Publications

Publications Summary


Document Type
Methodological Reports
Country(s)
Kenya, South Africa, India
Language
English
Recommended Citation
Allen, Courtney K., Julia Fleuret, Jehan Ahmed. 2020. Data Quality in Demographic and Health Surveys That Used Long and Short Questionnaires. DHS Methodological Reports No. 30. Rockville, Maryland, USA: ICF.
Download Citation
RIS format / Text format / Endnote format
Publication ID
MR30

Abstract:

Since the first phase of The DHS Program in 1984, DHS surveys have increased in scope and complexity as questionnaires are lengthened and survey modules are added. Increase in questionnaire length can increase interview time, and ultimately may present more burden for both the interviewer and the respondent. It seems intuitive that longer questionnaires would have different effects than shorter questionnaires on fieldwork, interviewer fatigue and performance, and survey implementation. Surveys that have two different lengths of questionnaires offer an opportunity to explore the extent to which questionnaires of different lengths may have these effects. This report aims to understand the effect, if any, of questionnaire length on data quality using the 2016 South Africa DHS, 2014 Kenya DHS, and 2015-16 India National Family Health Survey. We described the differences in fieldwork and interview length between the long and short questionnaires and examine data quality indicators to see whether the different length of questionnaires experience differing data quality. We used two types of data quality indicators: indicators that may reflect efforts on the part of fieldworkers to reduce survey burden (i.e., their workload), and those concerning age and date of birth that are notoriously difficult to collect accurately in household surveys. Additionally, we explore themes resulting from qualitative interviews with survey experts who worked on surveys with long and short questionnaires to understand how these surveys are implemented, and data quality considerations, if any, along the survey process. The long questionnaires in each country had large differences in the average number of variables per woman and interview length compared to short questionnaires. In Kenya, the long questionnaire lasted twice as long as the short questionnaire on average. Despite these differences, there is no evidence that interviewers may have intentionally reduced their workload. We found little evidence that having differing lengths of questionnaires resulted in data quality differences. Key informants agree that deploying long and short questionnaires to obtain estimates of a subset of survey indicators for nonstandard populations or at lower administrative levels solves a significant problem in survey design and implementation. While challenges remain, it is widely embraced as a useful approach to survey design. Key informants were unanimously supportive of using long and short questionnaires in the future. We recommend that the use of long and short questionnaires be included in the survey design options for future DHS surveys so that surveys can meet in-country data demands while maintaining feasibility.

Browse

Browse for Publications by:

Browse for Journal Articles based on DHS data by: