is an important concept for understanding fertility behaviour. In Australia, about 10% of live births between 2005-2008 are estimated to have been unwanted, and a further 10% were mistimed in that they occurred earlier than desired. One criticism of measures of unintended childbearing is that they are usually only based on the views of one parent, typically the mother. In recent years, there has been increasing awareness of the need to consider men’s roles and to study childbearing decision in the context of the couple. To this end, many surveys ask respondents to provide proxy answers on behalf of their partner. While proxy reports can be a cost-effective and simple way to gain information about the partner, without actually interviewing them, there has been little systematic research assessing their accuracy. Using data from a 2008 Australian survey we assess the accuracy of proxy reports of pregnancy wantedness by directly comparing them with the partner’s responses. We find that proxy reports are accurate in 87% of cases. In the remaining 13%, they either over or underestimated the degree to which the partner wanted the pregnancy. Proxy reports are often inaccurate when the partner did not want the child.

DHS surveys are the main source of estimates and analyses of fertility and under-five mortality in developing countries. All of these estimates and analyses are based on assumptions about the underlying accuracy of the events and dates in the birth histories. The main concerns of the proposed paper are whether the assumptions are valid and the sensitivity of various measures and analyses to deviations from the assumed accuracy. Three potential weaknesses in the data can be identified: the representativeness of the respondents and the accuracy of their reported ages; possible omission of births in the birth history; and systematic displacement of events. Each of these possible sources of distortion will be examined with statistical methods, building primarily on the leverage provided by multiple surveys in the same country. The second part of the assessment concerns the impact of errors on key outcomes. Those outcomes will be specified in advance and will include measures of levels and trends in rates such as the TFR and neonatal mortality rate and coefficients in statistical models.

In countries where the registration of deaths remains incomplete, the main sources of data on mortality are retrospective reports collected on the survival of close relatives. Birth and sibling histories are known to be plagued by recall errors but few methods exist to detect these errors. This paper introduces two simple approaches to assess the consistency between sibling histories and reports on the fertility of the previous generation. A comparison between the average size of sibships and the mean number of children ever born (CEB) to women of the previous generation indicates that a large proportion of siblings is omitted in Demographic and Health Surveys. On average the reported sibsizes are about 15 percent lower than expected from data on CEB. Discrepancies between reported and expected sibsizes are larger in sub-Saharan Africa than in other regions and they increase with the age of respondents. Second, a linkage of sibling histories reported by young women aged 15-18 with birth histories of their mothers when they live in the same household highlights frequent inconsistencies between mother's and daughter's reports on members of the same family. Daughters tend to report fewer siblings than expected from their mother's birth history, but more adult deaths.

Our study purpose is to assess data quality of self-reported current use of the Lactational Amenorrhea Method (LAM) in the Demographic and Health Surveys (DHS). LAM, an important contraceptive method during the postpartum period, is a modern contraceptive with an effectiveness rate of 98% for the six month postpartum period when used correctly. Our study’s specific objectives are: 1) to examine accuracy of self-reported LAM use compared to the constructed correct practice variable, and 2) to explore differentials in accuracy measures by characteristics at the individual-level and survey-level by analyzing data from 73 DHS conducted in 45 countries since 1998. Findings reveal that on average only 26% of reported LAM users met criteria of correct LAM practice across the surveys, indicating potentially unacceptably low user effectiveness at the population level. We discuss implications for future DHS data collection efforts, and implications for family planning and maternal and child health programming.