It is quite possible that analyses not yet thought of would yield even more dramatic differences between the two data sets.
Even if a identification method is found that maintains many of the observed research results from the original data set, there can be no guarantee that other analyses will not have been corrupted by identification.