Intervention group serves as its own control over time |
No true counterfactual, so cause and effect cannot be conclusively inferred |
Program is preexisting or full (above threshold) coverage |
No need to exclude any population or group from the intervention/program, so can be applied to programs with nation-wide coverage |
Multiple sources of data, analyses, and triangulations needed to establish plausible impact |
Pretest (baseline) data for the relevant indicators can serve as counterfactual scenario |
Differential selection bias and attrition risk to cause bias and dilution of impact |
No other plausible explanations for observed outcomes or any likely confounder effects can be adjusted for |
Can adapt to use existing data collected for other purposes (DHS, MICS) |
Data might not be as specific as required |
All-cause under-five mortality is a sensitive, specific, and time-sensitive proxy for changes in malaria-specific mortality in highly endemic countries |