Data quality in demographic surveys: tests and experiments

This paper analyzes the variables used to measure female empowerment using factor and multilevel analysis with data coming from Egyptian Demographic and Health surveys at different points in tome. The findings highlight the difficulty in measuring female empowerment in a meaningful way, with questions around both the reliability and the validity of the data, with particular issues raised around the temporal comparability of such measures. Although I specifically looked at Demographic and Health surveys in Egypt, the findings suggest that similar problems could be found in other surveys and settings.

Involuntary childlessness is a distressing condition for many men and women throughout the world. However, data on childlessness, especially in low-income countries is often poor. The Demographic and Health Surveys collect birth histories which include data on childlessness. However, the reported levels of childlessness in many countries fall below the level considered biologically feasible. Furthermore, patterns of childlessness often do not fit those expected from fertility patterns and DHS estimates are far lower than those reported by censuses. Despite these shortcomings, the DHS data are often considered sufficiently robust to reflect relative trends. This study seeks to identify the countries and regions where childless data are particularly problematic. The study will look at the variation in childlessness within countries in order to determine whether the data can be used for relative analysis and any adjustments which can be made to better reflect actual levels.

Background: Each year 15-30 million infants – up to 20 percent of all infants, are born with a low birth weight (LBW). This paper try to investigate the factors negatively correlated to LBW to minimize the prevalence of LBW in India based on the data (NFHS, 1992-2006). Methodology and findings: The analytic sample consisted of 0 to 36 months children with birthweight data obtained from health cards and maternal recall. The analysis done on sample in the distribution across household wealth, religion, and urban residence was compared using multilevel models. The chance of being low birth weight is almost ten percent more in poorest wealth quintile household than richest. This is consistently true from 1992 to 2006. Education is not playing direct role on LBW during 1992-2006. The place of residence is not distinctly playing role in the difference of prevalence of LBW children. India is classified here into six regions and the prevalence of LBW in north region is more than south region. The findings show that the north region is almost 50 percent more likely to have low birth weight children than south in India at 95 percent confidence interval. Conclusion: It can be concluded that the LBW is more prevalent in north, no educated women with higher birth order, and lowest wealth status household than their counterpart in India.

Estimation of inherent coverage error in census count is an highly addressed problem in statistical demography.
In this article we briefly discuss the statistical methodology to obtain the omission rate in Indian census using Dual System
Estimation(DSE) technique. We have explicitly studied the correlation bias factor involved in the estimate, its extent
and consequences. A new potential source of bias in the estimate is introduced. At the survey
time, more efficient enumerators (than census time) are appointed and
this fact may inflate the dependency between two lists and
lead to a significant bias. Some examples are given to
demonstrate this argument in various plausible situations. We have suggested one simple and flexible approach which can control the bias.
Our proposed estimator can efficiently overcome the potential bias by achieving desired degree of accuracy(almost unbiased) with relatively better efficiency. Overall improvements and consistency in the results are explored through simulation study on different populations. New estimator is also compared also with another type of dual system estimator used by Indian Sample Registration System(SRS) to estimate number of births and deaths.

Biomarkers as innovations look very attractive for inclusion in routine demographic health surveys. Biomarkers are highly sensitive biochemical indicators akin to routine investigations used in clinical diagnosis, are relatively new to demographic surveys. Such investigations are generally performed in laboratory situations with rigorous training and performance to predict with specificity and sensitivity. The paraphernalia and backup needs to be robust and dependable to get consistent output in tropical weather conditions in India with wide seasonal variations in temperature. Humidity and other atmospheric conditions affect the outcome of indicators. This paper reviews closely the methodological perspectives, human resources recruitment, choice of techniques and equipment for assays used in several recent surveys in India. An effort is made to understand the problems of logistics, quality of data and implications of such data used for policy perspectives. The issues of oversimplify a clinical parameter to fit a demographic survey intended to seek good data are discussed. Suggestions to remedy shortcoming based on long term clinical and practical field experiences are discussed, drawing comparisons with several very well acknowledged biomarkers surveys.

Data quality is an important issue with all data, be they observational records, survey data or specific checklists. There is a universal requirement by many governments around the world for data to be high quality and be better documented. Maintaining good quality of data is always a challenge in any survey. Therefore, the broad objective of the present study is to assess the quality of data in terms of misreporting and to understand the internal consistency using National Family Health Survey and District Level Household and Facility Survey. For this study, we have selected six states like Utter Pradesh, Bihar, Gujarat, West Bengal, Kerala and Tamil Nadu. Age accuracy will be measured by different methods such as Whipple’s Index, Myer’s Index, U. N. Joint Score, and Pullum Method. To check the quality of data the selected indicators are age reporting, women’s education, maternal health information (Contraception use, visit of ANC and other routine checkup, menstruation). Results indicate age misreporting in all the states and inconsistency is obtained for reporting of ANC visits and problem related to menstruation. Quality differs considerably by states. In terms of respondent’s characteristics, literacy comes out to be the highly influencing factor in reporting accurately.

The aim this study is to examine the nature and patterns of the reported age statistics in the North West Province using the 1996, 2001 and 2011 South African population censuses. First, the enumerated age-sex population distribution from the 1996 and 2001 censuses are examined to highlight the nature and patterns of age misreporting by means of such indices as Whipples, Myers and United Nations Age-Sex Score. Second, essential age-sex features of the population and their implications for development planning are highlighted. The study indicate that the quality of the reported age statistics in North West Province, like in the other provinces in South Africa, is quite good and contrast to findings from most other Sub-Saharan populations. Furthermore, the reported age statistics show a preference of terminal digits 6, 4, 8, 0 and 2, implying a preference of ages ending in even numbers. Further research should establish the factors responsible for this pattern.

The reliable data on vital events from Civil registration system is the key to track the progress of MDG IV and V. In India, Sample Registration System (SRS) is taken as a substitute to Civil Registration System and provide information to measure MDG. There has been some efforts to examine the completeness of SRS data until 2001. This paper examines completeness of registration of death by SRS, India for the period 1990-2009.The completeness of death registration is assessed using the method proposed by Bennett and Horiuchi (1981). Finding of the study suggested over registration of deaths during 1990-99 and under registration of death during 2000-09 by Sample Registration System of India. Result show that, at national level, during 1990-99, 22% deaths among men and 11% among female were over registered while during 2000-09, 1% of death among men and 13% among women were under registered by the sample registration system in India.

Population movement and its impact on other social, economic and demographic characteristics have been increasing attention of social scientists and demographers in recent years and a large number of studies on migration have been conducted. The majority of studies used macro level approach by operating on highly aggregate data for countries, district, states and the nation as a whole. In recent years the study of migration at the micro-level is useful and it is governed by human grouping on probabilistic and stochastic approach. Households play a very important role in decision of an individual to move or not to move from the household. The main objective of the paper is to develop a probability model for the total number of migrants from a household. The suitability of the model is tested through observer data. A primary source of data has been used for this study. The area of the study is Varanasi district in Eastern Uttar Pradesh, India. Result suggests that the proposed model under consideration is a better approximation to observed distribution of rural out migration at the micro level. Thus it may be a useful tool in calculating the various probabilities connected with the out-migration from the household.

The under-5 mortality rate (U5MR) is an important measure of the well-being of a country’s children, and its estimation is particularly critical as we approach the deadline of Millennium Development Goal 4 to reduce U5MR by two-thirds between 1990 and 2015. However, for the great majority of developing countries without well-functioning vital registration systems, estimating levels and trends in under-5 mortality is challenging, not only because of limited data availability but also because of issues with data quality. Global estimates of child mortality are often constructed without accounting for potential biases in data series which may lead to inaccurate point estimates and/or uncertainty intervals. We propose a Bayesian spline regression model for assessing levels and trends in the U5MR, whereby biases in data series are estimated for each source type through the inclusion of a multilevel model to improve upon the drawbacks of current methods. Preliminary results show that the proposed model is able to flexibly capture changes in U5MR over time and gives point estimates and uncertainty intervals that reflect potential biases in data series.