Testing for Differences in Two Datasets Assume that I have a dataset that contains the dates that measurements were made on a group of people. Thus, unique people identifiers, measurement results, and dates of measurements are available.  Now assume that I have a another version of the same dataset from a different source, and that there are differences between the two datasets. Is there a statistical test that can be used to tell if the differences between the two datasets are statistically significant?
 A: You have the entirety of both datasets, so it would not make sense to do a statistical test.  For example if you tried to test
Ho: the datasets are not different
Ha: the datasets are different
and you find even one difference between the datasets, they are different and Ho would be rejected with certainty (p-value = 0; the probability you find a difference given Ho is true would be zero).  Whether or not the datasets are the same is a deterministic property that can be observed, no need to test it.
A: If you have two datasets that are supposed to be the same data then testing for significance is silly - if there is the slightest possibility that the differences could be significant, then at least one of the data sets is junk, unless the data sets are so huge that even negligible differences are significant.
You have 'data mining' as a tag. Do you have some huge amount of data?
What you should be doing, in nearly all cases anyway, is looking at and trying to explain the differences, not seeing whether they are significant
