I realize that. But if there were two datasets that, when combined, provided the right amount of data, figuring that we already know how to interpret the data, we should be able to solve that problem.

However, finding a dataset that has the right missing data is probably only marginally more likely than finding a full dataset.