This paper provides detailed guidance on using big data to fill data gaps in impact evaluations. Data gaps can arise due to the inaccessibility of target populations, inadequate aggregation of data, data collection lag times, and data being missing in some contexts, like pandemics, conflicts, and humanitarian emergency situations. The paper includes a number of specific examples and additional references.
One issue to note when using the paper is that it uses the terms 'control group' and 'comparison group' inconsistently. The examples provided mostly refer to the use of comparison groups and quasi-experimental designs, not to randomly assigned control groups