Volume 38 - Article 55 | Pages 1663–1698

Measuring fertility through mobile‒phone based household surveys: Methods, data quality, and lessons learned from PMA2020 surveys

By Yoonjoung Choi, Qingfeng Li, Blake Zachary

Print this page  Facebook  Twitter

 

 
Date received:29 Sep 2017
Date published:23 May 2018
Word count:5056
Keywords:data quality, Demographic and Health Surveys (DHS), Performance Monitoring and Accountability 2020 (PMA2020), survey design, total fertility rate (TFR)
DOI:10.4054/DemRes.2018.38.55
 

Abstract

Background: PMA2020 is a survey platform with resident enumerators using mobile phones. Instead of collecting full birth history, total fertility rates (TFR) have been measured with a limited number of questions on recent births. Employing new approaches provides opportunities to test and advance survey methods.

Objective: This study aims to assess the quality of fertility data in PMA2020 surveys, focusing on bias introduced from the questionnaire and completeness and distribution of birth month and year, and to estimate TFR adjusted for identified data quality issues.

Methods: To assess underestimation from the questionnaire, we simulated births that would be counted using the PMA2020 questionnaires compared to births identified from full birth history. We analyzed the latest Demographic and Health Surveys in ten countries where PMA2020 surveys have been implemented. We assessed the level of reporting completeness for birth month and year and heaping of birth month, analyzing 39 PMA2020 surveys. Finally, TFR were calculated and adjusted for biases introduced from the questionnaire and heaping in birth month.

Results: Simple questions introduced minor bias from undercounting multiple births, which was expected and correctable. Meanwhile, incomplete reporting of birth month was relatively high, and the default value of January in data collection software systematically moved births with missing months out of the reference period. On average across the 39 surveys, TFR increased by 1.6% and 2.4%, adjusted for undercounted multiple births and heaping on January, respectively.

Contribution: This study emphasizes the importance of enumerator training and provides critical insight in software programming in surveys using mobile technologies.

Author's Affiliation

Yoonjoung Choi - Johns Hopkins University, United States of America [Email]
Qingfeng Li - Johns Hopkins University, United States of America [Email]
Blake Zachary - Johns Hopkins University, United States of America [Email]

Other articles by the same author/authors in Demographic Research

» Death distribution methods for estimating adult mortality: Sensitivity analysis with simulated data errors
Volume 21 - Article 9

» Neonatal mortality in the developing world
Volume 14 - Article 18

» Unconventional approaches to mortality estimation
Volume 13 - Article 12

Most recent similar articles in Demographic Research

» The quality of demographic data on older Africans
Volume 34 - Article 5    | Keywords: data quality, Demographic and Health Surveys (DHS)

» A Stata module for computing fertility rates and TFRs from birth histories: tfr2
Volume 28 - Article 38    | Keywords: Demographic and Health Surveys (DHS), total fertility rate (TFR)

» Probabilistic projection of subnational total fertility rates
Volume 38 - Article 60    | Keywords: total fertility rate (TFR)

» Collecting data from migrants in Ghana: Lessons learned using respondent-driven sampling
Volume 38 - Article 36    | Keywords: data quality

» Estimating mortality from external causes using data from retrospective surveys: A validation study in Niakhar (Senegal)
Volume 38 - Article 32    | Keywords: data quality