Disease Data and Development: Evidence from DHS and MICS Surveys

Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) are widely used in global health research. Cross-country evidence based on these surveys could be unreliable, if survey availability and quality exhibited systematic patterns. This paper documents whether such patterns exist in low- and middle-income countries (LMICs) over the period 1982–2016, examining countries’ economic, demographic, and political characteristics. Descriptively, DHS and MICS are very unevenly distributed across LMICs, but they show good data quality in terms of age and anthropometric mismeasurement. We explore country-level correlates of data availability and quality in cross-sectional and panel models (with country and time fixed effects). While economic and demographic factors show relatively weak influence, political factors do matter. Conflict is a negative predictor of DHS and MICS data availability. Democracy exhibits an inverted-U-shaped association with availability of USAID-financed DHS. Availability of UNICEF-financed MICS is negatively yet convexly associated with countries’ US-aligned voting in the UN. Predictors of data quality are unsystematic, but we detect regional differences. Overall, we recommend using DHS and MICS data together, and including country fixed effects in cross-country regressions. A crucial implication for higher-quality research is to harmonise and coordinate MICS with DHS, and make them jointly accessible.