Indicators to Examine Quality of Large Scale Survey Data: An Example through District Level Household and Facility Survey

Abstract

BackgroundLarge scale surveys are the main source of data pertaining to all the social and demographic indicators, hence its quality is also of great concern. In this paper, we discuss the indicators used to examine the quality of data. We focus on age misreporting, incompleteness and inconsistency of information; and skipping of questions on reproductive and sexual health related issues. In order to observe the practical consequences of errors in a survey; the District Level Household and Facility Survey (DLHS-3) is used as an example dataset.MethodsWhipple's and Myer's indices are used to identify age misreporting. Age displacements are identified by estimating downward and upward transfers for women from bordering age groups of the eligible age range. Skipping pattern is examined by recording the responses to the questions which precede the sections on birth history, immunization, and reproductive and sexual health.ResultsThe study observed errors in age reporting, in all the states, but the extent of misreporting differs by state and individual characteristics. Illiteracy, rural residence and poor economic condition are the major factors that lead to age misreporting. Female were excluded from the eligible age group, to reduce the duration of interview. The study further observed that respondents tend to skip questions on HIV/RTI and other questions which follow a set of questions.ConclusionThe study concludes that age misreporting, inconsistency and incomplete response are three sources of error that need to be considered carefully before drawing conclusions from any survey. DLHS-3 also suffers from age misreporting, particularly for female in the reproductive ages. In view of the coverage of the survey, it may not be possible to control age misreporting completely, but some extra effort to probe a better answer may help in improving the quality of data in the survey.

Validation study of the Alzheimer's Disease Assessment Scale‐Cognitive Subscale for people with mild cognitive impairment and Alzheimer's disease in Chinese communities

“…Smaller values of DI indicate less bias. Generally, DI values that are below 10 illustrate sufficient representativeness of the generalized population from the subset of sample individuals …”

How Reliable Are Reports of Early Adolescent Reproductive and Sexual Health Events In Demographic and Health Surveys?

“…Because the legal age of marriage in many countries is older than the traditional norm, 16 fear of repercussions may lead to overstatement of age at marriage. 17 In Bangladesh, however, women have been shown to intentionally lower their age at marriage, possibly because of the social desirability of early marriage for women there. 18 In communities in which sex or childbirth outside of marriage is censured, some women may feel the need to adjust their responses to make it appear that first sex or first birth occurred after marriage.…”

The Heaping Effect in the Survey Data on the Number of Friends

“…To observe the age dependence of the eect quantatively, the values of the Whipple index I W (a) are calculated in these three years, as dependent on the age a of the respondents. The formula for the 10-range is [22]:…”

“…According to the discussion in [22], the values of the Whipple index for the 10-year range between 100 and 150 mean that the eect is low, between 150 and 250 that the eect is moderate, and high above 250. The (1417) Fig.…”

“…For people above 50, the mean number of friends is not larger than 100 [20]. A series of indices have been proposed to evaluate the eect quantitatively [21]; below we use the Whipple index as an example [22].…”