ASSESSING THE VALIDITY OF MOBILE PHONE SURVEYS FOR MORTALITY DATA COLLECTION: A CASE STUDY FROM OUAGADOUGOU, BURKINA FASO.

Kassoum Dianou, Université Catholique de Louvain (UCL) - Centre for demographic research
Bruno Masquelier, Louvain University (UCL)
Bruno Lankoandé, Joseph Ki-Zerbo University (UJKZ)
Abdramane Soura, Université de Ouagadougou
Hervé Bassinga, Institut Supérieur des Sciences de la Population (ISSP) / Université de Ouagadougou
Ashira Menashe-Oren, Universite catholique de Louvain (UCL)
Malebogo Tlhajoane, London School of Hygiene and Tropical Medicine
Georges Reniers, London School of Hygiene and Tropical Medicine (LSHTM)

Mobile Phone Surveys (MPS) are increasingly being used, however, their comparability and validity in collecting mortality data is unclear. This study quantifies the impact of errors related to age, dating, and omissions in mortality rates from MPS. We conducted a validation study in the Ouagadougou Demographic and Health Surveillance System (OHDSS), interviewing adults over the phone, for whom we had high-quality data on the survival of their children, siblings, and parents. Individual records were matched, and we assessed the sensitivity and accuracy of mobile phone interviews in recording vital events, ages and dates. Our findings revealed that men tended to overestimate their age in MPS, while women tended to underestimate it. Our results also showed that data on the survival of children, siblings, and parents collected during MPS exhibited moderate specificity but high sensitivity in recording deaths. We applied an imputation method to analyze biases in survey estimates of child and adult mortality. Reporting errors in MPS data did not significantly impact estimates of under-five mortality rates but led to an underestimation of adult mortality. This study sheds light on the complex error patterns associated with MPS, emphasizing the need for thorough examination before considering their broader implementation.

Keywords: Data and Methods, Comparative methods , Mortality and Longevity, Population and Development

See extended abstract.