Mixing two data sources I'm working on my master thesis using the gravity equation. I have to choose a data source for the trade flows of Iran with it's trade partners. I have a panel data and for the intended period (over 20 years) there is a data source from  Iran's Custom Administration. My problem is, for some of the years, this data source contains no information about the trade transactions. My question: Could I get help from another data source for the missing Datas, and when yes, how can I check the robustness? Thank you guys!
 A: The easiest check you can make is to compare that the data that is present in both datasets is the same. If it differs, you can study where the differences come from and whether you need to apply some transformation on the data from the second set to reuse it in the context of the first dataset.
A common assumption that is made, and verified by the above test, is that the two datasets represent the same phenomenon. If this is the case, they should have similar distributions. This should hold for both the overall data and for data filtered according to any features (e.g. not only global exports, but also exports to one particular country, and exports of one particular commodity). So comparing the distributions of the two datasets is one strong robustness check you could make.
Another assumption that you could add is continuity over time. In this case you can use the present for the values missing in the first dataset, you can compare the extrapolation see if the second dataset is within certain vicinity from the extrapolation of the first.
Generally, discrepancies are not impossible/improbable, but you should try to have explanations of where they come from. This would give you a way to reason about the quality of the data and whether it could be used for your purposes.
